Trans-inactivation of receptor tyrosine kinases by novel angiotensin II AT2 receptor-interacting protein, ATIP.

Negative regulation of mitogenic pathways is a fundamental process that remains poorly characterized. The angiotensin II AT2 receptor is a rare example of a 7-transmembrane domain receptor that negatively cross-talks with receptor tyrosine kinases to inhibit cell growth. In the present study, we report the molecular cloning of a novel protein, ATIP1 (AT2-interacting protein), which interacts with the C-terminal tail of the AT2 receptor, but not with those of other receptors such as angiotensin AT1, bradykinin BK2, and adrenergic beta(2) receptor. ATIP1 defines a family of at least four members that possess the same domain of interaction with the AT2 receptor, contain a large coiled-coil region, and are able to dimerize. Ectopic expression of ATIP1 in eukaryotic cells leads to inhibition of insulin, basic fibroblast growth factor, and epidermal growth factor-induced ERK2 activation and DNA synthesis, and attenuates insulin receptor autophosphorylation, in the same way as the AT2 receptor. The inhibitory effect of ATIP1 requires expression, but not ligand activation, of the AT2 receptor and is further increased in the presence of Ang II, indicating that ATIP1 cooperates with AT2 to transinactivate receptor tyrosine kinases. Our findings therefore identify ATIP1 as a novel early component of growth inhibitory signaling cascade.

The potent vasoactive peptide angiotensin II (Ang II) 1 is also an important regulator of cellular proliferation and hypertro-phy. This peptide binds to two main subtypes of receptors (AT1 and AT2) that both belong to the superfamily of G proteincoupled receptors (GPCR) but display opposite biological and physiological effects. The AT1 receptor mediates most of the known cardiovascular and central actions of Ang II. This subtype has mitogenic and trophic effects in many tissues and cell types and transduces multiple intracellular signaling cascades typically associated with GPCR activation. In contrast, AT2 behaves like a "natural antagonist" of the AT1 subtype on most physiological functions, and induces anti-proliferative and proapoptotic effects in vitro and in vivo (for reviews, see Refs. [1][2][3][4][5]. The AT2 receptor activates unconventional signaling pathways that in most cases do not involve coupling to classical regulatory G proteins. A growing body of evidence indicates that anti-growth effects of the AT2 receptor are associated with activation of tyrosine phosphatases and inhibition of protein kinases, which ultimately lead to inhibition of extracellular regulated kinase (ERK2). In many cell types, AT2 is functionally coupled to the Src homology 2 domain-containing tyrosine phosphatase SHP-1 (6 -8). This phosphatase has been shown to play a central role in the AT2 signaling cascades leading to inhibition of AT1-induced PYK2 and Jun kinase (9), AT1-transactivated EGF receptor tyrosine kinase (10), and insulin-induced phosphatidylinositol 3-kinase and Akt activation (11).
AT2 negatively cross-talks with receptor tyrosine kinases (RTK) such as bFGF, EGF, and insulin receptors (10,12,13) by targeting a very early step of RTK activation, i.e. autophosphorylation of the receptor. In vascular smooth muscle cells (VSMC) from AT2-transgenic mice, EGF receptor trans-inactivation induced by AT2 stimulation was found to involve rapid activation of tyrosine phosphatase SHP-1 and its increased association with the EGF receptor (10). In Chinese hamster ovary (CHO) cells, however, AT2-mediated trans-inactivation of the insulin receptor does not involve protein dephosphorylation by orthovanadate-sensitive tyrosine phosphatases nor coupling to pertussis toxin-sensitive regulatory heterotrimeric G i /G o proteins (12), suggesting that another yet undefined mechanism may link AT2 receptor stimulation to growth inhibition.
A number of recent studies have revealed that GPCRs can mediate their intracellular effects through signaling pathways that are independent of G proteins (14,15). Over the past few years, many groups have identified novel intracellular proteins that directly interact with C-terminal tails of GPCRs and function as scaffolds to regulate receptor trafficking or signaling (16,17). ATRAP is one such example of a novel protein that selectively interacts with the AT1 receptor C terminus and down-regulates its activity (18,19). Regarding the AT2 receptor, recent studies have documented a direct interaction of its C-terminal tail with ErbB3, a member of the EGF receptor family (20,21), and with the transcription factor promyelocytic zinc finger containing protein (PLZF) abundantly expressed in the heart (22).
In the present study, we have used the C-terminal part of the AT2 receptor as bait in a two-hybrid system to identify new interacting partners of the receptor. We describe here the molecular cloning and functional characterization of ATIP1, a novel coiled-coil domain containing protein that selectively interacts with the AT2 receptor and mediates inhibition of growth factor-induced ERK2 activation and cell proliferation.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid and cDNA Cloning-The 52 C-terminal residues of the human AT2 receptor were PCR amplified and subcloned into pGBT9 vector in-frame with the Gal4-DNA binding domain. Independent transformants (3 ϫ 10 6 ) from a mouse fetal cDNA library containing inserts of 300 to 700 bp in VP16 vector (a kind gift of Dr. A. Vojtek) were screened by the two-hybrid system cloning method in the HF7 strain as described (23) using the 52 C-terminal residues of the AT2 receptor as a bait. A 354-bp cDNA clone designated ATIP was repeatedly isolated from two independent screenings of the library. Specificity of the interaction was verified with pGBT9 vectors containing lamin or RAS (kind gifts of Dr. J. Camonis) as unrelated baits. cDNA fragments corresponding to the 65 C-terminal residues of the human bradykinin B 2 receptor (kind gift of Dr. W. Mü ller-Esterl), the last 86 residues of the human ␤ 2 -adrenergic receptor (kindly provided by Dr. R. Jockers), or the last 62 residues of the rat AT1 receptor (a kind gift of Dr. K. Bernstein) were PCR amplified and subcloned into pGBT9 to assay for selectivity of the interaction.
For isolation of full-length murine mATIP1 cDNA, the 354-bp ATIP insert was used to screen a mouse fetal cDNA library (3.5 ϫ 10 5 colonies) of large inserts constructed into the pcDNA1 expression vector (24). For isolation of the human hATIP1 homolog, the 3Ј region (last 755 bp) of mATIP1 cDNA was used to screen 3 ϫ 10 5 colonies of a human lung cDNA library in pcDNA3 (Invitrogen). For isolation of full-length human hATIP2 cDNA, reverse transcriptase-PCR amplification was performed on poly(A ϩ ) from human myometrium with specific oligonucleotides: 5Ј-cgggatccgtatccagggctcatgttcacttg-3Ј (sense) and 5Ј-ccgctcgagtgctgatatacctcttgtgcccac-3Ј (antisense). The resulting cDNA fragment (1.37 kb) was subcloned into the BamHI and XhoI sites of the pcDNA3 vector (Invitrogen) and entirely sequenced.
Cell Lines and Transfections-Chinese hamster ovary cells expressing the human AT2 receptor (CHO-hAT2 cells) have been described elsewhere (12). These cells were stably transfected using Dosper liposomal transfection reagent (Roche Diagnostics) with the 354-bp ATIP-ID cDNA fragment subcloned into pcDNA3 (clones C11 and C12), or full-length mATIP1 cDNA in pcDNA3 (clones L11 and L14), or pcDNA3 vector alone (clones V11 and V13). Positive clones resistant to G418 (800 g/ml) were analyzed by immunoblotting using anti-ATIP polyclonal antibodies. AT2 receptor binding sites were analyzed by radioligand binding performed on whole cells (8 ϫ 10 4 cells/well in 24-well dishes) using 0.25 nM [ 125 I-Sar,Ile]Ang II (PerkinElmer Life Sciences) in the presence or absence of Ang II (1 M) or CGP 42112 (1 M), as described (25). All clones were used at passages 5 to 20. COS-hAT2 cells permanently expressing the human AT2 receptor have been previously described (26) and were used at passages 3 to 16. Adult rat VSMC were previously shown to express both AT1 and AT2 receptors (25). These cells were prepared as described (25) and used at passages 2 to 8.
For co-immunoprecipitation experiments, CHO-hAT2 cells stably transfected with either pcDNA3 or full-length mATIP1-Myc were starved in serum-free medium for 24 h and treated for 5 min at 37°C with or without Ang II (100 nM). Cell lysates (500 g proteins) were immunoprecipitated using 1 g of polyclonal rabbit anti-AT2 antibody (sc-9040, Santa Cruz) overnight. Western blots were revealed with anti-Myc monoclonal antibody 9E10 (Santa Cruz). Membranes were stripped and reprobed with goat anti-AT2 antibodies (sc-7421, Santa Cruz) for internal control.
For analyzing expression of endogenous ATIP proteins, whole lysates were prepared as described (27) from human tissues (kindly provided by Dr. Pascal Pineau, Institut Pasteur, Paris, and Dr. Michelle Breuiller-Fouche, Institut Cochin, Paris) and human cell lines. Proteins (20 g) were separated on a 10% SDS-PAGE and immunoblotted using rabbit anti-ATIP polyclonal antibodies (1:5000).
Production of Rabbit Anti-ATIP Polyclonal Antibodies-The 354-bp ATIP-ID cDNA fragment was subcloned into the pRSETA vector (Invitrogen) and the resulting polypeptide: His 6 -ATIP fused to six histidine residues, was purified from bacterial lysates by passage through a nickel column as described by the manufacturer. Purified His 6 -ATIP (100 g) was injected three times intradermally into rabbits at 2-week intervals for production of polyclonal antiserum. Anti-ATIP polyclonal antibodies were affinity purified by passage through glutathione-agarose beads coupled to the glutathione S-transferase-ATIP fusion protein (obtained by subcloning the 354-bp ATIP-ID insert into pGEX-4T1 (Amersham Biosciences) and purification of glutathione S-transferase-ATIP as described by the manufacturer). Specificity of anti-ATIP antibodies was verified by immunoblotting cell lysates from COS cells transfected with each ATIP cDNA. A single polypeptide migrating at the expected molecular weight (18,000, 55,000, 55,000, and 50,000 for ATIP-ID, mATIP1, hATIP1, and hATIP2, respectively) was detected in each case.
Transient Expression and Phosphorylation of Tagged ERK2-For measurement of ERK2 phosphorylation in transient transfections, COS-wt or COS-hAT2 cells were seeded at a density of 3 ϫ 10 5 cells/well in 6-well dishes and transfected with mATIP1 cDNA or empty vector (1 g) in 3 l of FuGENE (Roche) as indicated by the manufacturer. To avoid high background because of stimulation of endogenous ERK2 in non-transfected cells, co-transfections were performed with 0.5 g of a construct ("ERK2-Myc") encoding ERK2 fused to six Myc epitopes (a kind gift of Dr. Sabine Traver, Paris). The resulting ERK2-Myc polypeptide migrates at higher molecular weight (60,000) and can thus be easily distinguished from endogenous ERK2 (42,0000) in Western blot. Forty-eight hours after transfection, cells were starved in serum-free Dulbecco's modified Eagle's medium for 18 h before appropriate treatment with EGF as indicated (12), then lysed in 60 l of Laemmli's sample buffer and analyzed by immunoblotting (20 l) with anti-phospho-ERK antibodies (Cell Signaling). Membranes were stripped and reblotted with anti-Myc antibodies (Santa Cruz) to assess expression levels of ERK2-Myc in each lane.
Measurement of Thymidine Incorporation-DNA synthesis was assayed by measuring [ 3 H]thymidine incorporation as described (25). Stably transfected CHO-hAT2 cells (clones V11, V13, L11, and L14) were seeded at a density of 2 ϫ 10 5 cells/well in 24-well dishes (60% density) in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. On the following day, cells were set in serum-free medium for 24 h to induce quiescence. For VSMC, cells were seeded at a density of 7 ϫ 10 4 cells/well in 24-well dishes and transfected with mATIP1 cDNA or empty vector (0.4 g/well). Forty-eight hours after transfection, cells (80 -90% confluency) were cultured for 48 h in serum-free medium to induce quiescence. For measurement of DNA synthesis, quiescent cells were treated with the indicated growth factor for 40 h and pulsed with 1 Ci/ml [ 3 H]thymidine (PerkinElmer Life Sciences) for an additional 24 h. Cells were washed with ice-cold phosphatebuffered saline and treated as described previously (25) to measure radioactivity of the cell lysate.

Molecular Cloning of ATIP1 from Mouse and Man-
The yeast two-hybrid system cloning method was developed to identify novel interacting partners of the AT2 receptor. The last 52 amino acids of the human AT2 receptor were used as a bait to screen 3 ϫ 10 6 clones of a mouse fetal cDNA library constructed in the pVP16 cloning vector (23). One positive clone containing an insert of 354 bp was repeatedly isolated from two independent screenings of the library. This insert encoded an open reading frame of 118 amino acids that was designated ATIP-ID, for AT2 receptor-interacting protein-(interacting domain). As shown in Fig. 1A, ATIP-ID interacts with the C-terminal intracellular tail of the AT2 receptor but not with that of the AT1 receptor subtype, nor with C-terminal domains of other GPCR such as ␤ 2 -adrenergic or bradykinin B 2 receptors. The interaction between ATIP-ID and the AT2 C terminus was further confirmed by in vitro binding assays using purified domains of each protein (data not shown).
The 354-bp fragment of ATIP-ID was used as a probe to screen a mouse fetal cDNA library constructed in the pcDNA1 expression vector. A full-length cDNA clone of 1803 nucleotides was isolated (GenBank TM accession number AF173380). This clone contains an open reading frame of 1323 bp translating into a 440-amino acid polypeptide that we have designated mouse ATIP1 (mATIP1) (Fig. 1B). Several in-frame stop codons were found upstream of the initiating methionine. The human homolog of mATIP1 was further isolated by screening a human adult lung cDNA library, using as a probe a 755-bp cDNA fragment corresponding to the 3Ј end of the mATIP1 coding sequence. Full-length human ATIP1 (hATIP1) cDNA encompasses 1977 nucleotides (GenBank TM accession number AF293357) and encodes a 436-amino acid polypeptide that shares 86% amino acid sequence identity with mATIP1. While this manuscript was in revision, a nucleotide sequence (MTSG1) identical to hATIP1 was published in the data banks (GenBank TM accession number AF121259).
Mouse and human ATIP1 proteins are mainly hydrophilic and contain no transmembrane domain. The major part of ATIP1 is composed of a large coiled-coil domain (residues 106 to 375 of mATIP1) including two leucine zippers (Fig. 1B). ATIP1 is also characterized by a high proportion of basic residues (16%) and a stretch of 30 C-terminal residues rich in proline, serine/threonine, and arginine ("PSR" region). A BLAST search for homologous proteins in the data banks indicated that ATIP1 is a novel protein that shares 25% amino acid sequence identity with myosins in the coiled-coil region.
Immunoprecipitation experiments were carried out to investigate whether ATIP1 was also able to interact with the AT2 receptor in a cellular context. Full-length mATIP1 cDNA was fused to the Myc epitope and transfected into CHO cells expressing the AT2 receptor. Co-immunoprecipitation experiments (Fig. 1C) revealed constitutive interaction between ATIP1 and AT2, which is not significantly modified upon treatment with Ang II.
A Family of Homologous ATIP Proteins-We then analyzed the tissue distribution of ATIP1 mRNA. The 354-bp fragment of ATIP-ID was used as a probe to hybridize a Northern blot of mRNA from various human tissues ( Fig. 2A). Expression of a 1.9-kb transcript likely corresponding to hATIP1 mRNA was detected in all tissues examined. Additional hybridizing transcripts were detected at 4.2 and 6.9 kb in spleen, prostate, ovary, small intestine, colon ( Fig. 2A), as well as in heart, placenta, skeletal muscle, pancreas, and lung (data not shown), suggesting the existence of additional mRNAs homologous to ATIP1.
Western blot analysis of total lysates from human tissues and cell lines using a polyclonal antibody directed against the ATIP-ID domain ( Fig. 2A, right panel) further confirmed endogenous expression of hATIP1-like proteins. Four major polypeptides of apparent molecular weights 30,000, 60,000, 120,000, and 180,000, respectively, were expressed at variable levels in different tissues. Further work will be required to unambiguously identify each of these polypeptides, and determine the relationship between the various mRNA transcripts and polypeptides detected with the ATIP-ID probe and specific antibodies, respectively.
Nucleotide sequence comparisons with GenBank TM data bases allowed the identification of human cDNA sequences related to hATIP1 in uterus (accession number AL096842), brain (partial sequence, accession number AB033114), and fetal brain (accession number AK125188). All three sequences contained the 354-bp sequence of ATIP-ID and were therefore designated hATIP2 (uterus), hATIP3 (brain), and hATIP4 (fetal brain). Close examination of genomic sequences in the data bases revealed that all four hATIP mRNAs are derived from a single gene by alternative promoter utilization and exon/intron splicing. 2 Translated amino acid sequences of hATIP2, hATIP3, and hATIP4 comprise 415, 1270, and 517 residues, respectively, and are 100% identical to hATIP1 in their last 395 amino acids (Fig. 2B). It is of note that the C-terminal sequence shared by all four hATIP members includes the large coiled-coil domain and leucine zippers as well as the stretch of 118 amino acids (ATIP-ID) that interact with AT2. In the N terminus, hATIP proteins differ both in length and sequence, and exhibit specific motifs that suggest differential intracellular localizations and/or association with distinct cytosolic partners (Fig. 2B). hATIP2 thus exhibits a very short (20 amino acids) N-terminal region that contains a bipartite nuclear localization signal, and one polyproline-rich region (PPXXP) known to play an important role in protein-protein interactions with WW or Src homology 3 domains (28). hATIP3 has a long N-terminal region (874 residues) containing one nuclear localization signal and four polyproline-rich motifs (Fig. 2B). The N-terminal part of hATIP4 (122 amino acids) contains a stretch of 24 hydrophobic amino acids flanked on each side by a charged residue, which is a typical feature of membrane spanning regions. hATIP4 thus likely consists of a transmembrane protein with a short (36 residues) N-terminal extracellular domain and two polyproline-rich motifs at the inner face and close vicinity of the membrane (Fig. 2B).
Dimerization of ATIP Proteins-The presence of a large coiled-coil domain with two leucine zippers in the C-terminal part of all ATIP proteins suggested that these proteins may be able to dimerize. Co-immunoprecipitation experiments were undertaken to investigate this possibility. The cDNAs encoding hATIP1 and hATIP2 were fused either to the Myc or HA epitopes, and co-transfected into COS cells prior to immunoprecipitation with anti-Myc antibodies. As seen in Fig. 3 (left  panel), HA-hATIP2 was revealed in anti-Myc immunoprecipitates from cells co-transfected with Myc-hATIP2 and HA-hATIP2 (lane 3) but not from cells transfected either with Myc-hATIP2 or HA-hATIP2 alone (first and fourth lanes), therefore indicating that hATIP2 homodimerizes inside the cell. Similarly, HA-hATIP1 was detected in anti-Myc immunoprecipitates from COS cells co-transfected with Myc-hATIP1 and HA-hATIP1 (Fig. 3, right panel, fourth lane), but not from cells transfected with either Myc-hATIP1 or HA-hATIP1 alone (first and third lanes), therefore indicating homodimerization of hATIP1. Finally, the ability of hATIP1 and hATIP2 to heterodimerize was demonstrated by co-immunoprecipitation of HA-hATIP1 and Myc-hATIP2 using anti-Myc antibodies (Fig.  3, right panel, second lane).
Inhibitory Effect of mATIP1 on Growth Factor-induced ERK2 Activation-We then sought to examine the functional role of ATIP1, and we first analyzed whether ectopic expression of ATIP1 in eukaryotic cells was able either to block, or mimic, AT2 receptor activation. Our previous studies had demonstrated negative cross-talk between AT2 and growth factor No interaction was detected with empty vector (pLEX9) nor C-terminal regions of rat AT1 receptor (AT1R), human ␤ 2 -adrenergic receptor (hu␤2AR), and human bradykinin receptor B 2 (hBK2). B, amino acid sequence comparison of mouse and human ATIP1 using the one-letter code. Dashes indicate identity. Gaps have been introduced to maximize homology. The coiled-coil domain is boxed. Leucine zippers are indicated in bold. An arrow indicates the position of the C-terminal portion rich in Pro, Arg, and Ser/Thr residues (PSR). C, constitutive interaction between mATIP1 and AT2 receptor expressed in CHO cells. CHO-AT2 cells expressing mATIP1-Myc cDNA, or vector alone, were treated for 5 min in the presence (ϩ) or absence (Ϫ) of 100 nM Ang II and immunoprecipitated (IP) with rabbit anti-AT2 polyclonal antibodies. Western blots (WB) were revealed using monoclonal anti-Myc antibodies and reprobed with goat anti-AT2 antibodies for internal control.
As seen in Fig. 4B, ERK2 phosphorylation induced by a 5-min treatment with increasing doses of insulin was markedly reduced in clones L11 and L14 when compared with V13. Autophosphorylation of the insulin receptor ␤ chain migrating at 97 kDa was also significantly reduced in L11 and L14 as compared with V13 cells (Fig. 4B). These results therefore indicate that, in the same way as previously reported for the AT2 receptor (12), mATIP1 interferes at the initial step of the insulin receptor intracellular cascade (i.e. insulin receptor autophosphorylation), thereby blocking its downstream signaling.
The inhibitory effect of ATIP1 is not solely restricted to the insulin receptor, as ERK2 activation induced by bFGF was also decreased in clones L11 and L14 as compared with V13 (Fig.  4C). To examine the involvement of the AT2-interacting domain of mATIP1 in this effect, the 118-amino acid fragment corresponding to ATIP-ID was transfected into CHO-hAT2 cells and two independent clones (C11 and C12) were isolated and analyzed. Both clones C11 (Fig. 4D) and C12 (not shown) behaved like L11 and L14, showing reduced sensitivity to bFGF and insulin as compared with V13. Taken together, these data indicate that mATIP1 mimics the inhibitory effect of AT2 on growth factor-induced ERK activation and insulin receptor autophosphorylation, and that expression of the sole AT2-interacting domain of ATIP is sufficient to inhibit growth factor receptor-coupled signaling.
The Inhibitory Effect of mATIP1 Requires Expression of the AT2 Receptor-Transient transfections of COS cells were undertaken to further analyze the intracellular effects of mATIP1 expression and avoid any possible bias because of clonal selection of stable transfectants. To eliminate background because of high activation of endogenous ERK2 in response to growth factors, cells were co-transfected with a tagged ERK2-Myc cDNA construct as described under "Experimental Procedures." Experiments were conducted in parallel in wild-type COS cells (COS-wt) and in COS-hAT2 cells that permanently express the human AT2 receptor (26). As shown in Fig. 5 (left  panel), in COS-wt transfected with empty vector, EGF (50 ng/ml) induced early and transient activation of ERK2-Myc being maximal at 10 min. The same time course of ERK2 activation was observed in COS-hAT2 cells transfected with pcDNA3 (Fig. 5, right panel). Transient transfection of mATIP1 into COS-hAT2 cells led to a 50% reduction of EGF-induced ERK2 phosphorylation at 10 min (Fig. 5, right panel), in agreement with our findings on CHO-hAT2 cells (Fig. 4). Transfec- tion of mATIP1 into COS-hAT2 cells did not affect the time course of ERK2 phosphorylation, indicating that mATIP1 attenuates, rather than delays, growth factor-induced ERK2 activation. Interestingly, mATIP1 had no detectable effect on ERK2 activation in COS-wt cells (Fig. 5, left panel), suggesting that the inhibitory effect of mATIP1 requires the presence of AT2 receptors.
Inhibitory Effect of mATIP1 on Cell Proliferation-It is now well established that the AT2 receptor exerts anti-proliferative effects in a number of cell types (1,2,4,25,29), and we thus evaluated whether mATIP1 expression also leads to inhibition of cell proliferation. To address this question, transfected CHO-hAT2 clones were analyzed for DNA synthesis in response to various growth factors. As shown in Fig. 6A, thymidine incorporation induced by increasing doses of bFGF (upper panel) or PDGF (lower panel) was dramatically reduced in L11 cells as compared with V13 cells. Similar results were obtained with clones L14 and V11 (not shown).
mATIP1 Cooperates with AT2 Receptor to Inhibit RTK Signaling-We then analyzed whether mATIP1 can modulate the AT2 receptor signaling pathway. To this end, CHO-hAT2 cells transfected with empty vector (V13), mATIP1 (L14), or ATIP-ID (C11) were treated with insulin in the presence or absence of Ang II. Based on results presented in Fig. 4B, an elevated dose of insulin (1 g/ml) was used in these experiments to elicit detectable activation of ERK2 in clones L11 and C11. In these conditions, Ang II only weakly inhibited insulininduced ERK2 activity in clone V13 (Fig. 7A). This is consistent with our previous studies (12) showing that in CHO-hAT2 cells the inhibitory effect of AT2 receptor is maximal at a low dose (0.05 g/ml) of insulin and poorly detectable at a higher dose (0.5 g/ml) of insulin. ERK2 phosphorylation induced by 1 g/ml insulin was reduced in clones L14 and C11 as compared with V13, in agreement with results shown in Fig. 4B. In mATIP1-expressing clones L14 and C11, Ang II was still able to inhibit insulin-induced ERK2 phosphorylation (Fig. 7A), suggesting that ATIP1 and AT2 work in concert to negatively regulate the insulin receptor pathway.
The effect of mATIP1 on the AT2 response was also investigated in transiently transfected COS-wt and COS-hAT2 cells ( Fig. 7B). As expected, Ang II had no effect on EGF-induced ERK2 activation in COS-wt because these cells do not express the AT2 receptor (Fig. 7B, left panel). In agreement with data presented in Fig. 5, transfection of mATIP1 into these cells did not significantly affect ERK2 activation, either in the presence or absence of Ang II (Fig. 7B, left panel). As seen in Fig. 7B (right panel), in COS-hAT2 cells, AT2 stimulation (lane 3) as well as ATIP1 expression (lane 5) led to inhibition of EGFinduced ERK2 phosphorylation, and the inhibition of ERK2 was further increased by concomitant actions of ATIP1 and Ang II/AT2 receptor (lane 6).
Functional consequences of mATIP1 expression on the AT2 receptor response were further analyzed at the level of DNA synthesis in stably transfected CHO-hAT2 cell lines. As seen in Fig. 7C, Ang II elicited a weak anti-proliferative effect (13.6 Ϯ 4% inhibition, n ϭ 3) in V11 cells treated with 1 g/ml insulin and this inhibitory effect was increased in mATIP1 expressing clones L11 (82.6 Ϯ 13%, n ϭ 4) and L14 (35.7 Ϯ 10%, n ϭ 3). Altogether, these data indicate that mATIP1 cooperates with the AT2 receptor to inhibit growth factor-induced ERK2 activation and cell proliferation, and support a role for ATIP1 as an early determinant of the AT2 receptor inhibitory response.

DISCUSSION
It is now well established that the Ang II AT2 receptor negatively regulates cell proliferation via atypical signaling pathways that mainly involve activation of tyrosine phosphatases and inhibition of protein kinases (1)(2)(3)(4). However, the initial steps of the AT2 signaling cascade leading to growth inhibition still remain to be defined. The present study reports the molecular cloning of a novel AT2 receptor-interacting protein (ATIP1) that selectively interacts with the C-terminal tail of the receptor and inhibits insulin-, bFGF-, and EGF-induced intracellular signaling cascades, in the same way as previously reported for the AT2 receptor (12). Results shown here indicate that ATIP1 constitutively interacts with the AT2 receptor and reveal that expression, but not ligand-induced activation, of the receptor is required for functionality of ATIP1. Furthermore, ATIP1 and AT2 cooperate to inhibit growth factor-induced ERK2 activation and cell proliferation. Altogether these results support a role for ATIP1 as an early mediator of the AT2 receptor inhibitory pathway.
Northern blot experiments and examination of EST sequences in the data banks indicated that ATIP1 is expressed in essentially all tissues examined. This was surprising with regard to the restricted pattern of expression of the AT2 receptor, which is abundantly expressed in fetal tissues but found only at specific sites (adrenal, uterus, brain, and vasculature) in the adult (26,30). The large tissue distribution of ATIP1 suggests that in addition to its role in the AT2 anti-proliferative pathway, ATIP1 may play other important roles inside the cell, which may be broader than those fulfilled by AT2 receptors.
Molecular studies revealed that ATIP1 is the leader member of a family of at least four proteins (ATIP1 to ATIP4) that all exhibit the same C-terminal domain able to interact with the AT2 receptor. N-terminal parts of different ATIP members diverge both in length and sequence, and carry specific motifs for localization to the nucleus (ATIP2 and ATIP3) or spanning the plasma membrane (ATIP4). It will be interesting to determine whether ATIP2, ATIP3, and ATIP4 also participate in AT2 signaling pathways, and whether their functional roles also depend on the presence of AT2 receptors.
What is the mechanism for ATIP1-mediated inhibition of RTK signaling? We first investigated the role of tyrosine phosphatase SHP-1, a major component of the AT2 signaling pathway involved in inhibition of ERK (6 -8), EGF receptor phosphorylation (10), Pyk2 and c-Jun kinase (9), phosphatidylinositol 3-kinase and Akt (11), and JAK2 activity (31). Treatment with 0.1 mM sodium orthovanadate, a potent inhibitor of protein-tyrosine phosphatase activity, had no significant effect on mATIP1-FIG. 7. mATIP1 cooperates with AT2 receptor to inhibit ERK2 and cell proliferation. A, transfected CHO-hAT2 cells were treated for 5 min at 37°C with (ϩ) or without (Ϫ) insulin (1 g/ml) and Ang II (100 nM). Total cell lysates were analyzed by immunoblotting using anti-phospho-ERK antibodies and anti-ERK2 antibodies as described in the legend to Fig. 4. Values shown below are the results of densitometry scanning of ERK2 phosphorylation (anti-phospho-ERK) relative to the amount of endogenous ERK2 in each lane (anti-ERK2). Shown is one representative of three independent experiments. B, COS-wt (left panel) or COS-hAT2 (right panel) were co-transfected with ERK2-Myc and either empty vector or mATIP1 cDNA, as indicated. Treatments were for 5 min at 37°C with EGF (50 ng/ml) in the presence (ϩ) or absence (Ϫ) of Ang II (100 nM). Western blot analysis of total cell lysates and densitometry scanning were performed as described in the legend to Fig. 5. C, transfected CHO-hAT2 cells were treated for 48 h at 37°C with (ϩ) or without (Ϫ) insulin (1 g/ml) and Ang II (100 nM), and [ 3 H]thymidine incorporation was measured as described in the legend to Fig. 6. Results are expressed as mean Ϯ S.E. of three independent experiments performed in triplicate. induced inhibition of ERK activity nor cell proliferation in transfected CHO cells (data not shown), therefore ruling out a major effect of a phosphatase activity, including SHP-1, in the inhibitory effect of ATIP1. We further explored the possibility that ATIP1 may function as a scaffold protein to bring SHP-1 at the vicinity of the AT2 receptor, as previously shown for the G␣S protein (32). However, all attempts to visualize an interaction between ATIP1 and SHP-1, both at basal conditions and after stimulation with Ang II or growth factors, remained unsuccessful (data not shown).
The possibility was also investigated that ATIP1 may directly interact with intracellular parts of RTKs and thus provide a link between AT2 and RTKs. This was suggested by the observation that ATIP1 and AT2 attenuate RTK autophosphorylation, and by recent studies showing that ErbB3, a member of the EGF receptor family, interacts with the AT2 receptor C-terminal portion (20,21). Experiments performed on transfected CHO-hAT2 and COS-AT2 cells failed, however, to detect direct interaction between ATIP1 and the EGF receptor or the insulin receptor ␤-chain, either at basal levels or after stimulation with Ang II, EGF, or insulin (data not shown).
Results presented here show that the inhibitory effect of ATIP1 depends on the presence, but not extracellular activation, of AT2 and that ERK2 inhibition is further increased upon Ang II stimulation in ATIP1-expressing cells. One simple explanation would be that ATIP1 directly targets AT2 receptor expression or activation. Radioligand binding studies performed on transfected CHO-AT2 cells revealed that ATIP1 only slightly modifies AT2 receptor binding parameters K d and B max , suggesting that ATIP1 does not function mainly by increasing AT2 receptor expression, accessibility to the membrane, or affinity to the ligand. The possibility remains that by interacting with the C terminus of AT2, ATIP1 may induce "intracellular activation" of the receptor in a ligand-independent manner. Indeed, ligand-independent effects of the AT2 receptor on apoptosis have been demonstrated in various cell types (33). One main feature common to all ATIP members is the presence of a large C-terminal coiled-coil domain that allows homo-and hetero-dimerization of these proteins. This may suggest a role for ATIP1 in AT2 receptor dimerization, a process of major importance for GPCR regulation and function (34,35). AT2 receptors have indeed been shown to heterodimerize with the AT1 subtype, and thereby inhibit the AT1 intracellular response by a novel mechanism that is independent of ligand/AT2 receptor interaction (36).
Other studies using the two-hybrid system have led to the identification of signaling molecules that interact with the Cterminal tails of other GPCRs including AT1 and AT2 (18 -22), and regulate their functions (14,15,17). Of interest, although not structurally related, Homer proteins display striking similarities with the ATIP family. Homer is a family of non-transmembrane domain proteins that directly interact with the Cterminal tail of some, but not all, metabotropic glutamate receptors and provide a link with inositol 1,4,5-trisphosphate receptors and calcium signaling (37). Furthermore, members of the Homer family exhibit in their C-terminal portion a large coiled-coil domain allowing their dimerization and have been shown to intracellularly activate metabotropic receptors in an agonist-independent manner (38).
Intracellular cross-talk between GPCRs and RTKs has been extensively studied over the past few years and has been shown in most cases to involve EGF receptor trans-activation (39 -41). However, little is known about negative regulation of RTK activity. We report here the cloning of ATIP1, a novel AT2 receptor-interacting protein that functions as an early media-tor of growth-inhibitory signaling cascades. Further characterization of ATIP1 and other members of the ATIP family may provide important insights into novel intracellular pathways that negatively regulate mitogenic signals.