AIP1/DAB2IP, a novel member of the Ras-GAP family, transduces TRAF2-induced ASK1-JNK activation.

Previously we have shown that ASK-interacting protein 1 (AIP1, also known as DAB2IP), a novel member of the Ras-GAP protein family, mediates TNF-induced activation of ASK1-JNK signaling pathway. However, the mechanism by which TNF signaling is coupled to AIP1 is not known. Here we show that AIP1 is localized on the plasma membrane in resting endothelial cells (EC) in a complex with TNFR1. TNF binding induces release of AIP1 from TNFR1, resulting in cytoplasmic translocation and concomitant formation of an intracellular signaling complex comprised of TRADD, RIP1, TRAF2, and AIPl. A proline-rich region (amino acids 796-807) is critical for maintaining AIP1 in a closed form, which associates with a region of TNFR1 distinct from the death domain, the site of TNFR1 association with TRADD. An AIP1 mutant with deletion of this proline-rich region constitutively binds to TRAF2 and ASK1. A PERIOD-like domain (amino acids 591-719) of AIP1 binds to the intact RING finger of TRAF2, and specifically enhances TRAF2-induced ASK1 activation. At the same time, the binding of AIP1 to TRAF2 inhibits TNF-induced IKK-NF-kappaB signaling. Taken together, our data suggest that AIP1 is a novel transducer in TNF-induced TRAF2-dependent activation of ASK1 that mediates a balance between JNK versus NF-kappaB signaling.

differentiation, and apoptosis. TNF signals are initiated by binding to either of two different cell surface receptors, known as TNFR1 (CD122a) and TNFR2 (CD122b). The bestcharacterized signaling pathways are initiated by TNFR1. A current model postulates that in the absence of TNF, TNFR1 is kept in an inactive state through binding of an adaptor protein called silencer of death-domains (SODD). The interaction between TNFR1 and SODD is mediated by association of death domains on both proteins. TNF induces dissociation of SODD from TNFR1 with concomitant recruitment of TNFR-associated death-domain protein (TRADD) (3,4). TRADD functions as a platform adaptor to initiate assembly of a multiprotein complex that activates downstream signaling pathways. Specifically, TRADD recruits both receptorinteracting protein-1 (RIP1) and TNFR-associated factor (TRAF2), which in turn recruit the IB kinase complex (IKK) and a mitogen-activated protein kinase kinase kinase (MAP3K), leading to activation of NF-B and JNK pathways, respectively (5)(6)(7)(8). Recently it has been proposed that TNFR1-TRADD-RIP-TRAF2 complex function as an initial membrane bound complex (complex I) to specifically activate the NF-B cascade. The role of the complex I in the JNK cascade has not been addressed (8). Complex I is rapidly internalized and TRADD rapidly dissociates from TNFR1. After a lag of some several hours, internalized TRADD may recruit Fas-associated death domain protein (FADD) and procaspase-8 to form a cytoplasmic complex (complex II). Complex II promotes autocatalytic activation of pro-caspase-8 to initiate apoptosis unless the NF-B-induced long isoform of the component known as FLICE inhibitory protein (FLIP L) is also present in the complex (9). The events between the dissociation of complex I and the formation of complex II are not well established.
TRAF2 is a member of a family of signal transducing proteins. Six members of TRAF family have been extensively studied (10) and TRAF7 has been recently described (11). JNK and NF-B pathways can be activated by overexpression of TRAF 2, 5, and 6, but not of TRAF 1, 3, and 4, suggesting that different TRAFs, despite their structural homology, might perform very different functions. Studies from various cell types have identified TRAF2 as the bifurcation of two different kinase cascades leading to activation of NF-B and JNK: TRADD-TRAF2/RIP-IKK for NF-B activation, and TRADD-TRAF2-MAP3K-MAP2K for JNK activation (5,6). Recent studies of TRAF2 knockout mice or transgenic mice expressing dominant negative TRAF2 mutant protein have established that TRAF2 is absolutely required for activation of JNK by TNF, although it may be redundant in some cell types for NF-B activation (12,13) due to presence of TRAF5 (14). Studies in transfected cells have shown that the intact RING finger of TRAF2 is critical for TRAF2 function since a TRAF2 mutant with deletion of the N terminus (87-501) behaves as a dominant negative mutant for both NF-B and JNK activation (15). The TRAF2 gene may code for an alternatively spliced form, designated as TRAF-2A, which differs from TRAF2 in that it contains a seven amino acids insertion in its RING domain (16). TRAF-2A fails to activate NF-B pathway while still retains an ability to activate the JNK pathway (17). These results suggest that NF-B and JNK pathways may be differentially regulated by minor alterations in the effector domain of TRAF2 or by modulators that bind to different sites on the TRAF2 RING finger.
The mechanism by which TRAFs activates MAP3Ks remains unclear, and there is some disagreement about which MAP3K is critical for JNK activation. It has been shown that the TRAF domain of TRAF2 and TRAF6 interact with apoptosis signalregulating kinase-1 (ASK1), a member of the MAP3K family that specifically activates a cascade ending with JNK (but not NF-B) activation. The association of TRAFs with ASK1 is required for ASK1 activation (18). ASK1 is a 170-kDa protein containing an inhibitory N-terminal domain, an internal kinase domain, and the C-terminal domain interacting with TRAFs (18,19). The current model for ASK1 activation by TNF involves several critical steps including release of inhibitors (thioredoxin and 14-3-3) (20,21), TRAF-dependent homodimerization/polymerization (22), and ASK1 autophosphorylation at Thr-845 (23). We have recently shown that TNF can induce a TRAF2-dependent association of ASK1 with ASK-interacting protein 1 (AIP1, also called DAB2IP for DAB2-interacting protein), a novel member of the Ras-GAP family (24,25). AIP1 causes 14-3-3 release from ASK1, initiating ASK1-dependent JNK activation (24).
In the present study, we show that AIP1 is localized to the plasma membrane of EC where it associates with TNFR1 through a site distinct from the death domain involved in SODD and TRADD binding. In response to TNF, AIP1 dissociates from TNFR1 with concomitant cytoplasmic translocation and formation of a complex comprised of TRADD, RIP1, TRAF2, and AIP1, which is distinct from the complex I and complex II and which initiates specific activation of the ASK1-JNK pathway. Furthermore, we demonstrate that AIP1 specifically interacts with the effector domain RING finger of TRAF2 to enhance TNF-induced ASK1-JNK but to inhibit IKK-NF-B signaling. Our data suggest that AIP1 is a novel transducer of TRAF2 in TNF-induced ASK1-JNK signaling.

MATERIALS AND METHODS
Plasmid Construction-Expression plasmids for truncated TNFR1 was generously provided by Dr. Martin Kronke (Institute for Medical Microbiology, Immunology and Hygiene, Medical Center of the University of Cologne, Cologne, Germany) (26). Expression plasmids for TRAF2, ASK1, and AIP1 were described previously (24,27). Deletion constructs of TRAF2 and AIP1 were generated by PCR and cloning into FLAG vectors. TRAF2-CA was generated by site-directed mutagenesis using QuickChange TM site-directed mutagenesis kit (Stratagene) according to the protocol of the manufacturer. Constructs were confirmation by DNA sequencing and by Western blot for protein expression.
Transfection and Reporter Assay-Transfection of HUVEC was performed by DEAE-Dextran method as described previously (27). BAEC and EAhy926 were transfected by LipofectAMINE 2000 (Invitrogen, Life Technologies). Luciferase activity followed by Renilla activity was measured twice in duplicate using a Berthold luminometer. All data were normalized as relative luciferase light units/Renilla unit.
Isolation of Caveolae-enriched Membranes-Purification of caveolaeenriched membrane fractions was performed as described (29) with minor modifications. In brief, EAhy.926 cells were treated with 10 ng/ml TNF for the indicated time point, washed twice with ice-cold Dulbecco's phosphate-buffered saline and scraped into 1 ml of MBS (25 mM MES, pH 6.5, 0.15 M NaCl, 5 mM EDTA, and 0,2% Triton X-100) with protease 15 inhibitors (10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM sodium orthovanadate, 10 mM NaF, 1 mM Pefabloc) and left 20 min on ice. The suspension was subjected to 10 -15 strokes in a Dounce homogenizer and centrifuged for 10 min at 2000 rpm at 4°C to remove high density debris. Clarified post-nuclear supernatants were combined with 90% (w/v) sucrose prepared in MES, transferred to the bottom of a Beckman 12.5 ml ultracentrifuge tubes and overlaid gently with 6 ml of 35% and 3 ml of 5% sucrose respectively. The resulting 5-40% discontinuous sucrose gradients were centrifuged 18 -20 h at 40,000 rpm in a SW41 Beckman rotor at 4°C to allow the separation of the low density rafts/caveolae. After centrifugation, a floating light band, corresponding to the Triton X-100 insoluble material, was detectable at the interface between 35 and 5% of each gradient. Fractions were harvested from the top to the bottom of the gradients and analyzed either by SDS-PAGE followed by immunoblotting for caveolin-1, TNFR1, TRAF2, or AIP1 or for binding of cholera toxin subunit B (CTxB) by dot blot analysis.
Indirect Immunofluorescence Confocal Microscopy-Fixation, permeabilization, and staining of cultured EC were performed as described previously (27). Alexa Fluor 488 (green) or 594 (red) conjugated-secondary antibodies (Molecular Probes, Eugene, OR) were used. Confocal immunofluorescence microscopy was performed using an Olympus confocal microscope and acquired images were transferred to Photoshop 6.0 to generate the final figures.

TNF Induces a Translocation of AIP1 from Plasma Membrane to Cytoplasm in EC-
We have recently shown that AIP1/ DAB2IP is highly expressed in EC but not in several common tumor cell lines including human prostate cancer and breast cancer cell lines (24,30). AIP1 contains a pleckstrin homology (PH) and a PKC-conserved region 2 (C2), two motifs implicated in plasma membrane targeting. To determine if AIP1 is localized on EC plasma membrane, we first examined AIP1 localization in resting BAEC by indirect immunofluorescence microscopy with an anti-AIP1 antibody. AIP1 showed a distinct plasma membrane staining (Fig. 1a, top panel). Control staining with a normal rabbit serum did not show any staining. To further approve AIP1 membrane localization, an EC membrane protein CD31 was used as a positive control. AIP1 and CD31 showed co-localization on EC (Fig. 1a, bottom panel). AIP1 also showed plasma membrane localization in cultured HUVEC (not shown). To determine the critical domain for AIP1 for membrane localization, BAEC were transiently transfected with expression constructs for FLAG-tagged AIP1 containing various domains (AIP1-F for the full-length, AIP1-N for the N-terminal domain containing the PH, C2, and GAP domains, AIP1-PHC2, AIP1-PH, AIP1-⌬PH for the mutant with a deletion of the PH domain, and AIP1-C for the C-terminal domain).
AIP1-F, AIP1-N, and PHC2 are predominantly localized on membrane vesicles (or microdomains). AIP1-PH showed plasma membrane as well as cytoplasm with no detectable vesicles, and AIP1-PH also caused elongation of cells. In contrast, AIP1-C and AIP1-⌬PH are detected in the cytoplasm, suggesting that the PH and C2 domain are critical for AIP1 FIG. 1. TNF induces a translocation of AIP1 from plasma membrane to cytoplasm in EC. a, AIP1 is localized on EC plasma membrane. Top panel, endogenous AIP1 in BAEC was stained with anti-AIP1 followed by an Alexa Fluor 488conjugated anti-rabbit secondary antibody. Cells were then counterstained with 4Ј,6-diamidino-2-phenylindole (DAPI) for nucleus. Bottom panel, AIP1 and EC membrane protein CD31 were costained with anti-AIP1 (rabbit) and anti-CD31 (mouse) followed by an Alexa Fluor 488 conjugated anti-rabbit and an Alexa Fluor 594 conjugated anti-mouse secondary antibodies. Location of AIP1 and CD31 was visualized under a confocal fluorescence microscope. b, critical domain of AIP1 for membrane localization. AIP1 constructs containing different domains (AIP1-F, N, PHC2, PH, and C) were transfected into BAEC, and localization of AIP1 proteins was determined by indirect immunofluorescence microscopy with anti-FLAG followed by an Alexa Fluor 488 (green)-conjugated anti-mouse secondary antibody. c, TNF induces AIP1 translocation in a time-dependent manner. Human EC (HUVEC) were treated with TNF (10 ng/ml) for various time points (0, 5, 15, 30, 45, and 60 min). EC were stained with AIP1 followed by an Alexa Fluor 594-conjugated anti-rabbit secondary antibody. Localization of AIP1 at 0, 15, and 60 min are shown.
Previously we have shown that TNF induces association of AIP1 with ASK1, which is predominantly localized in cytoplasm, raising the possibility that TNF might induce a translocation of AIP1 from plasma membrane to cytoplasm. To test this hypothesis, HUVEC were treated with TNF (10 ng/ml) for various time points (0, 5, 15, 30, and 60 min) and localization of AIP1 was determined as described. TNF treatment for 15 min induces a translocation of AIP1 from plasma membrane to cytoplasm. However, AIP1 returned to plasma membrane at 60 min of post-treatment with TNF (Fig. 1c). The kinetics of internalization is similar to that of TNF-induced AIP1-ASK1 association and ASK1-JNK activation (Ref. 24, also see Fig. 2). These data suggest that TNF-induced AIP1 internalization correlates with formation of AIP1-ASK1 complex and activation of ASK1.
TNF Induces Release of AIP1 from TNFR1 with Concomitant Formation of AIP1 Complex (TRADD-RIP1-TRAF2-AIP1-ASK1) in EC-The membrane localization and rapid movement of AIP1 in EC prompted us to examine association of AIP1 with TNFR1 signaling components including TNFR1, TRADD, RIP1, and TRAF2. We found that AIP1 and TNFR1 are abundant in human EC cell line EAhy926 and we chose EAhy926 to determine endogenous AIP1-TNFR1 complexes.
EC were treated with TNF (10 ng/ml for 0, 15, and 60 min) and association of AIP1 with TNFR1 signaling components by coimmunoprecipitation assays. TNF treatment did not alter protein expression of TNFR1, TRADD, RIP1, TRAF2, AIP1, or ASK1 (Fig. 2a). AIP1-TNFR1 complex is readily detected in resting EC. In response to TNF, AIP1 is dissociated from TNFR1 at 15 min but reassociates with TNFR1 at 60 min (Fig.  2b), consistent with the kinetics of AIP1 translocation. Formation of AIP1-SODD complex shows a similar kinetics as that of AIP1-TNFR1 complex (not shown). In contrast, TNF treatment induced a strong interaction of AIP1 with TRADD, RIP1, and TRAF2 at 15 min, which declined by 60 min of treatment (Fig.  2b). Association of TRAF2 and AIP1 with ASK1 as well as ASK1 activation show a similar kinetics (Fig. 2c). These data suggest that TNF induces formation of a complex comprising of TRADD-RIP1-TRAF2-AIP1-ASK1 (so named AIP1 complex) in EC. Furthermore, we determined kinetics of TNFR1-TRADD complex, which peaks at 2 min, declines at 5 min and disappears by 15 min (Fig. 2d), consistent with previous studies of HUVEC (31). Interestingly AIP1, which associates with the unactivated TNFR1 molecule, is absent in the TNFR1-TRADD-RIP1-TRAF2 complex (complex I), which assembles on the plasma membrane and has been shown to trigger activation of NF-B pathway (9).

FIG. 2. TNF induces release of AIP1
from TNFR1 with concomitant formation of AIP1 complex (TRADD-RIP1-TRAF2-AIP1-ASK1) in EC. a, human EC EAhy926 were treated with TNF (10 ng/ml) for indicated times. Protein expression was determined with respective antibodies. b, cell lysates were immunoprecipitated with indicated antibodies (TNFR1, TRADD, RIP1, TRAF2) followed by Western blot with anti-AIP1. As controls, TRAF2-ASK1 and AIP1-ASK1 complex were determined by immunoprecipitation with anti-TRAF2 or anti-AIP1 followed by Western blot with anti-ASK1. c, ASK1 activity was determined by an in vitro kinase assay using GST-MKK4 as a substrate. d, TNFR1-TRADD association precedes AIP1 complex. EAhy926 were treated with TNF for indicated time (0, 2, 5, 15, or 30 min) and association of TNFR1-TRADD was determined by immunoprecipitation with anti-TNFR1 followed by Western blot with anti-TRADD. e, AIP1 is not localized in caveolae. EAhy926 cells were either untreated or treated with TNF (10 ng/ml for 15 min) and fractionated by sucrose gradient as described. No protein was detected in fractions 1-2 and 10 l of cell lysates from fractions 3-10 were subjected to Western blot with anti-TNFR1, anti-TRAF2 and anti-AIP1. TRADD, RIP1 and ASK1 were also determined. Anti-caveolin-1 was used as a control for caveolae.
EAhy926 displays an extensive caveolar system, which has been shown to be critical for signaling by TNFR1 and TRAF2 (32,33). To determine if AIP1 complex is localized in caveolae, EAhy926 were treated with TNF (10 ng/ml for 15) and cell fractionations (fractions 1-10) by sucrose gradient centrifugation following Dounce homogenization in Triton X-100 buffer. In resting EC, the majority of TNFR1 is detected in fractions 10 containing heavy membranes and a small portion of TNFR1 can be detected in fractions 4 -5 where caveolin-1 is typically distributed. TNF decreases distribution of TNFR1 in fractions 4 -5 at 15 min (Fig. 2e). TRAF2 is also detected in caveolae fractions in both resting and TNF-treated EC. However, AIP1 is only detected in fractions 10, but not in other fractions, suggesting that AIP1 is not localized in caveolae (Fig. 2e). Similar to AIP1, TRADD, RIP1, and ASK1 were not detected in caveolae fractions (Fig. 2e). Taken together, these data suggest that that AIP1 associates with TNFR1 in resting EC whereas associates with TRADD-RIP1-TRAF2-ASK1 in TNF-treated EC in different membrane microdomains other than caveolae.
AIP1 in a Closed Inactive Form Associates with TNFR1 Whereas in an Open Active Form Binds to TRAF2/ASK1-We have previously shown that in resting EC AIP1 is retained in a closed conformation by intramolecular interactions and AIP1 associates with ASK1 only after the disruption of these intramolecular interactions (24). To define the critical domains for AIP1 intramolecular loop formation, we first generated deletion constructs of AIP1 at the N terminus or the C terminus (AIP1-⌬PH with a deletion of the N-terminal PH, amino acids 81-1056), AIP1-⌬LZ with deletion of the C-terminal LZ motif, amino acids 1-910), and AIP1-⌬PR with a deletion of the PR region, amino acids 1-796) (Fig. 3a). EC were transfected with various AIP1 mutants, and association of ASK1 with various AIP1 was determined by co-immunoprecipitation assays. AIP1-⌬PR, but not AIP1-F, AIP1-⌬PH or AIP1-⌬LZ, binds to ASK1 in the absence of TNF (Fig. 3b, ϪTNF). However, TNF treatment (10 ng/ml for 15 min) induced association of ASK1 with these AIP1 proteins (Fig. 3b, ϩTNF). These data suggest that the PR region, but not the LZ motif or the PH domain, is a critical domain in retaining AIP1 in a closed inactive form.
Association of AIP1 mutants with endogenous TRAF2 and TNFR1 in EC were then examined by co-immunoprecipitation assays. AIP1-⌬PR, but not AIP1-F, AIP1-⌬PH, or AIP1-⌬LZ, binds to TRAF2 in the absence of TNF (Fig. 3b, ϪTNF). Like ASK1, TRAF2 associated with the AIP1 proteins in response to TNF (Fig. 3b, ϩTNF). In contrast, TNFR1 associates with AIP1-F, AIP1-⌬PH, and AIP1-⌬LZ, but not to AIP1-⌬PR (Fig.  3b). These data suggest that TNFR1 binds to a closed inactive whereas TRAF2/ASK1 bind to an open active form of AIP1. To further test this idea, we examined association of AIP1-F (the closed form) and AIP1-⌬PR (the open form) with TNFR1 and TRAF2 in response to TNF. BAEC were transfected with AIP1-F or AIP1-⌬PR in the absence or presence of TNF (10 ng/ml for 15 min). Association of AIP1 proteins with endogenous TNFR1 and TRAF2 were determined. AIP1-F (but not AIP1-⌬PR) binds to TNFR1 in resting EC, and AIP1-F/ TNFR1complex was significantly reduced in the presence of TNF (Fig. 3c). In contrast, association of AIP1-F withTRAF2 is only weakly detected in resting EC but is strongly induced in response to TNF. AIP1-⌬PR constitutively binds to TRAF2 and TNF has no effects on their interaction (Fig. 3c). These data strongly suggest that TNF induces an alteration of AIP1 conformation leading to release of AIP1 from TNFR1 with concomitant binding of AIP1 to the TRAF2/ASK1 complex.
AIP1 Associates with a Non-death Domain Region of TNFR1-We next determined the critical domain in TNFR1 for AIP1 binding. The intracellular domain of TNFR1 comprises of several conserved regions-a juxtamembrane (amino acids 205-307), a NSD (amino acids 308 -319), which binds to an adaptor protein FAN responsible for neutral sphingomyelinase activation domain (34), and the death-domain (amino acids 346 -426), which binds to SODD in resting state while recruits TRADD in response to TNF (Fig. 4a). To map the AIP1-binding domain (AID) in TNFR1, AIP1-F, and various TNFR1 truncates were transfected into BAEC. Expression of TNFR1 proteins was determined by an indirect fluorescence microscopy with anti-TNFR1 (Fig. 4b). Association of AIP1 with TNFR1 was determined by co-immunoprecipitation assay with anti-TNFR1 followed by Western blot with anti-FLAG (for AIP1-F). AIP1 associates TNFR1-WT and D212-308/346, but not with TNFR1-D205, D244, or D308 -346), suggesting that the sequence between amino acids 309 -346 of TNFR1 is critical for AIP1 interaction (Fig. 4c). AIP1 poorly binds to TNFR1-D320 with the intact NSD, which has been shown to be a FANbinding motif, suggesting that AIP1 and FAN bind to different sequences of TNFR1. As a control, TRADD interacts with the death domain-containing TNFR1 (TNFR1-WT and D308 -346) as determined by Western blot with anti-TRADD. The results were summarized in Fig. 4a. Taken together, these data suggest that AIP1 associates with TNFR1 at a site distinct from the TRADD-binding domain (DD) and the FAN-binding domain (NSD).
AIP1 via a PERIOD-like Domain Binds to the Intact RING Finger of TRAF2-TRAF2 associates with AIP1 in response to TNF (Fig. 2) and constitutively binds to an open form of AIP1 (AIP1-⌬PR) (Fig. 3). To further map the critical domain in AIP1 for TRAF2 binding, we determined association of TRAF2 with various truncated AIP1 proteins. Since we have previously shown that the C2 domain of AIP1 is critical for ASK1 association (24), and we first examined if TRAF2 binds to the Nterminal half of AIP1. Results show that TRAF2 did not interact with the N-terminal domains (AIP1-N, PHC2, and PH) (Fig.  5a), indicating that TRAF2 and ASK1 bind to different sites on AIP1. To determine TRAF2 binds to the C-terminal half of AIP1, we generated truncated AIP1-C (AIP1-C, C-⌬LZ with a deletion of the LZ motif, C-⌬PR with a deletion of the PR region). Results show that TRAF2 binds to AIP1-C-⌬PR, but not AIP1-C or C-⌬LZ (Fig. 5a). These data suggest that TRAF2 binds to an upstream sequence of the PR region in the Cterminal half of AIP1, and that the PR region is also critical to maintain AIP1-C in a closed conformation. Domain searching indicated that AIP1-C-⌬PR does not contain a TRAF2-binding consensus site but has a PERIOD-like domain (PER, amino acids 591-719). We further generated AIP1-C-PER (amino acids 522-719) and C-tPER (amino acids 522-620) with a truncation of PER domain. Association of TRAF2 with C-⌬PR, C-PER, and C-tPER was then determined by a co-immunoprecipitation assay. TRAF2 binds to AIP1-C-⌬PR and C-PER, but not C-tPER (Fig. 5b), suggesting that the PER domain in the C-terminal-half is involved in TRAF2 binding.
To define the critical domain of TRAF2 for AIP1 binding, we generated a series of mutant forms of FLAG-tagged TRAF2. DN-TRAF2 (amino acids 80 -531) contains a deletion of the whole RING finger whereas TRAF2-CA contains a single mutation at C31 within the RING finger leading to disruption of the RING structure (15). TRAF2-N (amino acids 1-249) contains the intact RING and the five zinc fingers of the N-terminal half of TRAF2 whereas TRFA2-C (amino acids 265-531) consists of the C-terminal TRAF domains of TRAF2 (Fig. 5c). AIP1-⌬PR was co-transfected with various TRAF2, and association of AIP1-⌬PR with TRAF2 molecules was determined. As previously described, TRAF2-N was detected in an insoluble fraction (27) and we cannot determine association of AIP1 with TRAF2-N. Results show that only TRAF2-WT binds to AIP1-⌬PR. However, DN-TRAF2, TRAF2-CA, or TRAF2-C did not interact with AIP1-⌬PR (Fig. 5d). These data indicate that the intact RING finger (the effector domain) of TRAF2 is required for AIP1 association.
AIP1 Is a Transducer of TRAF2 and Specifically Induces ASK1-JNK Activation While It Inhibits NF-B Activation-TRAF2 has been shown to be a critical adaptor in TNF-induced activation of both JNK and NF-B cascades. We first determined effects of AIP1 in TRAF2-induced JNK and NF-B activation in reporter gene assays. BAEC were transfected with a JNK-or NF-B-dependent reporter gene with AIP1 in absence or the presence of TNF as indicated. Reporter activity was measured for a luciferase assay. As shown previously, expression of TRAF2 in EC-induced activation of both JNK and NF-B reporter genes (Fig. 6a). Co-expression of AIP1 strongly induces JNK reporter gene activation. In contrast, AIP1 dramatically inhibits TRAF2-induced activation of B-reporter gene. A similar effect of AIP1 on TNF-induced activation of JNK and NF-B reporter genes was observed (data not shown). AIP1 on TNF-induced activation of ASK1 and IB kinase (IKK) were also determined by an in vitro kinase assay using GST-MKK4 and GST-IB␣ as a substrate, respectively. As shown previously, AIP1 enhanced TNF-induced ASK1 activation. In contrast, AIP1 strongly inhibited TNF-induced IKK activation (Fig. 6b). These data suggest that AIP1 interacts with the FIG. 3. AIP1 in a closed inactive form associates with TNFR1 whereas in an open active form binds to TRAF2/ASK1. a, schematic diagram of AIP1 domains and expression constructs (C2, protein kinase C-conserved domain 2; GAP, GTPase-activating protein; PER, period-like domain; PR, proline-rich; LZ, leucine-zipper; aa, amino acid). b, proline-rich region is critical for closed conformation of AIP1. BAEC were transfected with FLAG-tagged truncated AIP1 (AIP1-F, ⌬PH, ⌬LZ, ⌬PR), and cells were untreated or treated with TNF (10 ng/ml for 15 min). Association of endogenous ASK1, TRAF2, or TNFR1 with AIP1 domains was examined by immunoprecipitation with a respective antibody followed by Western blot with anti-FLAG. c, TNF responses of AIP1-TNFR1 and TRAF2-AIP1 complexes. BAEC were transfected with AIP1-F or ⌬PR and treated with TNF (10 ng/ml for 15 min). Association of AIP1 with TNFR1 and TRAF2 was determined as in b.
effector domain of TRAF2 to reciprocally regulate TNF-induced JNK and NF-B cascades.
We then determined the critical domains of AIP1 in regulating TNF-induced JNK and NF-B activation. BAEC were transfected with a JNK-or NF-B-dependent reporter gene with various AIP1 in absence or the presence of TNF as indicated. Reporter activity was measured for a luciferase assay. As shown previously, expression of TRAF2 in EC induced activation of both JNK and NF-B reporter genes (Fig. 6c). Expression of AIP1-F, N, or C-⌬PR (but not C-tPER) significantly augmented TNF-induced JNK activation. In contrast, AIP1-F, N or C-⌬PR (but not C-tPER) dramatically inhibited TRAF2induced NF-B reporter gene activity (Fig. 6d). A similar effect of AIP1 on TRAF2-induced JNK and NF-B was observed (data not shown). These data suggest that effects of AIP1-F, N, C-⌬PR, C-tPER) on JNK and B reporter correlate with their binding abilities for ASK1 and TRAF2.
We have previously shown that a short hairpin RNA (shRNA) down-regulated endogenous AIP1 level leading to enhanced TNF-induced ASK1-JNK activation and apoptosis (24). To determine a physiological role of AIP1 in TNF-induced NF-B signaling, we further examined TNF-induced ASK1 and IKK activation in AIP1-knockdown cells by in vitro kinase assays as described. Endogenous AIP1 expression was significantly (80%) reduced in BAEC expressing a short hairpin RNA of AIP1 (ShAIP1) while expression of TRAF2 or ASK1 was not altered (Fig. 6e). As we have shown previously (24), knockdown of AIP1 inhibited TNF-induced ASK1 activation, However, AIP1 knockdown cells showed enhanced TNF-induced IKK kinase activity (Fig. 6f). These data further support that physiological AIP1 differentially regulates TNF-induced ASK1-JNK and IKK-NF-B pathways. b and c, AIP1-interacting domain (AID) is located between NSD and DD. BAEC were transfected with FLAG-tagged AIP1 and various TNFR1 deletion constructs. Expression of TNFR1 proteins was determined by indirect immunofluorescence microscopy with anti-TNFR1 (goat, recognizing the extracellular domain of TNFR1) followed by an Alexa Fluor 488-conjugated anti-goat secondary antibody (b). Association of AIP1 domains with TNFR1 was determined by immunoprecipitation with anti-TNFR1 followed by Western blot with anti-FLAG (c). TNFR1-associated TRADD was also determined by anti-TRADD. Associations of these TNFR1 proteins with TRADD and AIP1 are summarized on the right panel in a.

DISCUSSION
Based on the data presented in this study, we propose the following model for the role of AIP1 in TNF signaling (Fig. 7). AIP1 via its PH/C2 domain is localized on plasma membrane microdomain where it exists in an inactive form complexed with TNFR1. In response to TNF, AIP is unfolded and dissociated from TNFR1 with concomitant of translocation of AIP1 to cytoplasm where it associates with TRAF2 and ASK1. More significantly, AIP1 associates with the effector domain (the RING finger) of TRAF2 and mediates TNF/TRAF2-induced ASK1-JNK activation while inhibits IKK-NF-B signaling. These data suggest that AIP1 is a novel transducer of TRAF2 in TNF signaling.
Intermediate Complex-Tschopp and co-workers (9) have recently dissected TNFR1 signaling complexes in more details and has proposed a sequential signaling complex model: the initial plasma membrane bound complex (complex I comprising of TNFR1, TRADD, RIP1 and TRAF2) for NF-B activation and a cytoplasmic complex (complex II) in which the internalized TRADD, RIP, TRAF2 recruits FADD and pro-caspase-8 for apoptotic signaling. However, it has not been determined if either complex I or complex II is active in the initiation of JNK signaling. Since TRAF2 is critical for JNK activation, it seemed reasonable to assume that complex I, which contains TRAF2, would recruit one or more MAP3Ks to induce JNK activation (8,9). Here we provide several lines of evidence to support the existence of an intermediate complex, which lacks TNFR1, as being responsible for TRAF2-AIP1-mediated ASK1-JNK activation. First, AIP1 forms a preexisting complex with TNFR1 in EC. However, in response to TNF AIP1 is dissociated from TNFR1 and in turn forms a complex with TRADD, RIP1, TRAF2, and ASK1 (so named AIP1 complex). Thus, AIP1 com-plex structurally different from complex I. Second, AIP1 complex is formed at 15 min in response to TNF and is dissociated by 60 min. However, complex II formation detected in certain cell types (HT1080) at 2-8 h post-treatment with TNF. Moreover, we could not detect association of AIP1 with FADD and pro-caspase-8 in response to TNF in EC (not shown). These data suggest that AIP1 complex is also structurally different from complex II. Most significantly, AIP1 overexpression enhances, whereas knockdown of AIP1 inhibits, TNF/TRAF2induced ASK1-JNK activation. However, AIP1 has opposite effects on TNF/TRAF2-induced IKK-NF-B pathway. These data suggest AIP1 complex is functionally different from complex I. In cell types including EC, ASK1-JNK activation induces apoptotic signaling which is dependent on intrinsic but not extrinsic pathways. We have recently shown that AIP1 enhances ASK1-mediated JNK activation and EC apoptosis. AIP1-suppressed IKK-NF-B activation further support that AIP1 is proapoptotic. Thus, AIP complex is functionally different from complex II. Whether or not complex II is present in EC are under investigation.
AIP1 Specifies TRAF2 toward to ASK1-JNK Pathway-TRAF2 has been shown to be the bifurcation point in TNFinduced activation of NF-B and activation of JNK (8,35,36). While the C-terminal TRAF domain is responsible for association with TNFR1-TRADD complex as well as the MAP3Ks, the N-terminal RING finger of TRAF2 is an effector domain in activation of both JNK and NF-B. Deletion of the RING domain (TRAF2-(87-501)) completely loses the ability to activate JNK and NF-B, and this mutant functions as a dominant negative. Specific downstream signaling events from TRAF2 are mediated by RIP/IKK and MAP3Ks. For example, IKK is specific for NF-B signaling, while ASK1 appear to direct sig-FIG. 5. AIP1 via a PERIOD-like domain binds to the intact RING finger of TRAF2. a and b, critical domains of AIP1 for TRAF2 binding. Truncated AIP-1N (N, PHC2, and PH) and AIP1-C (C, C-⌬PR, C-⌬LZ) were transfected into BAEC. Association of endogenous TRAF2 with AIP1 proteins was determined by immunoprecipitation with anti-TRAF2 followed by Western blot with anti-FLAG. b, period-like domain (PER, amino acids 591-719) in AIP1 is critical for TRAF2 binding. AIP1-⌬PR, C-PER (amino acids 522-719), and C-tPER (amino acids 522-620 with a truncated PER domain). TRAF2 was co-transfected with various AIP1-C constructs (C-⌬PR, C-PER and C-tPER). Association of TRAF2 with AIP1-C was determined by co-immunoprecipitation assay as described. c, diagrammatic representation of the domain structure of TRAF2 and expression constructs. The numbers refer to amino acid number indicating the boundary of the RING finger, five zinc fingers and the TRAF domains. d, intact RING finger of TRAF2 is critical for AIP1 association. BAEC were transfected with AIP1-⌬PR and various TRAF2 constructs. Association of AIP1-⌬PR with TRAF2 was determined by immunoprecipitation with anti-TRAF2 followed by Western blot with anti-FLAG.
naling exclusively to JNK. In contrast, TAK1 and MEKK1 can activate both NF-B and JNK (37,38). Recent studies suggest that TRAF2 engagement is not obligatory to trigger NF-B and JNK cascades simultaneously. For example, sphingosine kinase binds to TRAF domain of TRAF2 and specifically activates NF-B while inhibits JNK activation (39). In contrast, TNFinduced TRAF2 ubiquitination coincides with its translocation to an insoluble cellular fraction, which is critical for TNF-induced activation of JNK but not of NF-B (40). These data suggest that TRAF2 modulation/translocation or interactions with other proteins may provide a new layer of regulation to determine specific downstream signaling (activation of JNK and NF-B). Our study shows that AIP1 associates with TRAF2 in response to TNF and the intact RING finger of TRAF2 is required for TRAF2-AIP1 association. Moreover, TRAF2, AIP1, and ASK1 form a complex in response to TNF FIG. 6. AIP1 is a transducer of TRAF2 and specifically induces ASK1-JNK activation while inhibits IKK-NF-B activation. a, AIP1 specifically enhances TRAF2-induced JNK (but inhibits NF-B) activation. AIP1 together with a JNK or NF-B-dependent reporter gene was transfected into BAEC in the presence or absence of TRAF2 as indicated. A constitutive expression vector for Renilla was used as an internal control for transfection efficiency. Reporter gene activity was determined by luciferase assay as described. Data are presented as fold increase by expression of TRAF2 or AIP1 compared with the vector (as 1). Similar results were obtained from additional three independent experiments. b, AIP1 specifically enhances ASK1 whereas inhibits IKK activation by TNF. BAEC were transfected with a control vector or AIP1 and treated with TNF as indicated (10 ng/ml for 15 min). ASK1 and IKK activities were determined by an in vitro kinase assay using GST-MKK4 and GST-IKB␣, respectively. Relative ASK1 and IKK activities are presented by taking TNF-treated VC as 1.0. Similar results were obtained from two additional independent experiments. c and d, critical domains of AIP1 in TNF-induced activation of JNK and NF-B. BAEC were transfected with a JNKor NF-B-dependent reporter gene with various AIP1 (F, N, ⌬PR, or tPER) as indicated and a constitutive expression vector for Renilla was used as an internal control for transfection efficiency. 24 h post-transfection, cells were treated with TNF (10 ng/ml) for 12 h. Reporter gene activity was determined by luciferase assay as described. Data are presented as fold increase by AIP1 expression by taking untreated vector as 1 (Ϫ). Similar results were obtained from additional three independent experiments. e, knockdown of AIP1 in BAEC. BAEC were transiently transfected with pShag or pSh-AIP1 encoding a short-hairpin RNA of AIP1 (Sh-AIP1) as described previously (24). 48 h post-transfection, total cell lysates were used to determine AIP1 expression by Western blot with anti-AIP1. As expected, expression of TRAF2 and ASK1 was not altered by Sh-AIP1. f, knockdown of AIP1 suppresses ASK1 activity whereas enhances IKK activation by TNF. BAEC expressing pShag or pSh-AIP1 were untreated or treated with TNF (10 ng/ml for 15 min). ASK1 and IKK activities were determined as described in b. Relative ASK1 and IKK activities are presented by taking untreated pShag as 1.0. Similar results were obtained from additional two experiments. and overexpression of TRAF2 enhances association of AIP1 with ASK1, at least in part by unfolding both AIP1 and ASK1. Most importantly, AIP1 specifically mediates TNF/TRAF2-induced ASK1-JNK activation while inhibits IKK-NF-B signaling as demonstrated by both overexpression and knockdown of AIP1. Although many TRAF-binding proteins have been identified to modulate TRAF2 function, AIP1 represents a first molecule, which may specify TRAF2 toward to ASK1-JNK pathway. It needs to further determine the effects of AIP1 on TRAF2 ubiquitination/translocation and TRADD-TRAF2-RIP1 complex formation. FIG. 7. A proposed model for AIP1 as a transducer of TRAF2 in TNF-induced ASK1-JNK activation. AIP1 via its PH and C2 domains anchors to EC plasma membrane where it associates with a non-death domain region of TNFR1. In response to TNF, AIP1 is dissociated from TNFR1 with concomitant cytoplasmic translocation of AIP1 and formation of AIP1 complex comprising of TRADD-RIP-TRAF2-AIP1, which specifically activates ASK1-JNK signaling. As demonstrated by Tschopp and co-workers (9), TNFR1 recruits TRADD, RIP1, and TRAF2 to form a membrane-bound complex I, which specifically activates NF-B signaling. Since TNFR1-TRADD association precedes formation of AIP1 complex, it is likely that AIP1 complex is derived from complex I upon ubiquitination (ࡗ) and internalization of TRADD, RIP1, TRAF2, and AIP1. As we described previously (9), unfolded AIP1 binds to and activates ASK1-JNK at least in part by facilitating dephosphorylation of ASK1 at Ser 967 and dissociation of ASK1 from its inhibitor 14-3-3.