Involvement of TRAF4 in Oxidative Activation of c-Jun N-terminal Kinase
2002; Elsevier BV; Volume: 277; Issue: 31 Linguagem: Inglês
10.1074/jbc.m202665200
ISSN1083-351X
AutoresYou Xu, Ru Feng Wu, Ying Gu, Y S Yang, Meng-Chun Yang, Fiemu E. Nwariaku, Lance S. Terada,
Tópico(s)Immune Response and Inflammation
ResumoWe previously found that the angiogenic factors TNFα and HIV-1 Tat activate an NAD(P)H oxidase in endothelial cells, which operates upstream of c-Jun N-terminal kinase (JNK), a MAPK involved in the determination of cell fate. To further understand oxidant-related signaling pathways, we screened lung and endothelial cell libraries for interaction partners of p47 phox and recovered the orphan adapter TNF receptor-associated factor 4 (TRAF4). Domain analysis suggested a tail-to-tail interaction between the C terminus of p47 phox and the conserved TRAF domain of TRAF4. In addition, TRAF4, like p47 phox , was recovered largely in the cytoskeleton/membrane fraction. Coexpression of p47 phox and TRAF4 increased oxidant production and JNK activation, whereas each alone had minimal effect. In addition, a fusion between p47 phox and the TRAF4 C terminus constitutively activated JNK, and this activation was decreased by the antioxidant N-acetyl cysteine. In contrast, overexpression of the p47 phox binding domain of TRAF4 blocked endothelial cell JNK activation by TNFα and HIV-1 Tat, suggesting an uncoupling of p47 phox from upstream signaling events. A secondary screen of endothelial cell proteins for TRAF4-interacting partners yielded a number of proteins known to control cell fate. We conclude that endothelial cell agonists such as TNFα and HIV-1 Tat initiate signals that enter basic signaling cassettes at the level of TRAF4 and an NAD(P)H oxidase. We speculate that endothelial cells may target endogenous oxidant production to specific sites critical to cytokine signaling as a mechanism for increasing signal specificity and decreasing toxicity of these reactive species. We previously found that the angiogenic factors TNFα and HIV-1 Tat activate an NAD(P)H oxidase in endothelial cells, which operates upstream of c-Jun N-terminal kinase (JNK), a MAPK involved in the determination of cell fate. To further understand oxidant-related signaling pathways, we screened lung and endothelial cell libraries for interaction partners of p47 phox and recovered the orphan adapter TNF receptor-associated factor 4 (TRAF4). Domain analysis suggested a tail-to-tail interaction between the C terminus of p47 phox and the conserved TRAF domain of TRAF4. In addition, TRAF4, like p47 phox , was recovered largely in the cytoskeleton/membrane fraction. Coexpression of p47 phox and TRAF4 increased oxidant production and JNK activation, whereas each alone had minimal effect. In addition, a fusion between p47 phox and the TRAF4 C terminus constitutively activated JNK, and this activation was decreased by the antioxidant N-acetyl cysteine. In contrast, overexpression of the p47 phox binding domain of TRAF4 blocked endothelial cell JNK activation by TNFα and HIV-1 Tat, suggesting an uncoupling of p47 phox from upstream signaling events. A secondary screen of endothelial cell proteins for TRAF4-interacting partners yielded a number of proteins known to control cell fate. We conclude that endothelial cell agonists such as TNFα and HIV-1 Tat initiate signals that enter basic signaling cassettes at the level of TRAF4 and an NAD(P)H oxidase. We speculate that endothelial cells may target endogenous oxidant production to specific sites critical to cytokine signaling as a mechanism for increasing signal specificity and decreasing toxicity of these reactive species. tumor necrosis factor human immunodeficiency virus Jun N-terminal kinase untranslated region hemagglutinin green fluorescent protein glutathione S-transferase mitogen-activated protein kinase TNF receptor-associated factor 4 human umbilical vein endothelial cell The vascular endothelium is generally well supplied with oxygen and produces significant quantities of oxidants when stimulatedin vivo or in vitro (1Al-Mehdi A.B. Zhao G. Dodia C. Tozawa K. Costa K. Muzykantov V. Ross C. Blecha F. Dinauer M. Fisher A.B. Circ. Res. 1998; 83: 730-737Crossref PubMed Scopus (248) Google Scholar, 2Gu Y., Wu, R.F., Xu, Y.C. Flores S.C. Terada L.S. Virology. 2001; 286: 62-71Crossref PubMed Scopus (42) Google Scholar). As in other cell types, such tightly regulated oxidant bursts appear to transduce a variety of signals. Mechanical forces, growth factors, and cytokines stimulate oxidant production by endothelial cells, leading to migration, proliferation, apoptosis, or adhesion protein expression (3Bhunia A.K. Arai T. Bulkley G. Chatterjee S. J. Biol. Chem. 1998; 273: 34349-34357Abstract Full Text Full Text PDF PubMed Scopus (98) Google Scholar, 4Abid M.R. Kachra Z. Spokes K.C. Aird W.C. FEBS Lett. 2000; 486: 252-256Crossref PubMed Scopus (182) Google Scholar, 5Deshpande S.S. Angkeow P. Huang J. Ozaki M. Irani K. FASEB J. 2000; 14: 1705-1714Crossref PubMed Scopus (205) Google Scholar). However, the relatively broad biochemical reactivity of these oxidants poses a potential problem for signal specificity. As an example, a number of studies now support the participation of oxidants in both proliferative (6Ushio-Fukai M. Zafari A.M. Fukui T. Ishizaka N. Griendling K.K. J. Biol. Chem. 1996; 271: 23317-23321Abstract Full Text Full Text PDF PubMed Scopus (695) Google Scholar, 7Sundaresan M., Yu, Z.X. Ferrans V.J. Irani K. Finkel T. Science. 1995; 270: 296-299Crossref PubMed Scopus (2322) Google Scholar, 8Irani K. Xia Y. Zweier J.L. Sollott S.J. Der C.J. Fearon E.R. Sundaresan M. Finkel T. Goldschmidt-Clermont P.J. Science. 1997; 275: 1649-1652Crossref PubMed Scopus (1441) Google Scholar) and apoptotic (9Saitoh M. Nishitoh H. Fujii M. Takeda K. Tobiume K. Sawada Y. Kawabata M. Miyazono K. Ichijo H. EMBO J. 1998; 17: 2596-2606Crossref PubMed Scopus (2092) Google Scholar, 10Manna S.K. Zhang H.J. Yan T. Oberley L.W. Aggarwal B.B. J. Biol. Chem. 1998; 273: 13245-13254Abstract Full Text Full Text PDF PubMed Scopus (523) Google Scholar) pathways, depending on stimulus and context. The basis for the divergent responses to oxidants is not clear. Endothelial cells possess an NAD(P)H oxidase (11Mohazzab K.M. Kaminski P.M. Wolin M.S. Am. J. Physiol. 1994; 266: H2568-H2572PubMed Google Scholar) thought to participate in a number of these signal pathways. Inhibitors of this oxidase suppress growth factor, TNFα,1 HIV-1 Tat, and shear cessation-induced signaling (2Gu Y., Wu, R.F., Xu, Y.C. Flores S.C. Terada L.S. Virology. 2001; 286: 62-71Crossref PubMed Scopus (42) Google Scholar, 4Abid M.R. Kachra Z. Spokes K.C. Aird W.C. FEBS Lett. 2000; 486: 252-256Crossref PubMed Scopus (182) Google Scholar, 12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar, 13Wei Z. Costa K., Al- Mehdi A.B. Dodia C. Muzykantov V. Fisher A.B. Circ. Res. 1999; 85: 682-689Crossref PubMed Scopus (130) Google Scholar), and dominant negative Rac1 disrupts TNFα signaling (5Deshpande S.S. Angkeow P. Huang J. Ozaki M. Irani K. FASEB J. 2000; 14: 1705-1714Crossref PubMed Scopus (205) Google Scholar) in endothelial cells. Recently, both cytochrome subunits of the oxidase, p22 phox , and gp91 phox , were cloned from rat and human endothelial cells (14Bayraktutan U. Blayney L. Shah A.M. Arterio. Thromb. Vasc. Biol. 2000; 20: 1903-1911Crossref PubMed Scopus (209) Google Scholar, 15Gorlach A. Brandes R.P. Nguyen K. Amidi M. Dehghani F. Busse R. Circ. Res. 2000; 87: 26-32Crossref PubMed Scopus (543) Google Scholar). We subsequently cloned the oxidase adapter subunit p47 phox from HUVEC, demonstrating its participation in TNFα signaling (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). Unexpectedly, endogenous p47 phox was found to be constitutively associated with the cytoskeleton of unstimulated ECV-304 cells, contrasting the free cytosolic location of p47 phox in unstimulated neutrophils. Because most signaling proteins are associated with the cytoskeleton at some point in their activation cycle, the strong association of p47 phox with the endothelial cytoskeleton suggested specific localization of the oxidase with cytoskeletally anchored signaling complexes. Indeed, cytoskeletal disruption derailed both oxidase activation and downstream signaling (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). Spatial targeting of the oxidase may therefore potentially confer signal specificity to these evanescent radicals. To identify potential vicinal signaling elements associated with the endothelial NAD(P)H oxidase, we screened lung and HUVEC libraries for p47 phox -interacting proteins and recovered the orphan adapter TRAF4. This interaction appears to participate in downstream activation of JNK by the oxidase-activating endothelial agonists TNFα and HIV-1 Tat. All PCR amplifications for subcloning or mutagenesis were performed with Pfu Turbo (Stratagene). The bait vector pGBKT7-p47 was created by a single base mutation of p47 phox (T to C at −2), creating a newNcoI site. The NcoI-EcoRI fragment containing the coding region and 3′-UTR of p47 phox was then subcloned into pGBKT7 (CLONTECH) in-frame with the GAL4-BD. Full-length TRAF4 was PCR-amplified from a HUVEC library (Stratagene) between theEcoRI and SalI sites. It was directly ligated into the expression vector pCI (Promega) to create pCI-T4 and into the yeast shuttle vector pGBKT7 to create pGBKT7-T4. The C-terminal TRAF domain of TRAF4 was excised from the library prey plasmid pACT2-T4 using EcoRI and PshAI and ligated into pCIneo-FLAG (16Yang Y.S. Yang M.C. Wang B. Weissler J.C. Am. J. Respir. Cell Mol. Biol. 2001; 24: 30-37Crossref PubMed Scopus (27) Google Scholar) to create pCINF-T4(CT). pGBKT7-p47-(1–205) was constructed by removing the C-terminalBamHI-BamHI fragment from pGBKT7-p47, and pGBKT7-p47-(205–390) was obtained by isolation of the N-terminalSalI-BamHI fragment of p47 phox and ligation into pGBKT7. pGBKT7-p47-(1–346) was derived by excision of aSmaI-SmaI segment from pGBKT7-p47. pGBKT7-p47-(1–298) was derived by complete restriction of pGBKT7-p47 with EcoRI, partial restriction with NarI, T4 polymerase end fill-in, gel purification, and blunt-end ligation to reseal the plasmid. pGBKT7-p47-(347–390) was obtained by PCR deletion of p47-(1–346) from pGBKT7-p47 and frame correction by NcoI restriction, end fill-in, and blunt-end resealing. GAL4-BD fusions for p47-(153–286), (299–345), and (299–390) were derived by PCR amplification of segments between EcoRI and SalI sites, with insertion of appropriate stop codons, followed by ligation into pGBKT7. GAL4-AD fusions with TRAF4-(266–307) and TRAF4-(308–470) were produced by PCR amplification of segments between EcoRI and XhoI sites and ligation into pGADT7 (CLONTECH). The coding region of p47 phox was PCR-amplified betweenEcoRI and SalI sites and ligated into pCIneo-FLAG to yield pCINF-p47, and between two EcoRI sites with ligation into pGEX-2TK to yield pGEX-p47. HA-JNK2 was derived by reversal of HA-JNK2(APF) mutant (gift from Dr. Lynn Heasley) back to wild type with PCR mutagenesis, and HA-JNK1 was a gift from Dr. Stephen Dreskin (17Dreskin S.C. Thomas G.W. Dale S.N. Heasley L.E. J. Immunol. 2001; 166: 5646-5653Crossref PubMed Scopus (87) Google Scholar). pGAD424-TRAF1 and pGAD424-TRAF2 were gifts from Dr. Preet Chaudry. The frame of the former construct was corrected prior to use. The fusion construct pCINF-p47-T4 was produced by PCR amplification of p47 phox -(1–389) between EcoRI sites and ligation into pCINF-T4(CT) between the FLAG tag and TRAF4 C terminus. p47-GFP was constructed as previously described (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). All constructs were confirmed by direct sequencing. HUVEC were transfected with p47-GFP and plated onto fibronectin-coated slides. After 24 h, cells were fixed and permeablized (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar), counterstained with rhodamine-phalloidin (5 units/ml), and examined using a Zeiss Axiovert S100TV LSM 410 laser-scanning system. A commercial human lung library cloned into pACT2 was obtained (CLONTECH). An endothelial cell library was constructed using poly(A+) RNA from passage 4–5 HUVEC (Clonetics). Oligo(dT)-primed cDNA was cloned directionally into the EcoRI and XhoI sites of lambda phage HybriZAP 2.1XR (Stratagene). The primary library (4.8 × 106 plaque-forming units) was amplified once as lambda phage, and the library was dropped out in the yeast shuttle vector pAD-GAL4-2.1 by mass excision using a kit (Stratagene). The average insert size was 1.7 kb. Saccharomyces cerevisiae AH109 (CLONTECH) were stably transformed with the bait vector pGBKT7-p47 under tryptophan-deficiency selection, and full-length p47 phox was found to lack autonomous transactivation activity. Stable transformants were then secondarily transformed with the lung or endothelial cell libraries using lithium acetate. AH109 yeast contains two auxotrophic reporter genes (ADE2 and HIS3) andlacZ under the control of three distinct promoters. Therefore, colonies that survived selection using medium deficient in tryptophan (bait selection), leucine (library selection), adenine, and histidine (interaction selection) were tested for lacZexpression using a filter lift assay. Positive colonies were restreaked, and single clones were retested for auxotrophy andlacZ expression. Library plasmids from triple-positive clones were extracted and passaged through Escherichia coliwith selection for GAL4-AD plasmids and stably transformed back into AH109. Autonomous transactivation negative clones were then mated with Y187 yeast containing pGBKT7-p47 and diploids tested forlacZ expression. A similar approach was used to rescreen the HUVEC library using full-length TRAF4 as bait after stable transformation of AH109 with pGBKT7-T4. Phoenix-293 (Fx) cells electroporated with indicated vectors were incubated in lysis buffer (20 mmTris, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1% Triton X-100, 2.5 mm sodium pyrophosphate, 1 mm β-glycerophosphate, 1 mmNa3VO4, 1 μg/ml leupeptin, and 1 mm phenylmethylsulfonyl fluoride) for 30 min at 4 °C, sonicated for 5 s, and centrifuged at 6000 × gfor 20 min at 4 °C. Immunoprecipitation was then performed using antibodies against FLAG (Sigma), pelleting protein G-agarose (AmershamBiosciences, Inc.) conjugates at 1500 × g. Immunoblots were performed with antisera recognizing the C terminus of TRAF4 (Santa Cruz Biotechnology, C-20) or FLAG. Cytoskeletal fractions were collected as previously described (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar), and cytosolic fractions were recovered by acetone precipitation of supernatant. AH109 yeast stably transformed with GAL4-AD-TRAF4 constructs were tested first for autonomous transactivation; negative colonies were mated with Y187 yeast stably transformed with various GAL4-BD-p47 deletion constructs. Diploids were replated on medium selecting for both plasmids and tested forlacZ expression with a filter lift assay. Negative controls were Tyr-187 transformed with empty pGBKT7 and pGBKT7-lamin C; the positive control was AH109 transformed with holo-GAL4 (pCL1, CLONTECH). Positive interactions were identified by development of a blue color within 1 h, negative interactions remained white for >24 h. Direct interactions were confirmed in vitro (18Zapata J.M. Matsuzawa S. Godzik A. Leo E. Wasserman S.A. Reed J.C. J. Biol. Chem. 2000; 275: 12102-12107Abstract Full Text Full Text PDF PubMed Scopus (46) Google Scholar). BL21-RP E. coli (Stratagene) were transformed with either pGEX-2TK or pGEX-p47, induced for 3 h at 37 °C, and the GST proteins were captured on glutathione-Sepharose (Amersham Biosciences, Inc.). Approximately 2 μg of GST or GST-p47 was used per 200 μl of binding reaction. Full-length TRAF4 was transcribed and translated in vitro from pCI-T4 (TnT Quick Coupled, Promega) using [35S]methionine, and 5 μl (3 μCi) of translation mix was added to each binding reaction. In some reactions, p47-(299–390) was in vitro transcribed and translated from pGBKT7-p47-(299–390) without isotope and added directly to the binding reaction simultaneously with labeled TRAF4. A silent G360C mutation was introduced in p47 phox to eliminate a potential PI-CeuI site. The entire p47 phox cDNA was excised with XbaI and KpnI and ligated into pShuttle, and the expression cassette was then subcloned into the backbone pAdeno-X (CLONTECH). The linearized adenoviral DNA was then transfected into HEK-293 cells and replication-incompetent viruses harvested and titered. Ad-lacZ was similarly constructed according to the manufacturer's suggestions (CLONTECH). Robust expression of p47 phox was demonstrated by immunoblot at multiplicity of infections of 50–200 (not shown), and lacZexpression was demonstrated in >95% of cells. JNK activity of Fx cells was assessed using a traditional immunoprecipitation kinase technique using anti-JNK1 (Santa Cruz Biotechnology, C-17) and GST-Jun (Santa Cruz Biotechnology) (2Gu Y., Wu, R.F., Xu, Y.C. Flores S.C. Terada L.S. Virology. 2001; 286: 62-71Crossref PubMed Scopus (42) Google Scholar). Equivalent capture of JNK was assessed with immunoblot using a pan-specific anti-JNK (JNK-FL, Santa Cruz Biotechnology). To increase HUVEC transfection efficiency, passage 2–3 HUVEC (Clonetics) were synchronized at the G1-S transition with 3.5 mmthymidine overnight (19Schwachtgen J.L. Ferreira V. Meyer D. Kerbiriou-Nabias D. BioTechniques. 1994; 17: 882-887PubMed Google Scholar). Six hours after thymidine release, cells were electroporated with 10 μg of each plasmid, keeping total DNA concentration constant for each experiment. HUVEC were cotransfected with either HA-JNK1 or HA-JNK2. Cells were stimulated with either human TNFα (100 ng/ml, Peprotech) or HIV-1 Tat, prepared as a GST fusion as previously described (2Gu Y., Wu, R.F., Xu, Y.C. Flores S.C. Terada L.S. Virology. 2001; 286: 62-71Crossref PubMed Scopus (42) Google Scholar). The JNK activity of anti-HA immunoprecipitates was then assessed. We found first that similar to the situation with ECV-304 cells (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar), p47 phox appears to constitutively associate with the actin cytoskeleton of HUVEC. p47-GFP colocalized with submembranous lateral actin bundles as well as actin microspikes in unstimulated HUVEC (Fig. 1). To find binding partners for p47 phox , a whole lung library was chosen for an initial screen because of its high representation of endothelium. 1.7 × 105 transformants were screened, and 80 colonies survived initial auxotrophic selection and were restreaked. Of these, 23 single clones were found to be positive for lacZexpression. PCR amplification and HaeIII restriction digest pattern revealed nine independent clones. Only one clone represented an autonomous transactivation-negative interaction-positive clone containing an in-frame library insert; this clone was bidirectionally sequenced and found to encode the C-terminal 210 amino acid residues of TRAF4. To accomplish a more endothelial-specific screen, 1.9 × 106 clones were screened from the HUVEC library, and only 10 colonies were found to survive auxotrophic selection. Of these, four were lacZ-positive and two were subsequently found to be true positives. These clones were identical by restriction digest analysis. Sequence of one revealed the C-terminal 287 residues of TRAF4. Yeast mating demonstrated lack of binding of p47 phox to full-length TRAF1 and 2 (not shown). When overexpressed in Fx cells, TRAF4 preferentially remained with the detergent-insoluble fraction, with a smaller portion of the protein found in the detergent-soluble fraction (Fig.2 a), mimicking the cytoskeletal distribution of p47 phox (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). When coexpressed, TRAF4 was found to specifically coprecipitate with FLAG-p47 (Fig. 2 b). TRAF4 was also found to coprecipitate with p47-GFP, using anti-GFP, though at reduced efficiency (not shown). In addition, full-length TRAF4 coprecipitated with the C-terminal TRAF domain of TRAF4 (Fig. 2 c), suggesting self-association through this domain. Two-hybrid interactions were used to determine interacting domains of p47 phox and TRAF4. The C terminus of p47 phox -(299–390), containing a variant proline-rich (PR) sequence, an arginine-rich basic motif, and a C-terminal PR site, was both necessary and sufficient for interaction with TRAF4 (Fig. 3 a). Interestingly, the extreme C terminus (347–390) of p47 phox was necessary but by itself insufficient, and the adjacent segment (299–345) was also necessary but insufficient either by itself or attached to the rest of the protein N-terminal to it. The GST pull-down assay confirmed a direct interaction of p47 phox with full-length TRAF4in vitro (Fig. 3 b). In addition, in vitro-translated p47-(347–390) competed with full-length p47 phox for TRAF4 binding, consistent with a specific interaction of the C terminus of p47 phox with TRAF4. The smaller of the TRAF4 library inserts obtained encoded residues 261–470, corresponding to both the 6th exon and the TRAF domain of TRAF4, indicating an interaction of p47 phox with this domain. The TRAF domains of TRAFs 1–6 contain a C-terminal β-sheet-predominant subdomain (TRAF-C) preceded by a shorter coiled-coil subdomain (TRAF-N). By homology and secondary structure prediction, these regions span residues 308–470 and 267–307, respectively, of TRAF4 (20Regnier C.H. Tomasetto C. Moog-Lutz C. Chenard M.P. Wendling C. Basset P. Rio M.C. J. Biol. Chem. 1995; 270: 25715-25721Abstract Full Text Full Text PDF PubMed Scopus (205) Google Scholar). Yeast mating studies suggested that the isolated TRAF-N and TRAF-C domains were each insufficient to bind p47 phox , whereas the entire TRAF domain bound p47 phox (Fig.4). Overexpression of either p47 phox or full-length TRAF4 alone in Fx cells did not appreciably affect JNK activity. In contrast, overexpression of both p47 phox and TRAF4 increased JNK activity greater than 4-fold over control, suggesting a functional as well as physical interaction between the two proteins (Fig.5 a). Similarly, coexpression of p47 phox and TRAF4 increased DCF oxidation, consistent with an increase in oxidant production. To demonstrate that interaction of p47 phox with the TRAF4 TRAF domain was sufficient for JNK activation, we fused this TRAF domain to the C terminus of p47 phox . Overexpression of the fusion protein increased JNK activity in HUVEC (Fig. 5 b), whereas overexpression of either p47 phox or the TRAF4 TRAF domain alone did not. This activation was decreased by the antioxidant N-acetyl cysteine (NAC). To further implicate TRAF4-p47 phox interactions in endogenous endothelial cell signaling pathways, we investigated the effect of TRAF4 p47 phox binding domain overexpression on signaling by TNFα and HIV-1 Tat, two agonists that activate endothelial cell JNK through p47 phox -dependent oxidant production (2Gu Y., Wu, R.F., Xu, Y.C. Flores S.C. Terada L.S. Virology. 2001; 286: 62-71Crossref PubMed Scopus (42) Google Scholar, 12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). Overexpression of this TRAF4 truncation consistently decreased activation of both HA-JNK1 and HA-JNK2 by TNFα and HIV-1 Tat in HUVEC (Fig. 6), whereas full-length TRAF4 did not affect JNK activation (not shown). A secondary screen of the HUVEC library using full-length TRAF4 was performed on 5 × 106 AH109 transformants. Of 92 His+/Ade+/LacZ+ clones, 73 were thought to be unique by PCR and digest pattern. These clones were isolated, passaged through E. coli, and 71 clones were confirmed by mating back to Y187/pGBKT7-T4 and lacZ expression testing. These clones were all sequenced and found to represent 23 unique genes with coding regions in-frame with GAL4-AD, including 4 extracellular proteins, 4 nuclear proteins, and 3 unpublished cDNAs. Of the remaining 12 genes, 8 encoded proteins involved in the determination of cell death and/or proliferation (TableI).Table IcDNAs retrieved from TRAF4 protein interaction screenProteinAbbreviationNo. of clonesGenBankTMMelanoma-associated antigenMAGE1AF200348NADP+-dependent malic enzyme1X79440Non-smooth muscle α-actinin1M95178α-Ketoacid dehydrogenase kinaseBCKDK2XM 008106Ubiquitin conjugase 9UBC95X96427Phospholipid scramblase2NM 021105Hydrogen peroxide-inducible clone-5Hic-52NM 015927Tar-binding protein 2TARBP21NM 004178Arg/Abl binding protein 2ArgBP-21NM 021069Neurotrophin receptor-interacting MAGE homologNRAGE2NM 006986α-Helical coiled-coil rod homologHCR1NM 019052Vascular Rab-GAP/TBC domain-containing proteinVRP2NM 007063 Open table in a new tab TRAF4 is the least well understood of the traditional TRAF family members. Originally identified in a differential screen of metastatic breast cancer lymph nodes (21Tomasetto C. Regnier C. Moog-Lutz C. Mattei M.G. Chenard M.P. Lidereau R. Basset P. Rio M.C. Genomics. 1995; 28: 367-376Crossref PubMed Scopus (228) Google Scholar), it was subsequently shown to be widely expressed in normal human tissues, especially actively dividing epithelium (22Krajewska M. Krajewski S. Zapata J.M. Van Arsdale T. Gascoyne R.D. Berern K. McFadden D. Shabaik A. Hugh J. Reynolds A. Clevenger C.V. Reed J.C. Am. J. Pathol. 1998; 152: 1549-1561PubMed Google Scholar). Its closest relative in both domain organization and amino acid sequence is the Drosophila adapter DTRAF1, which interacts with Misshapen (Msn), a MAP4K acting upstream of JNK (23Liu H., Su, Y.C. Becker E. Treisman J. Skolnik E.Y. Curr. Biol. 1999; 9: 101-104Abstract Full Text Full Text PDF PubMed Scopus (133) Google Scholar). The biological function of TRAF4, however, is poorly understood, beyond its ability to prevent dimerization of the neurotrophin receptor p75NTR (24Ye X. Mehlen P. Rabizadeh S. VanArsdale T. Zhang H. Shin H. Wang J.J. Leo E. Zapata J. Hauser C.A. Reed J.C. Bredesen D.E. J. Biol. Chem. 1999; 274: 30202-30208Abstract Full Text Full Text PDF PubMed Scopus (163) Google Scholar) and to contribute to normal tracheal development in mice (25Shiels H., Li, X. Schumacker P.T. Maltepe E. Padrid P.A. Sperling A. Thompson C.B. Lindsten T. Am. J. Pathol. 2000; 157: 679-688Abstract Full Text Full Text PDF PubMed Scopus (68) Google Scholar). Full-length TRAF4 demonstrated a preference for the detergent-insoluble cell fraction, a distribution similar to that of p47 phox in endothelial cells (12Gu Y., Xu, Y.C., Wu, R.F. Souza R.F. Nwariaku F.E. Terada L.S. Exp. Cell Res. 2002; 272: 62-74Crossref PubMed Scopus (68) Google Scholar). The constitutive association of p47 phox with the cytoskeleton demonstrated in ECV-304 cells in this prior study and in HUVEC in the present study stands in marked contrast to its behavior in neutrophils. In the latter cell type, p47 phox migrates from a cytosolic location to the cytoskeleton and membrane skeleton only upon stimulation (26Nauseef W.M. Volpp B.D. McCormick S. Leidal K.G. Clark R.A. J. Biol. Chem. 1991; 266: 5911-5917Abstract Full Text PDF PubMed Google Scholar, 27El Benna J. Ruedi J.M. Babior B.M. J. Biol. Chem. 1994; 269: 6729-6734Abstract Full Text PDF PubMed Google Scholar). The constitutive association of p47 phox with the endothelial cytoskeleton suggests that the oxidase may exist in a preformed but inactive complex in endothelial cells. Abrupt cessation of flow, for instance, causes oxidant production by endothelial cell NADPH oxidase within 15 s (28Manevich Y., Al- Mehdi A. Muzykantov V. Fisher A.B. Am. J. Physiol. 2001; 280: H2126-H2135PubMed Google Scholar). Although the basis for such constitutive cytoskeletal association is not known, there are clear differences in cytoskeletal structure between adherent endothelial cells and suspended neutrophils. Adhesion-dependent reorganization of the actin cytoskeleton may create or expose p47 binding sites or initiate partial phosphorylation of p47 phox , resulting in cytoskeletal localization. Notably, adhesion primes neutrophils for a robust respiratory burst upon TNFα stimulation in a mechanism dependent upon actin polymerization (29Nathan C.F. J. Clin. Invest. 1987; 80: 1550-1560Crossref PubMed Scopus (684) Google Scholar). A similar mechanism may operate in adherent endothelial cells. The isolated TRAF domain of TRAF4 also demonstrated a preference for the detergent-insoluble fraction. Because the TRAF4 TRAF domain also bound p47 phox , these data do not allow us to determine whether TRAF4 is primarily associated with the cytoskeleton or whether this localization arises secondarily from its association with p47 phox . The retrieval of α-actinin as a potential TRAF4-interacting partner (Table I) is consistent with the former possibility. This situation may be somewhat different from the localization of TRAF5, because the zinc finger motifs of this latter protein rather that its TRAF domain appear to confer detergent insolubility (30Dadgostar H. Cheng G. J. Biol. Chem. 2000; 275: 2539-2544Abstract Full Text Full Text PDF PubMed Scopus (41) Google Scholar). It is equally plausible that TRAF4 and p47 phox each have independent cytoskeletal association domains. The C-terminal tail of p47 phox from residues 299–390 comprised the minimum TRAF4 binding domain we identified. This region encompasses a variant proline-rich site (299–302), a basic region (314–347), and a type II polyproline motif (362–368). In addition, this tail har
Referência(s)