Portal de Periódicos da CAPES

Truncation of Annexin A1 Is a Regulatory Lever for Linking Epidermal Growth Factor Signaling with Cytosolic Phospholipase A2 in Normal and Malignant Squamous Epithelial Cells

2007; Elsevier BV; Volume: 282; Issue: 49 Linguagem: Inglês

10.1074/jbc.m707538200

1083-351X

Masakiyo Sakaguchi, Hitoshi Murata, Hiroyuki Sonegawa, Yoshihiko Sakaguchi, Junichiro Futami, Midori Kitazoe, Hidenori Yamada, Nam Huh,

Biomarkers in Disease Mechanisms

Regulation of cell growth and apoptosis is one of the pleiotropic functions of annexin A1 (ANXA1). Although previous reports on the overexpression of ANXA1 in many human cancers and on growth suppression and/or induction of apoptosis by ANXA1 may indicate the tumor-suppressive nature of ANXA1, molecular mechanisms of the function of ANXA1 remain largely unknown. Here we provide evidence that ANXA1 mechanistically links the epidermal growth factor-triggered growth signal pathway with cytosolic phospholipase A2 (cPLA2), an initiator enzyme of the arachidonic acid cascade, through interaction with S100A11 in normal human keratinocytes (NHK). Ca2+-dependent binding of S100A11 to ANXA1 facilitated the binding of the latter to cPLA2, resulting in inhibition of cPLA2 activity, which is essential for the growth of NHK. On exposure of NHK to epidermal growth factor, ANXA1 was cleaved solely at Trp12, and this cleavage was executed by cathepsin D. In squamous cancer cells, this pathway was shown to be constitutively activated. The newly found mechanistic intersection may be a promising target for establishing new measures against human cancer and other cell growth disorders. Regulation of cell growth and apoptosis is one of the pleiotropic functions of annexin A1 (ANXA1). Although previous reports on the overexpression of ANXA1 in many human cancers and on growth suppression and/or induction of apoptosis by ANXA1 may indicate the tumor-suppressive nature of ANXA1, molecular mechanisms of the function of ANXA1 remain largely unknown. Here we provide evidence that ANXA1 mechanistically links the epidermal growth factor-triggered growth signal pathway with cytosolic phospholipase A2 (cPLA2), an initiator enzyme of the arachidonic acid cascade, through interaction with S100A11 in normal human keratinocytes (NHK). Ca2+-dependent binding of S100A11 to ANXA1 facilitated the binding of the latter to cPLA2, resulting in inhibition of cPLA2 activity, which is essential for the growth of NHK. On exposure of NHK to epidermal growth factor, ANXA1 was cleaved solely at Trp12, and this cleavage was executed by cathepsin D. In squamous cancer cells, this pathway was shown to be constitutively activated. The newly found mechanistic intersection may be a promising target for establishing new measures against human cancer and other cell growth disorders. Annexin A1 (ANXA1) 2The abbreviations used are: ANXA1, annexin A1; NHK, normal human keratinocytes; EGF, epidermal growth factor; EMSA, electromobility shift assay; siRNA, small interfering RNA; GST, glutathione S-transferase; TGF, transforming growth factor; cPLA2, cytosolic phospholipase A2; GFP, green fluorescent protein; SCC, squamous carcinoma cell. 2The abbreviations used are: ANXA1, annexin A1; NHK, normal human keratinocytes; EGF, epidermal growth factor; EMSA, electromobility shift assay; siRNA, small interfering RNA; GST, glutathione S-transferase; TGF, transforming growth factor; cPLA2, cytosolic phospholipase A2; GFP, green fluorescent protein; SCC, squamous carcinoma cell. belongs to the annexin superfamily composed of 13 proteins sharing a feature of Ca2+-dependent binding to phospholipids (1Gerke V. Moss S.E. Physiol. Rev. 2002; 82: 331-371Crossref PubMed Scopus (1610) Google Scholar, 2Lim L.H. Pervaiz S. FASEB J. 2007; 21: 968-975Crossref PubMed Scopus (323) Google Scholar). ANXA1 plays pleiotropic roles in various biological contexts, including membrane organization, membrane traffic, and regulation of intracellular Ca2+. Physiologically, it is involved in inflammation and at least partly mediates the function of glucocorticoid (1Gerke V. Moss S.E. Physiol. Rev. 2002; 82: 331-371Crossref PubMed Scopus (1610) Google Scholar, 2Lim L.H. Pervaiz S. FASEB J. 2007; 21: 968-975Crossref PubMed Scopus (323) Google Scholar, 3Parente L. Solito E. Inflamm. Res. 2004; 53: 125-132Crossref PubMed Scopus (238) Google Scholar). ANXA1 is also relevant to regulation of cell growth and apoptosis. Reduced expression of ANXA1 has been observed in many different human malignancies, including esophageal cancer (4Hu N. Flaig M.J. Su H. Shou J.Z. Roth M.J. Li W.J. Wang C. Goldstein A.M. Li G. Emmert-Buck M.R. Taylor P.R. Clin. Cancer Res. 2004; 10: 6013-6022Crossref PubMed Scopus (73) Google Scholar), prostate cancer (5Kang J.S. Calvo B.F. Maygarden S.J. Caskey L.S. Mohler J.L. Ornstein D.K. Clin. Cancer Res. 2002; 8: 117-123PubMed Google Scholar), breast cancer (6Shen D. Chang H.R. Chen Z. He J. Lonsberry V. Elshimali Y. Chia D. Seligson D. Goodglick L. Nelson S.F. Gornbein J.A. Biochem. Biophys. Res. Commun. 2005; 326: 218-227Crossref PubMed Scopus (63) Google Scholar), and B-cell lymphoma (7Vishwanatha J.K. Salazar E. Gopalakrishnan V.K. BMC Cancer. 2004; 4: 8Crossref PubMed Scopus (60) Google Scholar). In addition, overexpression of ANXA1 in malignant tumor cells resulted in growth suppression and/or apoptosis (8Hsiang C.H. Tunoda T. Whang Y.E. Tyson D.R. Ornstein D.K. Prostate. 2006; 66: 1413-1424Crossref PubMed Scopus (55) Google Scholar). Thus, ANXA1 may be regarded as a tumor suppressor gene (8Hsiang C.H. Tunoda T. Whang Y.E. Tyson D.R. Ornstein D.K. Prostate. 2006; 66: 1413-1424Crossref PubMed Scopus (55) Google Scholar). Molecular mechanisms of the tumor-suppressive function of ANXA1, however, remain largely unknown.The fact that ANXA1 inhibits the activity of cytosolic phospholipase A2 (cPLA2; 3) could be a clue for a better understanding of the function of ANXA1 in cell growth regulation, because cPLA2 is an initiator enzyme of the arachidonic acid cascade and thus plays a critical role in the regulation of growth of many different types of cells (9Flower R.J. Blackwell G.J. Biochem. Pharmacol. 1976; 25: 285-291Crossref PubMed Scopus (522) Google Scholar, 10Nakanishi M. Rosenberg D.W. Biochim. Biophys. Acta. 2006; 1761: 1335-1343Crossref PubMed Scopus (97) Google Scholar). Inhibition of cPLA2 activity is thought to be because of direct binding of ANXA1 to cPLA2 (11Kim K.M. Kim D.K. Park Y.M. Kim C.K. Na D.S. FEBS Lett. 1994; 343: 251-255Crossref PubMed Scopus (82) Google Scholar, 12Croxtall J.D. Choudhury Q. Tokumoto H. Flower R.J. Biochem. Pharmacol. 1995; 50: 465-474Crossref PubMed Scopus (69) Google Scholar) rather than to sequestering of substrate phospholipids by the binding of ANXA1 to them (13Davidson F.F. Dennis E.A. Powell M. Glenney Jr., J.R. J. Biol. Chem. 1987; 262: 1698-1705Abstract Full Text PDF PubMed Google Scholar, 14Haigler H.T. Schlaepfer D.D. Burgess W.H. J. Biol. Chem. 1987; 262: 6921-6930Abstract Full Text PDF PubMed Google Scholar). It is also known that ANXA1 is phosphorylated by the ligand-activated EGF receptor at Tyr21 (14Haigler H.T. Schlaepfer D.D. Burgess W.H. J. Biol. Chem. 1987; 262: 6921-6930Abstract Full Text PDF PubMed Google Scholar, 15Pepinsky R.B. Sinclair L.K. Nature. 1986; 321: 81-84Crossref PubMed Scopus (216) Google Scholar) and that phosphorylated ANXA1 becomes labile to tryptic cleavage at an N-terminal site(s) in vitro (14Haigler H.T. Schlaepfer D.D. Burgess W.H. J. Biol. Chem. 1987; 262: 6921-6930Abstract Full Text PDF PubMed Google Scholar). However, the functional significance of the phosphorylation and intracellular cleavage of ANXA1 for cell growth induced by EGF and the possible mechanistic link to the interaction between ANXA1 and cPLA2 are not well understood.On the other hand, ANXA1 is known to bind another Ca2+-binding protein, S100A11, via the N terminus of ANXA1 and the C terminus of S100A11 (16Mailliard W.S. Haigler H.T. Schlaepfer D.D. J. Biol. Chem. 1996; 271: 719-725Abstract Full Text Full Text PDF PubMed Scopus (106) Google Scholar, 17Seemann J. Weber K. Gerke V. Biochem. J. 1996; 319: 123-129Crossref PubMed Scopus (71) Google Scholar). Although a model for membrane rearrangement by the ANXA1-S100A11 complex has been proposed (1Gerke V. Moss S.E. Physiol. Rev. 2002; 82: 331-371Crossref PubMed Scopus (1610) Google Scholar), the biological relevance and action mechanism of the complex formation remain to be clarified. We recently demonstrated that S100A11 is involved in TGFβ-triggered signaling (18Sakaguchi M. Miyazaki M. Sonegawa H. Kashiwagi M. Ohba M. Kuroki T. Namba M. Huh N.H. J. Cell Biol. 2004; 164: 979-984Crossref PubMed Scopus (69) Google Scholar, 19Sakaguchi M. Sonegawa H. Nukui T. Sakaguchi Y. Miyazaki M. Namba M. Huh N.H. Proc. Natl. Acad. Sci. U. S. A. 2005; 102: 13921-13926Crossref PubMed Scopus (36) Google Scholar). On exposure of normal human keratinocytes (NHK) to TGFβ, S100A11 was phosphorylated by protein kinase Cα and transferred to the nucleus, resulting in induction of p21WAF1. The growth inhibition was almost completely mitigated when this pathway was functionally blocked. The S100A11-mediated pathway was deteriorated in squamous carcinoma cell lines that are resistant to TGFβ (20Sonegawa H. Nukui T. Li D.W. Takaishi M. Sakaguchi M. Huh N.H. J. Mol. Med. 2007; 85: 753-762Crossref PubMed Scopus (5) Google Scholar). These results indicate that, in addition to the well characterized Smad-mediated pathway, the S100A11-mediated pathway is involved in and essential for the growth inhibition of NHK by TGFβ. The newly found pathway, however, does not involve the interaction of S100A11 with ANXA1.In this study, we addressed the questions of whether and how the three observations described above, i.e. EGF-triggered phosphorylation and cleavage of ANXA1, inhibition of cPLA2 by ANXA1, and binding of ANXA1 to S100A11, are mechanistically linked.EXPERIMENTAL PROCEDURESCells—NHK (KURABO, Osaka, Japan) were cultured in EpiLife medium (Cascade Biologics, Portland, OR) with the growth supplement HKGS (Cascade Biologics). A human vulvar epidermoid carcinoma cell line, A431, and human cutaneous squamous carcinoma cell lines, BSCC-93 and DJM-1, were purchased from ATCC. A431 was cultured in Dulbecco's modified Eagle's medium with 4 mm/liter glutamate and 4.5 g/liter glucose. BSCC-93 and DJM-1 were cultured in Dulbecco's modified Eagle's medium. Both media were supplemented with 10% fetal bovine serum. NHK of passage 2–4 were used throughout the experiments. For monitoring DNA synthesis, tritiated thymidine (1 μCi/ml; American Radiolabeled Chemicals, St. Louis, MO) was added to the cultures 1 h before harvest.Materials—An EGF receptor inhibitor (AG1478), arachidonyltrifluoromethyl ketone, and an synthetic cPLA2α inhibitor were purchased from Calbiochem. Recombinant human EGF and TGFβ were purchased from PeproTech EC (London, UK) and Sigma, respectively.Plasmid Constructs and Preparation of Recombinant Proteins—pGEX vectors (GE Healthcare) were used to produce glutathione S-transferase (GST) fusion proteins of full-length human S100A11 and ANXA1 in Escherichia coli. cDNAs of phosphorylation-mimic mutants of S100A11 were obtained by replacing either Thr10 (P-N) or Ser94 (P-C) or both sites (P-NC) with Asp by conventional site-directed mutagenesis. The fused proteins were purified using a Sephadex 4B column (GE Healthcare) under conventional conditions. When necessary, GST was released by cleaving with PreScission protease (GE Healthcare) and removed from the final preparations using a Sephadex 4B column.Adenovirus Constructs—Adenovirus constructs were prepared using Adeno-X Expression System 2 (Clontech) under the conditions recommended by the manufacturer. S100A11 and its variants were made as a fusion protein of nuclear export signal-GFP-S100A11 to avoid the growth-interfering function of S100A11 in nuclei (19Sakaguchi M. Sonegawa H. Nukui T. Sakaguchi Y. Miyazaki M. Namba M. Huh N.H. Proc. Natl. Acad. Sci. U. S. A. 2005; 102: 13921-13926Crossref PubMed Scopus (36) Google Scholar). An S100A11 variant lacking Ca2+-binding capacity (ΔCa; 21Sakaguchi M. Miyazaki M. Takaishi M. Sakaguchi Y. Makino E. Kataoka N. Yamada H. Namba M. Huh N.H. J. Cell Biol. 2003; 163: 825-835Crossref PubMed Scopus (100) Google Scholar) was produced by mutating Asp at the 28, 68, 72, and 76 sites and Glu at the 38 and 79 sites to Ser. ANXA1 and its variants were made as a fusion protein of Ds-Red-ANXA1-GFP. An ANXA1 variant lacking the phosphorylation site of Tyr21 (ΔP) and an N-terminal truncated form (ΔN) were made by replacing with Phe and by deleting N-terminal 39 amino acids, respectively, using conventional site-directed mutagenesis.Immunological Analyses—Western blot analysis and immunoprecipitation were performed under conventional conditions. To get optimal pictures, we usually repeated Western blotting three times for each set of samples. The following antibodies were used: rabbit anti-human S100A11 antibody (raised in our laboratory 21); rabbit anti-human annexin A1 antibody (raised in our laboratory; confirmed to specifically recognize native and denatured human ANXA1); mouse anti-phosphotyrosine antibody (clone PY20; Chemicon, Temecula, CA); mouse anti-human cPLA2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA); mouse anti-EGF receptor antibody (Lab Vision, Fremont, CA); goat anti-human cathepsin B antibody (Santa Cruz Biotechnology); goat anti-human cathepsin D antibody (Santa Cruz Biotechnology); goat anti-cathepsin L antibody (Santa Cruz Biotechnology); and mouse anti-human tubulin antibody (Sigma). The second antibody used was horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG antibody (Cell Signaling Technology). Protein bands were visualized by a chemiluminescence system (ECL Plus, GE Healthcare).RNA Interference and Protein Transduction—Validated small interfering RNAs (siRNAs) for human cathepsins B (ID 105579), D (ID 105581 and ID 4180), and L (ID 105042) and control siRNA (glyceraldehyde-3-phosphate dehydrogenase) were purchased from Ambion (Austin, TX). Transfection of siRNA was performed using Lipofectamine 2000 (Invitrogen). Cathepsin D (Sigma) was labeled with Cy3 using a FluoroLink-Ab Cy3 labeling kit (GE Healthcare) and transduced into cells using a Chariot kit (Active Motif, Carlsbad, CA).Electromobility Shift Assay (EMSA)—EMSA was performed as described previously (21Sakaguchi M. Miyazaki M. Takaishi M. Sakaguchi Y. Makino E. Kataoka N. Yamada H. Namba M. Huh N.H. J. Cell Biol. 2003; 163: 825-835Crossref PubMed Scopus (100) Google Scholar). Briefly, a 32P-labeled Myc/Max consensus DNA probe (Santa Cruz Biotechnology) was mixed with 2 μg of crude nuclear extracts in a reaction mixture (20 μl) containing 60 mm KCl, 5 mm MgCl2, 0.1 mm EDTA, 1 μgof poly(dI-dC), 1 mm dithiothreitol, 5% glycerol, and 20 mm Hepes, pH 7.9. DNA-protein complexes were then separated by electrophoresis in a 5% polyacrylamide gel under nondenaturing conditions.Affinity Column Chromatography with ANXA1 N-terminal Peptide—Human ANXA1 N-terminal peptide (AMVSEFLK-QAWFIENEEQ) was covalently conjugated to Tresyl beads (TOYOPEAL AF-Tresyl-650 M; Tosoh, Tokyo, Japan). Cell extracts prepared with either neutral (0.5% Triton X-100, 1 mm dithiothreitol, 50 mm Tris-HCl, pH 7.4) or acidic buffer (0.5% Triton X-100, 1 mm dithiothreitol, 50 mm citric acid, pH 4.5) were used, and eluates with the same buffers were analyzed by electrophoresis.Mass Spectrometry—Protein bands visualized by Coomassie Brilliant Blue staining were treated with a trypsin profile IGD kit (Sigma) and analyzed using an Agilent 1100 LC/MSD Trap XCT Ultra series system with an ionizing system of HPLC-Chip-MS (Agilent Technologies, Santa Clara, CA). Data base searching and identification of proteins were carried out by a software Spectrum Mill MS Proteomics Workbench (version Rev A. 03. 02.060b/Agilent Technologies) and accessing to the NCBInr public data bases.Acquisition and Processing of Images—For fluorescence-labeled cells, images were acquired by a laser-scanning microscope (type, Axioplan 2; objective lens, Plan-Apochromat63 × 1.4 oil DC and Plan-Neofluar40x/0.75; Carl Zeiss, Oberkochen, Germany) and processed using Adobe Photoshop 6.0.RESULTSInteraction of S100A11, ANXA1, and cPLA2 and Its Effect on the Growth of NHK—First, we confirmed binding between S100A11 and ANXA1. Recombinant S100A11 was pulled down by GST-ANXA1 in vitro (Fig. 1A). The binding appears Ca2+-dependent because the addition of EGTA or the use of an S100A11 variant lacking Ca2+-binding capacity (ΔCa) largely reduced the amount of S100A11 precipitated with GST-ANXA1.Phosphorylation-mimic variants of S100A11 at two potential sites, i.e. Thr10 and Ser94 (18Sakaguchi M. Miyazaki M. Sonegawa H. Kashiwagi M. Ohba M. Kuroki T. Namba M. Huh N.H. J. Cell Biol. 2004; 164: 979-984Crossref PubMed Scopus (69) Google Scholar), showed a binding mode similar to that of the wild-type protein. Endogenous S100A11 and cPLA2 were pulled down by GST-ANXA1 added to the cell extracts (supplemental Fig. S1). Recombinant S100A11 added to the incubation mixture dose-dependently increased the amount of cPLA2 bound to and hence recovered by GST-ANXA1, indicating that the formation of ANXA1-S100A11 complex facilitates the binding of ANXA1 to cPLA2. Under similar conditions, GST-S100A11 pulled down endogenous ANXA1 and cPLA2 (data not shown). Immunoprecipitation of endogenous ANXA1 in resting NHK co-precipitated S100A11 and cPLA2 (Fig. 1B and 2A, left-hand columns). Overexpression of wild-type S100A11 using an adenovirus vector slightly increased the co-precipitable cPLA2 (Fig. 1B). ΔCa S100A11 abrogated even the binding of endogenous S100A11 to ANXA1 and cPLA2, thus functioning as a dominant negative agent (supplemental Fig. S2). Phosphorylation-mimic S100A11 variants behaved in a manner similar to that of the wild type in cells.FIGURE 2Cleavage of ANXA1 by EGF stimulation abrogates its binding capacity to S100A11 and cPLA2. A, cleavage of ANXA1 and loss of binding capacity of ANXA1 to S100A11 and cPLA2 by EGF. NHK were cultivated with 10 ng/ml EGF and analyzed by Western blotting with or without prior immunoprecipitation with an antibody against ANXA1. P-Y, an antibody against phosphotyrosine. Only a shifted band region is shown for EMSA using a Myc probe as an indicator for cell growth. B, ANXA1 was purified using an affinity column with anti-ANXA1 antibody from extracts of NHK cultivated with 10 ng/ml EGF in HKGS-free EpiLife medium and stained with Coomassie Brilliant Blue (CBB) after electrophoresis. Open arrowhead, intact ANXA1; closed arro-w head, cleaved ANXA1. C, NHK were infected with an adenovirus vector carrying DR-ANXA1-GFP. Thirty six hours later, 10 ng/ml EGF was added, and cells were observed after further incubation for 6 h. Two independent experiments gave rise to similar results. D, effect of overexpression of S100A11 and ANXA1 on EGF-induced ANXA1 cleavage. NHK were treated in a similar manner to that described in C. Each cell extract was analyzed by Western blotting with or without prior immunoprecipitation. Only a region of shifted bands is shown for EMSA using a Myc probe. DR, Ds-red; NES, nuclear export signal; ter, terminal. For other abbreviations, see the legend for Fig. 1.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Next, we examined the effect of binding of ANXA1 to cPLA2 that was facilitated by S100A11. Intracellular cPLA2 activity was monitored by determining radioactivity released from cells labeled with [3H]arachidonic acid in advance. Treatment of NHK with EGF resulted in marked activation of cPLA2 (Fig. 1C, upper panel). Overexpression of wild-type S100A11 and ANXA1 significantly suppressed the activation. NHK expressing the ΔCa S100A11 variant showed a higher level of cPLA2 activity than that of the control cells, indicating a dominant negative effect of ΔCa S100A11 on endogenous S100A11 with respect not only to binding to ANXA1 (Fig. 1B) but also to inhibition of cPLA2 activity. A truncated form of ANXA1 lacking the N-terminal 39 amino acids (ΔN) lost the capacity to inhibit cPLA2 as expected from the lack of binding to S100A11 (Fig. 1C, upper panel). In contrast, overexpression of a phosphorylation site mutant of ANXA1 (ΔP: Tyr21 to Phe21) inhibited cPLA2 activity even to a higher extent than that in the case of the wild type.Growth of NHK was shown to largely depend on the arachidonic acid cascade. Addition of cPLA2 inhibitors, arachidonyltrifluoromethyl ketone and an synthetic cPLA2α inhibitor (Calbiochem), abrogated DNA synthesis induced by EGF, and this abrogation was canceled by adding arachidonic acid to the medium (supplemental Fig. S3). cPLA2 is an enzyme triggering the arachidonic cascade (9Flower R.J. Blackwell G.J. Biochem. Pharmacol. 1976; 25: 285-291Crossref PubMed Scopus (522) Google Scholar), and arachidonic acid was reported to play a critical role in growth of mouse epidermal keratinocytes (22Lo H.H. Teichmann P. Furstenberger G. Gimenez-Conti I. Fischer S.M. Cancer Res. 1998; 58: 4624-4631PubMed Google Scholar). Overexpression of cPLA2 alone stimulated DNA synthesis, which was blocked by the cPLA2 inhibitor (supplemental Fig. S4). DNA synthesis of NHK overexpressed with the various variants of S100A11 and ANXA1 was strongly correlated with the activity of cPLA2 except for the cells with ΔCa S100A11 (Fig. 1C, lower panel), and this was probably due to saturation of cPLA2 activity. These results indicate that Ca2+-dependent formation of the ANXA1-S100A11 complex greatly facilitates the binding of ANXA1 to cPLA2 and that ANXA1 efficiently inhibits cPLA2 activity only when complexed with S100A11.Cleavage of ANXA1 on Exposure of NHK to EGF and Resulting Abrogation of Its Binding to S100A11 and cPLA2—When NHK were exposed to EGF, a fast migrating ANXA1 appeared depending on time (Fig. 2A). No change was observed in amounts and electromobility of S100A11 and cPLA2. In resting NHK, ANXA1 was not phosphorylated, and immunoprecipitation of ANXA1 pulled down S100A11 as well as cPLA2. On exposure of NHK to EGF, phosphotyrosine was detected in both forms of ANXA1. As expected from the fact that ANXA1 binds to S100A11 at its N-terminal region (16Mailliard W.S. Haigler H.T. Schlaepfer D.D. J. Biol. Chem. 1996; 271: 719-725Abstract Full Text Full Text PDF PubMed Scopus (106) Google Scholar), the truncation of ANXA1 was associated with a decrease in the amounts of S100A11 and cPLA2 pulled down by anti-ANXA1 antibody. The two forms of ANXA1 in EGF-treated NHK were immunoprecipitated and separated by electrophoresis (Fig. 2B). Conventional automated amino acid sequencing failed to reveal the protein recovered from the upper band probably because of the N-terminal capping. The protein was identified to be ANXA1 by tandem mass spectrometry. The N-terminal sequence of the lower band was determined to be FIENE, indicating that ANXA1 was cleaved solely at the C-terminal side of Trp12 in NHK on exposure to EGF in vivo. Cleavage of ANXA1 induced by EGF was also confirmed visually by expressing a fusion protein of Ds-Red, ANXA1, and GFP (Fig. 2C). Intracellular localization of the fusion protein, i.e. preferential accumulation on the cell membrane with diffuse distribution in the cytoplasm, was similar to that of endogenous ANXA1 (not shown) in accordance with a report by Radke et al. (23Radke S. Austermann J. Russo-Marie F. Gerke V. Rescher U. FEBS Lett. 2004; 578: 95-98Crossref PubMed Scopus (29) Google Scholar). On exposure of NHK to EGF, the signal from Ds-Red with the N-terminal portion segregated from the signal from GFP fused to the bulk of ANXA1.To further investigate the interaction between S100A11 and ANXA1 in relation to binding to cPLA2, we prepared a series of adenovirus constructs carrying various variants of S100A11 and ANXA1 (see "Experimental Procedures"). NHK were infected with both S100A11 and ANXA1 adenoviruses and analyzed by Western blotting with or without prior immunoprecipitation (Figs. 1B and 2D). In NHK not treated with EGF, overexpression of wild-type S100A11 or/and ANXA1 enhanced their binding to cPLA2, and overexpression of ΔCa S100A11 functioned as a dominant negative agent for endogenous S100A11. On exposure to EGF, exogenous ANXA1 was also cleaved. This cleavage was suppressed by concomitant expression of wild-type S100A11, probably because the complex formation conferred ANXA1 resistance to proteolysis induced by EGF. Phosphorylation-mimic S100A11 variants showed characteristics similar to those of wild-type S100A11 (Fig. 1C). ΔP ANXA1 was never cleaved, indicating that phosphorylation of ANXA1 at Tyr21 is a prerequisite for subsequent cleavage. These results indicate that on exposure of NHK to EGF, ANXA1 is phosphorylated at Tyr21 and eventually cleaved at Trp12 and that cleaved ANXA1 loses its binding capacity to S100A11 and hence to cPLA2, resulting in abrogation of the inhibition of cPLA2 activity. It should be noted that this mechanism is in accordance with the results of cell growth monitored by [3H]thymidine incorporation (Fig. 1C) or by electromobility shift assay (EMSA) using a c-Myc probe (Fig. 2D, bottom).Identification of a Protease for ANXA1 Cleavage as Cathepsin D—To identify a protease(s) responsible for the EGF-induced cleavage of ANXA1, we screened proteins binding to the N-terminal peptide covering the cleavage site of ANXA1. A band with a molecular mass of ∼30 kDa showing a differential binding between the control and N-terminal peptide columns was identified (Fig. 3A). Tandem mass spectrometry resulted in robust identification of the protein as human cathepsin D. When cathepsin D was depleted using siRNA, cleavage of ANXA1 induced by EGF was completely inhibited (Fig. 3B). Inhibition of the cleavage of ANXA1 resulted in recovery of the binding capacity of ANXA1 to S100A11 and cPLA2 as demonstrated by immunoprecipitation followed by Western blot analysis. The function to cleave ANXA1 appears to be specific to cathepsin D, because down-regulation of cathepsin B or L showed no effect (Fig. 3C). Inhibition of ANXA1 cleavage resulted in growth inhibition as monitored by EMSA using a c-Myc probe (Fig. 3C, bottom). Blocking of EGF-induced truncation of ANXA1 by depletion of cathepsin D using siRNA was canceled by transducing exogenous cathepsin D protein (Fig. 3D). siRNAs targeted to two different sites of cathepsin D showed the same effect.FIGURE 3Identification of protease responsible for EGF-induced cleavage of ANXA1 as cathepsin D. A, identification of protein that binds to an N-terminal peptide of ANXA1. Untreated NHK were lysed with either an acidic or neutral buffer and applied onto the affinity column. Eluates were electrophoresed and stained with Coomassie Brilliant Blue. B, effect of siRNAs on EGF-induced cleavage of ANXA1. After transfection with siRNAs, NHK were cultivated in EpiLife with HKGS for 24 h and then in HKGS-free EpiLife for another 24 h, washed twice with a 3-h interval, and then exposed to EGF for 6 h before harvest. Cat, cathepsin; Y-P, anti-phosphotyrosine antibody. IP, immunoprecipitate; WB, Western blot. C, effect of siRNA of cathepsin D on the response of NHK to EGF. NHK were treated in a similar manner to that described in B. Only a region of shifted bands is shown for EMSA using a Myc probe. D, rescue of cathepsin D-depleted cells by transduction of the protein. NHK were cultivated and treated under conditions similar to those described in C, except for transduction of exogenous cathepsin D protein 6 h prior to addition of EGF. GST was used as a control. Left panel, Cy3-labeled cathepsin D in NHK. Two independent experiments gave rise to similar results. Right panel, Western blot analysis.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Effect of High Ca2+ or TGFβ on Growth of NHK—High Ca2+ and TGFβ are representative growth suppressors of NHK. Addition of either agent overrides growth of NHK supported by EGF as confirmed in an experiment for which results are shown in Fig. 4A. When NHK were cultured in a medium with 1.5 mm Ca2+, EGF-induced phosphorylation and subsequent cleavage of ANXA1 were completely blocked (Fig. 4B). High Ca2+ induced phosphorylation of S100A11, which was not affected by EGF (supplemental Fig. S5). Phosphorylation of S100A11 at Thr10 is a triggering event for growth suppression of NHK (21Sakaguchi M. Miyazaki M. Takaishi M. Sakaguchi Y. Makino E. Kataoka N. Yamada H. Namba M. Huh N.H. J. Cell Biol. 2003; 163: 825-835Crossref PubMed Scopus (100) Google Scholar). Increased amounts of S100A11 and cPLA2 were recovered by immunoprecipitation using an anti-ANXA1 antibody, indicating that high Ca2+ facilitates more efficient formation of S100A11-ANXA1 complex (Fig. 4C) and hence inhibits liberation and reactivation of cPLA2 by EGF. Intracellular Ca2+ concentration increased in NHK exposed to high Ca2+ but not in those exposed to TGFβ (data not shown). In TGFβ-treated NHK, the phosphorylation, cleavage, and binding mode of ANXA1 were similar to those in the control cells (Fig. 4, B and C). However, the amount of cPLA2 protein decreased depending on time, and a reduced level of cPLA2 mRNA was specifically observed in NHK treated with TGFβ (supplemental Fig. S6).FIGURE 4Effect of co-treatment with EGF and either Ca2+ or TGFβ. A, abrogation of EGF-induced growth stimulation of NHK by Ca2+ or TGFβ. NHK were treated under conditions similar to those described in the legend for Fig. 1C. Ca2+ (1.5 mm) and TGFβ (1 ng/ml) were added simultaneously with EGF (10 ng/ml). Two independent experiments were performed, giving rise to similar results. Standard deviations were obtained from triplicate determinations. B, NHK were cultured in HKGS-free EpiLife for 24 h, washed twice with an interval of 3 h, and then incubated with the additives at the same concentrations as those stated in A. Each cell extract was analyzed by Western blotting (WB) with or without prio