HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 21, 19008-19018, May 24, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/21/19008    most recent M112472200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Matsuzaki, T. Articles by Sugino, H. Search for Related Content PubMed PubMed Citation Articles by Matsuzaki, T. Articles by Sugino, H. Social Bookmarking                      What's this?

Regulation of Endocytosis of Activin Type II Receptors by a Novel PDZ Protein through Ral/Ral-binding Protein 1-dependent Pathway^*

Takashi Matsuzaki, Sayuri Hanai, Hisashi Kishi, ZhongHui Liu^§, YongLi Bao, Akira Kikuchi^¶, Kunihiro Tsuchida, and Hiromu Sugino

From The Institute for Enzyme Research, The University of Tokushima, 3-18-15 Kuramoto, Tokushima 770-8503, Japan and ^¶ Department of Biochemistry, Faculty of Medicine, Hiroshima University, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan

Received for publication, December 31, 2001, and in revised form, February 27, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Using yeast two-hybrid screening, we have identified a mouse Postsynaptic density 95/Discs large/Zona occludens-1 (PDZ) protein that interacts with activin type II receptors (ActRIIs). We named the protein activin receptor-interacting protein 2 (ARIP2). ARIP2 was found to have one PDZ domain in the NH₂-terminal region and interact specifically with ActRIIs among the receptors for the transforming growth factor  family by the PDZ domain. Interestingly, overexpression of ARIP2 enhances endocytosis of ActRIIs and reduces activin-induced transcription in Chinese hamster ovary K1 cells. In addition, immunofluorescence co-localization studies indicated the direct involvement of ARIP2 in the intracellular translocation of ActRIIs by PDZ domain-mediated interaction. Moreover, we have identified that the COOH-terminal region of ARIP2 interacts with Ral-binding protein 1 (RalBP1). RalBP1 is a potential effector protein of small GTP-binding protein Ral and regulates endocytosis of epidermal growth factor and insulin receptors. The studies using deletion mutants of RalBP1 and constitutively GTP and GDP binding forms of Ral indicate that ARIP2 regulates endocytosis of ActRIIs through the Ral/RalBP1-dependent pathway, and the GDP-GTP exchange of Ral is critical for this regulation.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Activin, a member of the TGF-¹ superfamily, has a broad range of physiological activities including hematopoiesis, bone morphogenesis, neurogenesis, and hormone action (1-5). These various actions on cell proliferation, differentiation, and apoptosis are dependent upon target cells. Activin transduces its signal via heteromeric complexes composed of two different serine/threonine kinase receptors, termed type I and type II (6-8). Upon ligand binding, the type II receptor transphosphorylates and activates the type I receptor kinase at the membrane region. Then, the type I receptor cytoplasmic domain interacts with intracellular signaling molecules, Smads, which regulate transcription of selected genes in a cell-specific manner (9, 10).

In the current model, the functions of the ActRIIs are limited to ligand binding, type I receptor recruitment, and transphosphorylation. However, it is speculated that there are specific roles of the ActRIIs in activin signaling. For example, the serine/threonine kinase domains of ActRIIs are constitutively activated, and ActRIIs exist as homooligomers (probably homodimers) even in the absence of ligands (11, 12). These data suggest the existence of a strict monitoring mechanism for type II receptor kinase activity. Furthermore, multiple forms of ActRIIs exist; two subtypes of activin type II receptors, ActRIIA and ActRIIB, each of which is encoded by individual genes, are known (13, 14). Several alternative splicing variants have also been found in each subtype (14-16). For example, activin type IIA-N receptor, a splicing product of ActRIIA, is specifically expressed in neural cells and is thought to mediate neuronal-specific activin action. In a previous report, we noted the identification of a PDZ protein called activin receptor-interacting protein 1 (ARIP1), which contains five PDZ domains and two WW domains and interacts specifically with ActRIIA among the receptors for the TGF- family (17). ARIP1 is a scaffolding protein that unites activin receptors with an intracellular signaling molecule, Smad, and thereby regulates activin-mediated signaling in neuronal cells. These findings clearly indicate that a delicate regulation of the functions of the ActRIIs is important for maintenance of proper activin signaling.

In this study, we report the identification of a novel cytoplasmic protein, which we named activin receptor-interacting protein 2 (ARIP2), by a yeast two-hybrid screening using the cytoplasmic region of the ActRIIA as bait. Different from ARIP1, ARIP2 has only a single PDZ domain and specifically interacts with ActRIIA and ActRIIB by the PDZ domain. More importantly, expression of ARIP2 enhances endocytosis of ActRIIs and suppresses activin-induced transcription. Furthermore, we found that RalBP1 interacts with the COOH-terminal region of ARIP2 and makes a ternary complex with ActRIIs and ARIP2 in vivo. Because RalBP1 is a downstream molecule of Ral and regulates endocytosis of EGF and insulin receptors (18), we examined whether Ral and RalBP1 regulate ARIP2-mediated endocytosis of ActRIIs. We were able to show that Ral signaling regulates ARIP2-mediated endocytosis of ActRIIs. ARIP2 interacts with ActRIIs and RalBP1 simultaneously. RalBP1 associates with POB1, which forms a complex with EGF receptor pathway substrate 15 (Eps15), Eps15-interacting protein (Epsin), adapter complex 2, and clathrin to regulate endocytosis of EGF and insulin receptors (18-22). Therefore, we proposed that ARIP2 is a scaffolding PDZ protein that unites ActRIIs and the regulatory proteins for endocytosis and controls activin-mediated signaling.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cell Culture and Antibodies-- COS-7 cells were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum and antibiotics (100 units/ml penicillin and 100 µg/ml streptomycin; Wako, Osaka, Japan), and CHO-K1 cells were maintained in minimum essential  medium (Invitrogen) with 10% fetal calf serum and antibiotics. The anti-ActRIIA and anti-ActRIIB polyclonal antibodies were purchased from R&D Systems (Minneapolis, MN). The anti-RalBP1 and anti-GAL4 polyclonal antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-FLAG (M2) antibody was from Sigma. The anti-C-Myc (9E11) antibody was from NeoMarkers (Fremont, CA). The anti-HA polyclonal antibody was from Upstate Biotechnology (Lake Placid, NY).

Yeast Two-hybrid Screening-- Yeast two-hybrid screening was performed using a commercially available system (Matchmaker Two-Hybrid System 2; CLONTECH, Palo Alto, CA) according to the manufacturer's protocol (17). Approximately 4 × 10⁶ clones of a mouse brain cDNA library constructed in pAct2 (CLONTECH) were screened using bait constructs, pAS-ActRIIA, which contained the nucleotide sequences for the entire cytoplasmic region of mouse ActRIIA (17), and pAS-ARIP2, which contained the full coding region of ARIP2 (nucleotides 127-588, Fig.1).

Cloning of Mouse ARIP2 cDNA-- To obtain a full-length cDNA for ARIP2, a mouse brain cDNA library in the ZAPII vector (STRATAGENE, La Jolla, CA) was screened using a fragment of ARIP2 as the probe. One clone covering the full-length cDNA of ARIP2 was identified.

DNA Constructs-- DNA constructs for the yeast two-hybrid screening were made using the plasmid pAS2-1 to express the fusion protein with the GAL4 DNA binding domain. pAS-ActRIIA was made by introducing a cytoplasmic region of mouse ActRIIA into pAS2-1, as described previously (17). pAS-ARIP2 was made by introducing nucleotides 127-588 of ARIP2 into pAS2-1. DNA constructs for the mammalian two-hybrid assay were made using the plasmid pBIND to express the fusion protein with the GAL4 DNA binding domain and using pACT to express the fusion protein with the VP16 activation domain (17). pBIND-ActRIIA, pBIND-ActRIIB, pBIND-TGF- type II receptor (TGFRII), and pBIND-bone morphogenetic protein type II receptor (BMPRII), encoding the cytoplasmic regions of mouse ActRIIA, mouse ActRIIB, human TGFRII, and human BMPRII, respectively, have been described previously (17). In pBIND-ActRIIAESSL and pBIND-ActRIIBESSI, four amino acid residues of ActRIIA and ActRIIB COOH terminus, respectively, were deleted. To make pACT/pBIND-ARIP2 and pACT/pBIND-ARIP2C, cDNA fragments composed of nucleotides 127-588 and 127-426 of ARIP2, respectively, were prepared by PCR and ligated into pACT and/or pBIND. DNA constructs for pACT-RalBP1-(499-647) and pcDNA3-RalBP1-(499-647) were made by introducing a cDNA fragment composed of amino acid 499-647 of RalBP1 (23) into pACT and pcDNA3. Expression constructs of ARIP2 were made by subcloning nucleotides 127-588 of ARIP2 into pcDNA3 (Invitrogen). A series of deletion mutants of ARIP2 was made by assembling PCR-generated fragments. pcDNA3-ARIP2PDZ, lacking the PDZ domain, contained nucleotides 226-588, and pcDNA3-ARIP2C, lacking the COOH-terminal region, contained nucleotides 127-426 of ARIP2. To make the glutathione S-transferase-Ral binding domain of RalBP1 (GST-RalBD), cDNA covering amino acids 391-499 of rat RalBP1 (23) was PCR-amplified with specific primers harboring the BamHI site at 5' terminus and the EcoRI site at 3' terminus and cloned into pGEX4T-1 (Amersham Biosciences). pCGN-HA-RalB, pCGN-RalB^G23V, pCGN-RalB^S28N, pBJ-Myc-RalBP1, pBJ-Myc-RalBP1-(364-647), pBJ-HA-POB1 were as described (19, 24, 25). The 5' and 3' junctional regions between the insert and the vector of each construct were sequenced to ensure that the inserts were introduced properly.

Mammalian Two-hybrid Assay-- Mammalian two-hybrid assay was performed using CheckMate mammalian two-hybrid system (Promega, Madison, WI), according to the manufacturer's protocol (17). In brief, CHO cells were cotransfected with the plasmids of interest, a cytomegalovirus promoter-driven -galactosidase and a reporter plasmid, pG5luc, which drives the luciferase gene under the control of GAL4-responsive promoter. Luciferase activity was measured and normalized to the -galactosidase activity as described previously (6).

Immunoprecipitation and Western Blotting-- COS-7 cells were cotransfected with indicated plasmids using TransFast liposome reagent (Promega) according to the manufacturer's protocol. Two days after transfection, COS-7 cells were harvested in a lysis buffer (1% Nonidet P-40, 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 4 µg/ml leupeptin, and 1 µg/ml aprotinin). The lysate was centrifuged, and the protein concentration of each supernatant was determined by Coomassie protein assay reagent (Pierce). Proteins (200 µg) were mixed with protein G-Sepharose beads (Amersham Biosciences) for 2 h at 4 °C. After centrifugation, supernatants were incubated with the primary antibody by slow rotation at 4 °C overnight and mixed with protein G-Sepharose beads at 4 °C for 2 h. Immunoprecipitated proteins were washed three times with the lysis buffer. The precipitated proteins were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. Then, the membranes were incubated with indicated antibodies followed by incubation with horseradish peroxidase-conjugated secondary antibodies. Immunoprecipitated proteins were detected by chemiluminescence (ECL; Amersham Biosciences). To show the endogenous association of ARIP2 and ActRIIs, mouse liver was homogenized in 5 volumes of homogenization buffer (0.15 M NaCl, 2 mM EDTA (pH 7.2), 5 mM benzamide-HCl (pH 7.2), 2 mM N-ethylmaleimide, 2 mM phenylmethylsulfonyl fluoride, 2 mM diisopropyl fluorophosphate) and centrifuged at 600 × g for 10 min at 4 °C. The supernatant was then centrifuged at 10,000 × g for 30 min. Next, the pellet was resuspended in five volumes of the homogenization buffer and centrifuged at 10,000 × g for 30 min at 4 °C. The pellet was lysed in a lysis buffer and immunoprecipitated with anti-ActRIIA, anti-ActRIIB, or anti-ARIP2 antibody. The precipitated proteins were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. Then the membranes were probed with the anti-ARIP2 antibody. To study the endogenous association of ARIP2, ActRIIs, and RalBP1, mouse brain lysate was prepared as above and immunoprecipitated with either anti-ActRIIA or anti-ActRIIB. The precipitated proteins were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. Then the membranes were probed with either anti-ARIP2 or anti-RalBP1 antibody.

Antibody Production-- To make polyclonal anti-ARIP2 antibodies, GST fusion proteins of COOH-terminal ARIP2 (nucleotides 427-588) were purified and used for immunization of New Zealand White rabbits. ARIP2-specific antisera were purified by passing through protein A-Sepharose (Amersham Biosciences) and used for immunoprecipitation.

Northern Blotting-- Northern blot filter with 2 µg of poly(A)+ RNA from various adult mouse tissues (mouse Multiple Tissue Northern blot, CLONTECH) was hybridized with cDNA probe containing either a full-length (nucleotides 127-850) or the COOH-terminal fragment (427-850) of ARIP2. Each probe was made by PCR from ARIP2 cDNA. Hybridization, washing, and detection were performed as described previously (17).

Activin-responsive Promoter Assay-- The CAGA-lux construct has been described previously (26). CAGA-lux, cytomegalovirus--galactosidase, and various amounts of either pcDNA3-ARIP2, pcDNA3-ARIP2PDZ, or pcDNA3-ARIP2C were introduced into CHO-K1 cells using TransFast liposome reagent. Stimulation by activin and measurement of luciferase activity were performed as described previously (6).

Activin Binding and Internalization Assay-- CHO-K1 cells grown in 6-well plates were transfected with the indicated plasmids using TransFast liposome reagents. To examine the surface activin binding activity of the cells, transfected cells were washed three times with cold binding medium (minimum essential  medium containing 20 mM HEPES-NaOH (pH 7.4) and 0.1% bovine serum albumin) and incubated in 800 µl of cold binding medium containing 15 ng of ¹²⁵I-activin A for the indicated time periods on ice. Then unbound ligands were removed by washing three times with cold PBS(), the cells were solubilized in 1 N NaOH, and the radioactivities of the cells were counted in a  counter. To study the internalization of ActRIIs, transfected cells were washed three times with binding medium and incubated in 800 µl of binding medium containing 15 ng of ¹²⁵I-activin A at 37 °C, 5% CO₂ for the indicated time periods. After incubation, the plates were placed on ice, washed three times with cold PBS(), and then incubated with cold acid washing buffer (0.2 M acetic acid and 0.5 M NaCl) for 5 min on ice. Then the cells were washed three times with cold PBS() and solubilized in 1 N NaOH, and the radioactivities of the lysates were counted in a  counter. Cell-associated radioactivity after the acid wash represented internalized ActRIIs during incubation at 37 °C. The amount of surface ActRIIs was determined by incubating the cells in 800 µl of cold binding medium with 15 ng of ¹²⁵I-activin A for 4 h on ice followed by three washes with cold PBS(). Internalization of ActRIIs was determined as the ratio of internalized ActRIIs/surface ActRIIs. Nonspecific binding and internalization activities were determined in the presence of a 100-fold excess of unlabeled activin A.

Immunohistochemistry-- CHO-K1 cells were grown on coverslips and transfected with indicated plasmids by TransFast liposome reagent. The next day coverslips were fixed and permeabilized as described previously (17). The cells were double-immunostained with goat anti-ActRIIA antibody to detect ActRIIs and FLAG-ARIP2 or FLAG-ARIP2C. Then the cells were incubated with fluorescein-conjugated secondary antibodies. Fluorescent images of the cells were captured using a confocal microscope (Zeiss 510).

Isolation of GST-Ral Binding Domain of RalBP1-- For purification of GST-RalBD, protein expression was induced in BL21 protease-negative bacteria strain using 0.2 mM isopropyl-1-thio-D-galactopyranoside for 4 h at 37 °C. Isopropyl-1-thio-D-galactopyranoside-treated bacteria were collected and lysed in ice-cold PBS() containing 1% Triton X-100 and protease inhibitors. The lysate was sonicated 3 times for 20 s and centrifuged at 10,000 × g for 20 min to remove insoluble material. Cleared lysate was incubated with glutathione-Sepharose beads (Amersham Biosciences), and GST-RalBD protein was eluted from the beads in buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10% glycerol, and 10 mM glutathione. The eluted protein was dialyzed for 20 h in the same buffer without glutathione.

Ral-GTP Precipitation Assays-- The GTP-bound form of Ral was isolated using GST-RalBD and subsequently quantified, as previously described (27). CHO-K1 cells were transfected with pCGN-HA-RalB and then 1 day after transfection were serum-starved for 24 h. The cells were then treated with or without 50 ng/ml activin A for the indicated time periods and lysed in Ral binding buffer (10% glycerol, 1% Nonidet P-40, 50 mM Tris-HCl (pH 7.4), 200 mM NaCl, 2.5 mM MgCl₂, and protease inhibitors). Lysates were clarified by centrifugation, and the supernatants of each sample were adjusted to the same protein concentration with Ral binding buffer in a final volume of 0.5 ml. Samples were incubated with 15 µg of GST-RalBD precoupled with glutathione beads and incubated for 2 h with rotation at 4 °C. Beads were washed four times in Ral binding buffer, and the precipitated proteins were separated by SDS-PAGE and then transferred onto a polyvinylidene difluoride membrane. The membranes were then probed with anti-HA antibody.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cloning of ARIP2-- Using yeast two-hybrid screening, we searched for proteins that specifically interacted with the cytoplasmic region of ActRIIA. Thirty candidate clones were obtained and sequenced. In a previous study, we reported the characterization of one of the proteins, called ARIP1. ARIP1 is a large protein that has two WW domains and five PDZ domains and is able to associate with ActRIIA and Smad (17). In this study, a clone encoding a novel PDZ protein was selected. We refer to the protein as ARIP2. A full-length cDNA for ARIP2 was isolated by screening mouse brain cDNA libraries. A full-length cDNA clone composed of 862 bp was obtained (Fig. 1A). The ARIP2 cDNA encoded a protein of 153 amino acids and contained a single PDZ domain in the NH₂-terminal region. Data base searches for ARIP2-related sequences revealed the presence of a highly homologous sequence in expressed sequence tags in human tissue, which is likely to be a human counterpart of ARIP2 (Fig. 1B).

View larger version (32K): [in this window] [in a new window]   Fig. 1.   Primary structure of ARIP2. A, nucleotide and amino acid sequences of mouse ARIP2. The predicted amino acid sequence is shown below in the single-letter code. A PDZ domain of mouse ARIP2 is indicated by the underline. Nucleotides are numbered on the right. The sequence is available form GenBankTM under accession number AF414433. B, an alignment of mouse ARIP2 and a putative human homologue. The partial amino acid sequences of putative human ARIP2 are derived from the composite sequences of human EST. The amino acid residues shared by mouse and human homologues are shown by shaded boxes. Missing amino acids are indicated by dashes. Amino acid residues of mouse and human ARIP2 are numbered on the right.

ARIP2 Interacts with COOH-terminal Amino Acids of ActRIIs through PDZ Domain-- In a previous study (17) we reported that ActRIIs have a class I consensus PDZ binding motif at the COOH-terminal region and have an ability to interact with a PDZ protein, ARIP1. We first tested by immunoprecipitation assay whether the specific interaction with ActRIIs is mediated by the PDZ domain of ARIP2 in COS-7 cells. Lysates of COS-7 cells cotransfected with FLAG-ARIP2 and either ActRIIA or ActRIIB were immunoprecipitated with an anti-ActRIIA or anti-ActRIIB antibody. The coimmunoprecipitated FLAG-ARIP2 was detected by an anti-FLAG antibody. (Fig. 2A, both of the top panels, lane 1). Similarly, FLAG-ARIP2C was coimmunoprecipitated in cotransfected COS-7 cells (Fig. 2A, both of the bottom panels, lane 1). By contrast, FLAG-ARIP2PDZ was not coimmunoprecipitated in cotransfected cells (Fig. 2A, both of the middle panels, lane 1). These data confirmed that ARIP2 interacts with ActRIIs through its PDZ domain. Because ActRIIs have consensus PDZ binding motifs at COOH-terminal regions, we next tested the interaction of COOH-terminal deletion mutants of ActRIIs with ARIP2 by mammalian two-hybrid assay. As shown in Fig. 2B, ARIP2 did not interact with mutated ActRIIs that lacked the COOH-terminal ESSL (ActRIIA) or ESSI (ActRIIB). Thus, the COOH-terminal amino acids of the receptors are required for the association of ActRIIs with ARIP2. Thus, we concluded that the PDZ domain of ARIP2 interacts with the COOH-terminal regions of ActRIIs, which have consensus PDZ binding motifs.

View larger version (33K): [in this window] [in a new window]   Fig. 2.   ARIP2 interacts specifically with ActRIIs among various type II receptor serine/threonine kinases by PDZ domain. A, interactions of wild-type and deletion mutants of ARIP2 with ActRIIs (left panels for ActRIIA; right panels for ActRIIB) analyzed by immunoprecipitation assay in COS-7 cells. COS-7 cells were transfected with the indicated plasmids (in both of the top panels, FLAG-ARIP2 (lane 2), ActRII (lane 4), FLAG-ARIP2 and ActRII (lanes 1, 3, and 5)), immunoprecipitated (IP) with anti-ActRIIA (IIA) or anti-ActRIIB (IIB) antibody (lanes 1 and 2) or anti-FLAG (F) antibody (lanes 3 and 4), separated by SDS-PAGE, blotted onto membranes, and probed with anti-FLAG antibody. Similar experiments were performed in cells cotransfected with FLAG-ARIP2PDZ and ActRIIs (both of the middle panels) or with FLAG-ARIP2C and ActRIIs (both of the bottom panels). In lane 5 of all panels, total lysates of cotransfected COS-7 cells were analyzed. Molecular mass markers are indicated on the right. B, interactions of ARIP2 with COOH-terminal deletion mutants of ActRIIs and various serine/threonine kinase receptors analyzed by mammalian two-hybrid assay in CHO-K1 cells. Relative interactions are shown on the left. In the inset, expression of ActRIIA and ActRIIB in CHO-K1 cells were analyzed. CHO-K1 cells were transfected with Bind-ActRIIA or Bind-ActRIIB. Cell lysates were separated by SDS-PAGE and probed with anti-GAL4 antibody. C, interactions of ARIP2 with ActRIIs analyzed by immunoprecipitation assay in mouse liver. A lysate of mouse liver was immunoprecipitated either with anti-ActRIIA (lane 1), anti-ActRIIB (lane 3), or anti-ARIP2 antibody (lane 5), separated by SDS-PAGE, blotted onto membrane, and probed with anti-ARIP2 antibody. In lane 7, total lysate was analyzed. In lane 9, lysate of COS-7 cells transfected with pcDNA3-ARIP2 was analyzed. In lanes 2 and 4, buffers with IgGs (lane 2, anti-ActRIIA; lane 4, anti-ActRIIB; lane 6, anti-ARIP2) were loaded. In lane 8, only buffer was loaded. Molecular mass markers are indicated on the right.

ARIP2 Interacts Specifically with ActRIIs among Various Type II Receptor Serine/Threonine Kinases-- In a mammalian two-hybrid assay, ARIP2 interacted with both ActRIIA and ActRIIB. By contrast, ARIP2 did not show interaction with either TGF- type II receptor or bone morphogenetic protein type II receptor (Fig. 2B). In addition, ARIP2 also interacted with kinase-deficient ActRIIA and ActRIIB (data not shown). Because ActRIIA and ActRIIB expressed at approximately the same level (Fig. 2B, inset), The relative level of interaction of ARIP2 with ActRIIB is weaker than that with ActRIIA. We did not detect any association of ARIP2 with any type I family of receptor serine/threonine kinases (data not shown). These results indicate that ARIP2 interacts specifically with ActRIIs among serine/threonine kinase receptors for the TGF- superfamily and suggest that ARIP2 may have specific roles in the activin signaling.

ARIP2 Interacts with ActRIIs in Mouse Liver-- To determine whether ARIP2 interacts with ActRIIs in a native environment, we performed immunoprecipitation of ARIP2 from a detergent-soluble extract of mouse liver lysate. As shown in Fig. 2C, either anti-ActRIIA or anti-ActRIIB antibody coimmunoprecipitated ARIP2, which migrates approximately at 28 kDa, whereas the control IgGs did not show any signal. This result indicates that ARIP2 interacts with ActRIIs in vivo.

ARIP2 mRNA Is Widely Distributed in Mouse Tissues-- We performed a series of Northern blot experiments to determine the tissue distribution of ARIP2 mRNA. We used two probes derived from the full coding region and the COOH-terminal region, which did not include the PDZ domain sequence. As shown in Fig. 3, ARIP2 mRNA is widely expressed in various mouse tissues. A probe from the full coding region detected the prominent 0.9-kb signal as well as the large 3.9-kb signal (Fig. 3A), whereas the probe from the COOH-terminal region hybridized only to the small 0.9-kb bands (Fig. 3B). These results indicated that the isolated ARIP2 cDNA corresponded to the small 0.9-kb transcript but not to the large 3.9-kb transcript, which is likely to encode a PDZ domain and may be produced by an alternative splicing event excluding the COOH-terminal region of ARIP2.

View larger version (65K): [in this window] [in a new window]   Fig. 3.   Tissue distribution of ARIP2 mRNA. ARIP2 transcripts in various adult mouse tissues were detected by Northern blot analysis with the probes containing the whole open reading frame (A) or the COOH-terminal region (B). Molecular mass markers are indicated in kilobases in the center.

ARIP2 Has an Inhibiting Effect on Activin-induced Transcriptional Response-- CHO-K1 cells were transfected either with pcDNA3-ARIP2, pcDNA3-ARIP2PDZ, or pcDNA3ARIP2C together with a reporter plasmid, CAGA-lux, and activin-induced luciferase activity was measured. Overexpression of ARIP2 in CHO-K1 cells decreased the activin-induced transcriptional activity in a dose-dependent manner (Fig. 4). By contrast, ARIP2PDZ, which did not interact with ActRIIs, had no effect. However, ARIPC, which is the COOH-terminal deletion mutant of ARIP2 and has an ability to interact with ActRIIs, increased activin-induced transcriptional activity in a dose-dependent manner. These data suggest that ARIP2 has an inhibiting effect on activin-induced transcriptional response and that the COOH-terminal region of ARIP2 is involved in the regulation of activin signaling in a mechanism other than that of interacting with receptors.

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Effect of ARIP2 on activin-induced transcription. ARIP2, ARIP2PDZ, and ARIP2C have differential effects on activin-induced transcription in CHO-K1 cells. CHO-K1 cells were transfected with CAGA-lux, cytomegalovirus--galactosidase, and various amounts of either ARIP2, ARIP2PDZ, or ARIP2C, then incubated with 50 ng/ml activin A for 24 h. The luciferase activity of each cell lysate was measured and normalized to the -galactosidase activity. The values presented are the mean ± S.E. of triplicate determinations.

ARIP2 Regulates Endocytosis of ActRIIs-- Because many PDZ domain-containing proteins have a role in receptor-mediated endocytosis (28-31, 33), we next examined the possibility that ARIP2 is involved in endocytosis of ActRIIs. In transiently ActRIIs-transfected CHO-K1 cells, ¹²⁵I-activin A bound to the cells in a time-dependent manner. The binding was about 20-fold higher than that in non-transfected cells (Fig. 5, A and D). Expression of ARIP2 reduced surface activin binding. However, expression of ARIP2C did not (Fig. 5, A and D). These results indicated that ARIP2 reduced the ActRII cell surface expression levels by the COOH-terminal region. To determine the effect of ARIP2 on ActRII internalization, CHO-K1 cells were cotransfected with ActRIIs and either ARIP2 or ARIP2C and incubated with ¹²⁵I-activin A for the indicated time periods at 37 °C. Expression of ARIP2 increased internalization of ActRIIs by ~20% at 60 min (Fig. 5, B and E). This effect was not observed when ARIP2 was cotransfected with the COOH-terminal deletion mutant of ActRIIs (ActRIIAESSL or ActRIIBESSI), which did not bind to ARIP2 (Fig. 5, C and F). Furthermore, expression of ARIP2C decreased internalization of ActRIIs, but this effect was not observed when either ActRIIAESSL or ActRIIBESSI was used (Fig. 5, B, C, E, and F). Taken together, these results indicated that ActRIIs entered into the cells by interacting with the PDZ domain of ARIP2 and that the COOH-terminal region of ARIP2 also played a role in endocytosis of ActRIIs.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   Effect of ARIP2 on internalization of ActRIIs. A and D, surface activin binding activity. CHO-K1 cells transfected with the indicated plasmids (A: , ActRIIA; , ActRIIA and ARIP2; , ActRIIA and ARIP2C; , empty vector. D: , ActRIIB; , ActRIIB and ARIP2; , ActRIIB and ARIP2C; , empty vector) were treated in 800 µl of cold binding medium with 15 ng of 125I-activin A on ice for the indicated time periods, and then the cells were washed and lysed. The radioactivities of the lysates counted in a  counter. B, C, E, and F, internalization of ActRIIs. CHO-K1 cells were transfected with the indicated plasmids (B: , ActRIIA; , ActRIIA and ARIP2; , ActRIIA and ARIP2C. C: , ActRIIAESSL; , ActRIIAESSL and ARIP2; , ActRIIAESSL and ARIP2C. E: , ActRIIB; , ActRIIB and ARIP2; , ActRIIB and ARIP2C. F: , ActRIIBESSI; , ActRIIBESSI and ARIP2; , ActRIIBESSI and ARIP2C), treated in 800 µl of binding medium with 15 ng of 125I-activin A and incubated at 37 °C for the indicated time periods. The graphs show the internalization of ActRIIs calculated by the ratio of internalized ActRIIs/surface ActRIIs as described under "Experimental Procedures." The values shown are the mean ± S.E. of triplicate determinations.

ARIP2 Translocates ActRIIs to the Perinuclear Region by PDZ Domain-mediated Interaction-- Indirect immunofluorescence was utilized to assess the intracellular localization of ARIP2 and ActRIIs to obtain further evidence for the translocation of ActRIIs by ARIP2. CHO-K1 cells were transiently transfected with ActRIIA and either FLAG-ARIP2 or FLAG-ARIP2C. When expressed with only ActRIIA, ActRIIA immunoreactivity was found both in cytoplasmic and plasma membrane regions (Fig. 6A). However, expressed with ARIP2, ActRIIA immunoreactivity was found to be clustered in the perinuclear region and was not found in the membrane region (Fig. 6, B-D). Intriguingly, the ARIP2 immunoreactivity was significantly colocalized with ActRIIA (Fig. 6D). By contrast, ARIP2C expression did not affect the distribution of ActRIIA (Fig. 6, E-G). Furthermore, when ActRIIAESSL was expressed with ARIP2, the immunoreactivity of ActRIIAESSL was not clustered by ARIP2 but was visualized in the membrane region (Fig. 6, H-J). These data are consistent with the results obtained from the internalization assay and indicate the direct involvement of ARIP2 in intracellular translocation of ActRIIA by PDZ domain-mediated interaction. Similar results were obtained when ARIP2 was expressed with ActRIIB or ActRIIBESSI (data not shown).

View larger version (34K): [in this window] [in a new window]   Fig. 6.   ARIP2 translocates ActRIIA to the perinuclear region by PDZ domain-mediated interaction. CHO-K1 cells were transiently transfected with the indicated plasmids (A, ActRIIA; B-D, ActRIIA and ARIP2; E-G, ActRIIA and ARIP2C; H-J, ActRIIAESSL and FLAG-ARIP2). The cells were fixed and permeabilized, and the immunoreactivities were visualized using fluorescent-conjugated secondary antibodies. Antibodies used are anti-FLAG antibody (C, F, and I), shown as green, and anti-ActRIIA antibody (A, B, E, and H) shown as red. The superposition of images B and C is shown in D, the superposition of images E and F is shown in G, and the superposition of images H and I is shown in J. Bar, 10 µm.

COOH-terminal Region of ARIP2 Interacts with RalBP1-- To examine the underlying mechanism of ARIP2-mediated internalization of ActRIIs, we searched for proteins that specifically interacted with ARIP2 by a yeast two-hybrid screening. More than 100 clones were obtained and sequenced. One clone encoded a COOH-terminal sequence of RalBP1 (amino acids 529-647). RalBP1 is a potential effector protein of Ral, a member of small GTP-binding protein (23). RalBP1 has a Rho-GAP homology domain, a Ral binding domain, and a POB1 binding region (Fig. 7A). The Ral binding domain of RalBP1 binds to the GTP-bound form of Ral but not to the GDP-bound form (23). Importantly, RalBP1 associates with POB1, which forms a complex with endocytic machinery to regulate endocytosis of EGF and insulin receptors (18-22). We first studied whether the COOH-terminal region of RalBP1 interacts with ARIP2 by mammalian two-hybrid assay. As shown in Fig. 7B, the COOH-terminal region of RalBP1 specifically interacts with the COOH-terminal region of ARIP2. Because this region of RalBP1 is the POB1 binding region (19), we next carried out an immunoprecipitation assay to study whether ActRIIs, ARIP2, RalBP1, and POB1 could form a complex in mammalian cells. Lysates of COS-7 cells that were cotransfected with ActRIIA, myc-RalBP1, HA-POB1, and either FLAG-ARIP2 or FLAG-ARIP2C were incubated with an anti-ActRIIA antibody. The immunoprecipitated proteins were probed with anti-FLAG, anti-Myc, or anti-HA antibodies. As shown in Fig. 7C, either ARIP2, RalBP1, or POB1 was coimmunoprecipitated with ActRIIA. By contrast, neither RalBP1 nor POB1 coimmunoprecipitated in the lysates of cells that were cotransfected with FLAG-ARIP2C (Fig. 7C, lane 2). These data confirmed that ActRIIs, ARIP2, RalBP1, and POB1 could form a complex in vitro and that the interaction of ARIP2 and RalBP1 did not interfere with the interaction between RalBP1 and POB1. In other words, ARIP2 is capable of linking ActRIIA with a complex of RalBP1 and POB1 through interaction with RalBP1. We obtained similar results when cells were coexpressed with ActRIIB, RalBP1, POB1, and either ARIP2 or ARIP2C (data not shown). To determine whether ActRIIs, ARIP2, and RalBP1 form a complex in vivo, we performed an immunoprecipitation assay from a detergent-soluble extract of mouse brain lysate. As demonstrated in Fig. 7D, ActRIIA coimmunoprecipitated both ARIP2 and RalBP1, indicating that ActRIIA, ARIP2, and RalBP1 form a complex in vivo. Similar results were obtained when the lysate was immunoprecipitated with anti-ActRIIB antibody (data not shown). Thus, ActRIIs, ARIP2, and RalBP1 could form a complex in vivo.

View larger version (31K): [in this window] [in a new window]   Fig. 7.   ARIP2 interacts with RalBP1 and forms a complex with ActRIIA, RalBP1, and POB1 in vitro. A, schematic representation of RalBP1. B, ARIP2 interacts with RalBP1. Interactions of the COOH-terminal region of RalBP1 (amino acids 499-647) with ARIP2 or ARIP2C were analyzed by mammalian two-hybrid assay in CHO-K1 cells. Interaction levels are shown as relative luciferase activities. The expression vectors used were pACT-RalBP1-(499-647), pBIND-ARIP2, and pBIND-ARIP2C. The values presented are the mean ± S.E. of triplicate determinations. C, ARIP2 forms a complex with ActRIIA, RalBP1, and POB1 in COS-7 cells. COS-7 cells were transfected with ActRIIA, Myc-RalBP1, HA-POB1, and either with FLAG-ARIP2 (lane 1) or FLAG-ARIP2C (lane 2). The lysates were immunoprecipitated (IP) with anti-ActRIIA antibody, separated by SDS-PAGE, blotted onto membranes, and probed with indicated antibodies. D, ARIP2 forms a complex with ActRIIA and RalBP1 in mouse brain. Lysate of mouse brain was immunoprecipitated with anti-ActRIIA antibody (lane 1), separated by SDS-PAGE, blotted onto membrane, and probed with anti-ARIP2 antibody. In lane 2, total lysate was analyzed.

The Ral/RalBP1 Pathway Regulates ARIP2-mediated Internalization of ActRIIs-- We first investigated the effect of interaction between ARIP2 and RalBP1 on ARIP2-mediated internalization of ActRIIA. CHO-K1 cells were transfected with ActRIIA, ARIP2, and a series of deletion mutants of RalBP1, and then an internalization assay was performed. Expression of either full-length RalBP1 or RalBP1-(364-647), which lacked Rho-GAP homology domain, did not affect ARIP2-mediated ActRIIA internalization. However, expression of RalBP1-(499-647), which lacked both the Rho-GAP homology and Ral binding domains, significantly reduced ARIP2-mediated ActRIIA internalization (Fig. 8A).

View larger version (28K): [in this window] [in a new window]   Fig. 8.   The Ral/RalBP1 pathway regulates ARIP2-mediated ActRII internalization. A and B, effects of RalBP1 and Ral on ARIP2-mediated ActRII internalization. CHO-K1 cells transfected with the indicated plasmids (A; , ActRIIA and ARIP2; , ActRIIA, ARIP2, and RalBP1; , ActRIIA, ARIP2, and RalBP1-(364-647); , ActRIIA, ARIP2, and RalBP1-(499-647). B: , ActRIIA and ARIP2; , ActRIIA, ARIP2, and wild-type Ral; , ActRIIA, ARIP2, and RalG23V; , ActRIIA, ARIP2, and RalS28N) were treated in 800 µl of binding medium with 15 ng of 125I-activin A and then incubated at 37 °C for the indicated time periods. The graphs show the internalization of ActRIIA calculated by the ratio of internalized ActRIIA/surface ActRIIA as described under "Experimental Procedures." The values shown are the mean ± S.E. of triplicate determinations. C, activation of Ral by activin signaling. CHO-K1 cells were transfected with HA-Ral, then treated with 50 ng/ml activin A for the indicated time periods. The GTP-bound form of Ral precipitated with GST-RalBD (upper panel) and total lysate (lower panel) were analyzed by immunoblotting with anti-HA antibody. D, calcium-dependent activation of Ral by activin signaling. CHO-K1 cells were transfected with HA-Ral. Before activin stimulation, the cells were pretreated with (lane 2) or without (lane 1) 30 µM BAPTA-AM for 30 min. Then the cells were treated with 50 ng/ml activin A for 15 min. The GTP-bound form of Ral associated with GST-RalBD was purified and analyzed by immunoblotting with anti-HA antibody. In the bottom, total lysate was analyzed.

To clarify the possible role of Ral in ARIP2-mediated internalization of ActRIIA, we next performed an internalization assay in cells cotransfected with ActRIIA, ARIP2, and either wild-type Ral, Ral^G23V (the GTP-bound form), or Ral^S28N (the GDP-bound form). As shown in Fig. 8B, expression of wild-type Ral did not affect ARIP2-mediated internalization of ActRIIA, whereas expression of either Ral^S28N or Ral^G23V resulted in an ~20% reduction of ARIP2-mediated ActRIIA internalization. Because only the GTP-bound form of Ral could bind to the Ral binding domain of RalBP1, the most likely explanation for these results is that the proper GDP-GTP exchange of Ral is critical for ARIP2-mediated internalization of ActRIIs. We have obtained similar results when internalization of ActRIIB by ARIP2 was studied (data not shown). Thus, these results indicate that the Ral/RalBP1 pathway regulates ARIP2-mediated internalization of ActRIIs and suggest that the GDP-GTP exchange of Ral is critical for this regulation.

Activin Signaling Activates the GDP-GTP Exchange of Ral by a Calcium-dependent Pathway-- Because the GDP-GTP exchange of Ral is critical for ARIP2-mediated internalization of ActRIIs, we next studied whether activin signaling affected the GDP-GTP exchange of Ral. As shown in Fig. 8C, activin stimulation significantly augmented association of Ral with the GST-RalBD. Because only the GTP-bound form of Ral associates with the Ral binding domain of RalBP1, this result indicated that activin signaling accelerated the GDP-GTP exchange of Ral and resulted in augmented association of Ral with GST-RalBD. In addition, we demonstrated that pretreatment with an intracellular calcium chelator, BAPTA-AM, reduced activin-mediated Ral activation (Fig. 8D). Thus, activin signaling can activate the GDP-GTP exchange of Ral by a calcium-dependent pathway.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We have isolated and characterized a novel mouse PDZ protein, ARIP2, which interacts with ActRIIs by the PDZ domain. Importantly, ARIP2 regulates the translocation of ActRIIs and controls endocytosis of ActRIIs in a small G protein Ral-dependent manner. This is the first report of a molecule that regulates endocytosis of ActRIIs. In a previous study, we reported the characterization of a PDZ protein, ARIP1, which has five PDZ domains and two WW domains. ARIP1 associates with ActRIIA and with Smad3 and regulates activin signaling in neuronal cells (17). Different from ARIP1, ARIP2 interacts with both ActRIIA and ActRIIB. In addition, ARIP2 interacts with RalBP1 at its COOH-terminal region, forming a complex with ActRIIs, RalBP1, and POB1, and then regulates the endocytosis of ActRIIs. We concluded that ARIP2 is a scaffolding protein that unites ActRIIs with the regulatory proteins required for endocytosis.

Internalization assay and immunohistochemical analysis indicated the direct involvement of ARIP2 in ActRII endocytosis. Because ActRIIs are primary ligand binding receptors and are constitutively activated serine/threonine kinases (9, 11), the cell surface levels of ActRIIs must be tightly controlled. Indeed, overexpression of ActRIIs enhances functional association with ActRIB, resulting in ligand-independent activation (Ref. 8 and data not shown). In animal cap cells of Xenopus embryo, which sense different concentrations of activin to form different cell types, the cells recognize activin concentrations by counting the absolute number of occupied receptors but not by sensing a change in the ratio of occupied to unoccupied receptors (34). These findings suggest that the cell surface level of ActRIIs is a key factor determining the cellular response to activin. Furthermore, recent studies using chimeric granulocyte macrophage-colony stimulating factor (GM-CSF)/TGF- receptors indicated that homodimers of TGF- type II receptor induced by GM-CSF stimuli are internalized in AKR-2B fibroblast cells (35, 36). Thus, it is reasonable that pathways regulating the endocytosis of not only heteromeric ActRIIs·ActRIB complexes but also homomeric ActRIIs complexes do exist. In the present study, we performed a series of internalization assays using cells transfected with ActRIIs. We showed that ARIP2 regulated the endocytosis of ActRIIs via PDZ domain-mediated interaction (Figs. 5, 6, and 8). Because CHO-K1 cells express ActRIB endogenously (data not shown), it remains to be elucidated whether ARIP2 regulates the endocytosis of homomeric ActRIIs complexes or heteromeric complexes composed of ActRIIs and ActRIB. Different from the full-length ARIP2, the COOH-terminal deletion mutant of ARIP2 (ARIP2C) failed to induce ActRII endocytosis (Fig. 5, B and E). Although we do not know the mechanism of ARIP2C-mediated reduction of ActRII endocytosis, it is possible that once the PDZ domain of ARIP2 binds to ActRIIs, the endocytosis of ActRIIs is triggered by the Ral/RalBP1 pathway. Because ARIP2C cannot bind RalBP1 and the Ral/RalBP1 pathway directly regulates ARIP2-mediated endocytosis of ActRIIs, it is highly likely that ARIP2C was not capable of initiating Ral/RalBP1-dependent ActRII endocytosis. Interestingly, we found one COOH-terminal splicing variant of ARIP2 equivalent to ARIP2C (data not shown).

Our reporter assays suggest a functional role of ARIP2-mediated ActRII endocytosis in activin signaling. The dose-dependent decrease of signal transduction with increasing ARIP2 cDNA clearly indicated that ARIP2 interfered with activin signaling (Fig. 4). Taken together with the data obtained by internalization assay and immunohistochemical analysis, our finding suggests that increased amounts of ARIP2 decrease the expression levels of ActRIIs in the plasma membrane, resulting in a decreased response to the ligands. By contrast, the dose-dependent increase in signal with increasing ARIP2C cDNA may result from an increased number of ActRIIs bound to ligands on the membrane. Dyson and Gurdon (34) report that in Xenopus blastula cells, activin receptors that bound to ligands are not internalized and continue signaling at the membrane. Whether ARIP2 reduces activin signal transduction and increases internalization of ActRIIs under the physiological condition remains to be determined. Nevertheless, it is a fascinating idea that ARIP2 could deactivate activin signaling by regulating the endocytosis of ActRIIs.

In addition to a Ral binding domain, RalBP1 also contains a Rho-GAP homology domain and exhibits GAP activity toward Rac1 and CDC42 but not RhoA (37). Rho subfamily members are also involved in receptor-mediated endocytosis (38, 39). However, it is unlikely that the Rho-GAP homology domain of RalBP1 is involved in endocytosis of ActRIIs since expression of RalBP1-(364-647), which lacked the Rho-GAP homology domain did not affect ARIP2-mediated ActRII endocytosis (Fig. 8A). By contrast, expression of RalBP1-(499-647), which lacked both the Rho-GAP homology domain and the Ral binding domain, reduced ARIP2-mediated ActRII endocytosis (Fig. 8A). Furthermore, expression of both Ral^G23V (the GTP-bound form) and Ral^S28N (the GDP-bound form) inhibited ARIP2-mediated endocytosis of ActRIIs (Fig. 8B). These results are in complete agreement with the effects of the Ral in EGF and insulin receptor endocytosis (18) and indicate that the GDP-GTP exchange of Ral is critical for the regulation of receptor-mediated endocytosis via the Ral/RalBP1-dependent pathway. Similarly, both GTP- and GDP-bound forms of Rab2, another small GTP-binding protein, inhibit vesicle transport from the endoplasmic reticulum to the Golgi complex (40). We have shown that activin signaling activates the GDP-GTP exchange of Ral in a calcium-dependent manner (Fig. 8, C and D). Although Ral-GDP dissociation stimulator (RalGDS) activates the GDP-GTP exchange of Ral in a Ras-dependent manner (41, 42), Ras-independent and calcium-dependent activation of Ral was also reported (43). Activin has been demonstrated to increase intracellular calcium concentrations in the multiple types of cells (44-47). Thus, it is likely that ARIP2-mediated endocytosis of ActRIIs occurs through a ligand- and calcium-dependent activation of Ral. In the case of EGF and insulin receptor internalization, the Ral/RalBP1 pathway also has been demonstrated to act in a ligand-dependent process (18).

In the present study, we reported that a PDZ protein, ARIP2, regulated endocytosis of ActRIIs through the Ral/RalBP1-dependent pathway. Recent studies suggest existence of other endocytotic pathways regulating endocytosis of ActRIIs. Ehrlich et al. (48) report that TGF- type II receptor is constitutively endocytosed by a di-leucine-based internalization signal. We have found that both ActRIIA and ActRIIB contain the di-leucine-based internalization sequence as well. Whether this sequence regulates endocytosis of ActRIIA and ActRIIB is not known. However, because only ActRIIs have the PDZ binding sequence among the receptors for the TGF- family, the endocytosis and sorting of ActRIIs are likely to be regulated uniquely by PDZ proteins. It is also likely that ActRII-binding proteins other than ARIP2 may regulate endocytosis of ActRIIs. Sorting nexins (SNXs), originally identified as homologues of yeast proteins including vacuolar sorting protein 5 (Vps5p) and multicopy suppressor of vacuolar sorting protein 1 (Mvp1p), are reported to interact with wild-type and kinase-deficient TGF-/activin receptors and thought to play roles in TGF-/activin receptor endocytosis (49). Although overexpression of SNXs reduced TGF-/activin-induced transcription, the definite function of SNXs in the TGF-/activin receptor endocytotic pathway is not fully determined. TGF- receptor-associated protein 1 (TRAP1) may have similar functions as SNXs. TRAP1 also has an ability to interact with wild-type and kinase-deficient TGF-/activin receptors and acts as a Smad4 chaperone (50). TRAP1 shows a 25% identity and 40% similarity to the human ortholog of the yeast vacuolar sorting protein 39 (Vps39)/vacuole morphology protein 6 (Vam6p). Thus, it is suggested that TRAP1 may play a role in lysosome biogenesis (50). Different from SNXs and TRAP1, ARIP2 specifically interacts with ActRIIs. Therefore, endocytosis of ActRIIs is uniquely regulated by ARIP2, and ARIP2-mediated endocytosis of ActRIIs may have specific roles in activin signaling.

Activins have been characterized to act as morphogens in Xenopus animal pole blastomeres by inducing a range of mesodermal tissues in a concentration-dependent manner (51-53). Although passive diffusion can form an activin gradient (54), recent studies highlight the role of endocytosis in shaping morphogen gradients. Successive rounds of endocytosis and resecretion have been proposed as one model for morphogen movement through tissue (55). A study using biologically active Drosophila bone morphogenetic protein homolog decapentaplegic (dpp)-green fluorescent protein has shown that endocytosis of dpp is required for gradient formation of dpp (56). Furthermore, changing the ratio between degradation versus recycling of the endocytosed ligand would determine the eventual shape of the morphogen gradient. Hence, it is also noteworthy that Dubois et al. (57) showed that the wingless gradient between posterior and anterior cells in the engrailed domain of Drosophila embryo is due to the more rapid degradation of wingless proteins in posterior cells. Taken together, these observations show that regulation of receptor-mediated endocytosis of ligand is a key factor shaping and controlling the morphogen gradient. ARIP2 regulates the endocytosis of activin by ActRIIs-mediated internalization (Fig. 5), and ActRIIs are shared by nodal, which also acts as a morphogen in the zebrafish embryos (32, 58). Thus, ARIP2 is likely to be one of the candidate molecules that plays roles in shaping activin and nodal gradients by regulating the endocytosis of ActRIIs.

	ACKNOWLEDGEMENTS

We thank Dr. C.-H. Heldin for the CAGA-lux plasmid, Dr. Y. Eto for recombinant human activin A, Dr. Y. Hasegawa for bovine activin A. We also thank Drs. H. Shoji, T. Nakamura, K. Sugino, and T. Murakami for helpful discussions. We also thank The Institute for Genome Research, University of Tokushima for use of Zeiss 510 confocal microscope.

	FOOTNOTES

^* This research was supported by the Ministry of Education, Science, Sports, Culture, and Technology of Japan and also by grants from The Inamori Foundation for Research and Kyowa Hakko Kogyo Co., Ltd.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.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EBI Data Bank with accession number(s) AF414433.

Present address: Dept. of Animal Reproduction, National Institute of Animal Industry, Ministry of Agriculture, Forestry and Fisheries, Tsukuba, Ibaraki 305-0991, Japan.

^§ Present address: Dept. of Immunology, School of Basic Medical Sciences, University of Jilin, 2 Xinmin St., Changchun 130021, People's Republic of China.

To whom correspondence should be addressed. Tel.: 81-88-633-7439; Fax: 81-88-633-7440; E-mail: tsuchida@ier.tokushima-u.ac.jp.

Published, JBC Papers in Press, March 6, 2002, DOI 10.1074/jbc.M112472200

	ABBREVIATIONS

The abbreviations used are: TGF-, transforming growth factor ; TGFR, TGF- receptor; PDZ, Postsynaptic density 95/Discs large/Zona occludens-1; ActRIIA, activin type IIA receptor; ActRIIB, activin type IIB receptor; ActRIB, activin type IB receptor; ARIP2, activin receptor-interacting protein 2; ARIP1, activin receptor-interacting protein 1; RalBP1, Ral-binding protein 1; POB1, partner of RalBP1; CHO, Chinese hamster ovary; GST, glutathione S-transferase; EGF, epidermal growth factor; PBS, phosphate-buffered saline; SNX, Sorting nexin; TRAP1, TGF- receptor-associated protein 1; HA, hemagglutinin; GAP, GTPase activating protein.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES