Soluble Axl Is Generated by ADAM10-Dependent Cleavage and ...

MOLECULAR AND CELLULAR BIOLOGY, Nov. 2005, p. 9324–9339 Vol. 25, No. 210270-7306/05/$08.00�0 doi:10.1128/MCB.25.21.9324–9339.2005Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Soluble Axl Is Generated by ADAM10-Dependent Cleavage andAssociates with Gas6 in Mouse Serum†

Vadim Budagian,1‡* Elena Bulanova,1‡ Zane Orinska,1 Erwin Duitman,1 Katja Brandt,1Andreas Ludwig,2 Dieter Hartmann,3 Greg Lemke,4 Paul Saftig,2 and Silvia Bulfone-Paus1

Department of Immunology and Cell Biology, Research Center Borstel, D-23845 Borstel,1 and Institute of Biochemistry,Christian Albrechts University, D-24118 Kiel,2 Germany; Department of Human Genetics, KU Leuven, and

Flanders Interuniversity Institute for Biotechnology (VIB-4), B-3000 Leuven, Belgium3; and MolecularNeurobiology Laboratory, Salk Institute for Biological Studies, San Diego, California 921864

Received 15 May 2005/Returned for modification 8 June 2005/Accepted 13 August 2005

Axl receptor tyrosine kinase exists as a transmembrane protein and as a soluble molecule. We show thatconstitutive and phorbol 12-myristate 13-acetate-induced generation of soluble Axl (sAxl) involves the activityof disintegrin-like metalloproteinase 10 (ADAM10). Spontaneous and inducible Axl cleavage was inhibited bythe broad-spectrum metalloproteinase inhibitor GM6001 and by hydroxamate GW280264X, which is capableof blocking ADAM10 and ADAM17. Furthermore, murine fibroblasts deficient in ADAM10 expression exhib-ited a significant reduction in constitutive and inducible Axl shedding, whereas reconstitution of ADAM10restored sAxl production, suggesting that ADAM10-mediated proteolysis constitutes a major mechanism forsAxl generation in mice. Partially overlapping 14-amino-acid stretch deletions in the membrane-proximalregion of Axl dramatically affected sAxl generation, indicating that these regions are involved in regulating theaccess of the protease to the cleavage site. Importantly, relatively high circulating levels of sAxl are present inmouse sera in a heterocomplex with Axl ligand Gas6. Conversely, two other family members, Tyro3 and Mer,were not detected in mouse sera and conditioned medium. sAxl is constitutively released by murine primarycells such as dendritic and transformed cell lines. Upon immobilization, sAxl promoted cell migration andinduced the phosphorylation of Axl and phosphatidylinositol 3-kinase. Thus, ADAM10-mediated generation ofsAxl might play an important role in diverse biological processes.

Receptor tyrosine kinases (RTKs) play fundamental roles indiverse cell functions, including proliferation, differentiation,survival, migration, and metabolism (16). Axl RTK (alsoknown as Ark, Ufo, and Tyro7) is the prototype of a family oftransmembrane receptors, which also includes Tyro3 (alsoknown as Sky, Brt, Etk, Tif, Dtk, and Rse) and Mer (c-Eyk,Nyk, and Tyro12) (34, 44, 64). They share a distinct molecularstructure characterized by two immunoglobulin-like motifs andtwo fibronectin type III repeats in their extracellular domainand a cytoplasmic domain that contains a conserved catalytickinase region (34, 44). Axl, Tyro3, and Mer are variably ex-pressed in neural, lymphoid, vascular, and reproductive tissuesand in different primary cells and tumor cell lines (11, 41, 42).Mutant mice that lack these three receptors have a defectivephagocytic clearance of apoptotic cells and impaired spermat-ogenesis (41) and develop a severe lymphoproliferative disor-der accompanied by broad-spectrum autoimmunity (42).

A common heterophilic ligand for these RTK family mem-bers is Gas6, a vitamin K-dependent protein that is widelysecreted by most tissues, including the lungs, intestine, andvascular endothelium (43). Gas6 is the product of growth ar-rest-specific gene 6, which was initially cloned from serum-

starved fibroblasts and shares about 44% sequence identity andsimilar domain organization with protein S, a negative regula-tor of blood coagulation (48). Recent studies indicate that theGas6/Axl system plays an important role in vascular biology(46). A large amount of experimental evidence supports a rolefor Gas6/Axl signaling in cell growth and protection from ap-optosis in normal and cancer cells (10, 24, 31). Axl activationresults in autophosphorylation and phosphorylation of cyto-plasmic substrates, including phosphatidylinositol 3-kinase(PI3K), Akt, S6K, Src kinase, ERK, p38, STAT3, and NF-�B(2, 29, 32, 35, 62, 68).

The extracellular regions of Axl, Tyro3, and Mer containsimilar combinations of structural motifs, which are also ob-served in the receptor-type protein tyrosine phosphatases andadhesion molecules of the cadherin and immunoglobulin su-perfamily (67). Several studies demonstrated that Axl couldmediate cell adhesion and aggregation through homotypicectodomain associations (9, 23). Both murine and human Axlproteins undergo proteolytic processing to yield a soluble formof this molecule. Murine Axl is cleaved extracellularly to gen-erate a soluble ectodomain of approximately 65 kDa (23),whereas cleavage of human Axl is mapped to the 14-amino-acid (aa) stretch in the extracellular region and corresponds tothe soluble form with a higher molecular mass of 80 kDa (50).Soluble Axl (sAxl) is present in cell-conditioned medium oftumor cells growing in vivo and in vitro and in the sera ofhumans, mice, and rats (23, 50). However, the identities of thesAxl-generating protease(s) and the mechanism(s) that ac-count for this process remain unknown.

Ectodomain shedding has emerged as an important post-

* Corresponding author. Mailing address: Department of Immunol-ogy and Cell Biology, Research Center Borstel, Parkallee 22, D-23845Borstel, Germany. Phone: 49 4537 188564. Fax: 49 4537 188403. E-mail: [email protected].

† Supplemental material for this article may be found at http://mcb.asm.org/.

‡ V.B. and E.B. contributed equally to this work.

9324

RETRACTED

on January 30, 2018 by guesthttp://m

cb.asm.org/

Dow

nloaded from


cb.asm.org/

Dow

nloaded from


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

http://mcb.asm.org/

http://mcb.asm.org/

translational mechanism to regulate the functions of variousintegral membrane-bound proteins, including adhesion mole-cules, cytokines, growth factors, and their receptors (57, 60).Both membrane-bound and soluble members of the proteasesuperfamily can mediate this process (14). Yet, metallopro-teinases of the zinc-dependent ADAM family (for a disintegrinand metalloproteinase) were shown to be responsible for thecleavage of the majority of shed proteins (27, 57). The ADAMsare type I membrane-anchored glycoproteins which play im-portant roles in fertilization, neurogenesis, and angiogenesisand mediate the shedding of various membrane-bound mole-cules (14, 57). Among these family members, ADAM10/Kuz-banian and ADAM17/TACE (for tumor necrosis factor alpha[TNF-�]-converting enzyme) are particularly important in thecontext of ectodomain generation (57). They are involved inthe proteolysis of various substrates, including epidermalgrowth factor receptor ligands (14, 55), TNF-� and its recep-tors (12, 25), amyloid precursor protein (61), Notch (13),Notch ligand Delta (60), N-cadherin (54), fractalkine (38), andmany others.

In the present study, we demonstrate that a soluble variantof murine Axl RTK is constitutively generated from the full-length transmembrane molecule through defined proteolyticcleavage and identify ADAM10 as the major metalloprotein-ase responsible for this process. We provide several lines ofexperimental evidence which support a functional role for thissoluble molecule. Importantly, sAxl is constitutively present inmouse serum and cell-conditioned medium in combinationwith Gas6, which may affect Gas6 bioavailability and activity.Thus, our results link the proteolytic cleavage of Axl and Gas6function, suggesting that maintenance of a dynamic equilib-rium between serum levels of sAxl and Gas6 might play animportant regulatory role.

MATERIALS AND METHODS

Cytokines, antibodies (Abs), inhibitors, and recombinant proteins. Recombi-nant human interleukin-15 (IL-15), granulocyte-macrophage colony-stimulatingfactor (GM-CSF), and TNF-� were purchased from TEBU. Lipopolysaccharide(LPS) derived from Salmonella enterica serovar Friedenau was kindly providedby H. Brade (Research Center Borstel, Borstel, Germany). Phorbol 12-myristate13-acetate (PMA) and anti-�-actin Abs were purchased from Sigma-Aldrich.Abs against Axl (M-19), Gas6 (N-20), and PI3K (Z-8) were from Santa CruzBiotechnology; antiphosphotyrosine Abs (RC20) were from BD Biosciences, andanti-ADAM10 Abs were from Chemicon. Recombinant murine Gas6, rat anti-mouse Gas6, biotinylated goat anti-mouse Gas6, Axl, Mer, and Tyro3 (Dtk) Absand Axl-Fc and IL-3R–Fc chimeras were purchased from R&D Systems.

The broad-spectrum hydroxamic acid-based metalloproteinase inhibitorGM6001 and the respective negative control compound were purchased fromCalbiochem. GW280264X (a potent inhibitor of TACE and ADAM10 metallo-proteinases) was previously described (1, 38).

Cell culture, stimulation, and transfection conditions. Ras-Myc-retrovirus-immortalized TACE�/� (generously provided by S. Rose-John, Christian-Albre-chts-University, Kiel, Germany), simian virus large T antigen (SV-T)-immortal-ized ADAM10�/�, primary Axl�/� mouse embryo fibroblasts (MEFs), andrespective wild-type (WT) cells were generated and characterized as describedelsewhere (3, 36, 53). Axl�/� MEFs were immortalized by stable transfectionwith the simian virus 40 large T antigen expression vector pMSSVLT (58).Murine plasmacytoma J558L (19), murine mammary adenocarcinoma TS/A(17), murine fibrosarcoma L929 (ECACC 85011425), its TNF-�-resistant deriv-ative L929R, and primate fibroblast COS-7 (ATCC CRL-1651) cell lines weremaintained in RPMI 1640 medium, and MEFs were cultured in Dulbecco mod-ified Eagle medium. Culture medium was supplemented with 10% fetal calfserum (FCS), 2 mM L-glutamine, 100 U/ml penicillin, and 100 �g/ml streptomy-cin. Cells were stimulated with PMA (200 ng/ml) in FCS-free medium.

Bone marrow-derived dendritic cells (DCs) were generated as previously de-

scribed (15). Briefly, bone marrow cells were flushed from the femurs and tibiasof mice and erythrocytes were lysed. Cells were grown in RPMI 1640 mediumsupplemented with 10% FCS, 2 mM L-glutamine, 50 �M �-mercaptoethanol,antibiotics, and 20 ng/ml GM-CSF. On day 3, fresh medium with 20 ng/mlGM-CSF was added and on day 6, half of the cell-free culture supernatant wasreplaced with fresh medium containing 10 ng/ml GM-CSF. On day 8, cells wereharvested, washed twice, and incubated with LPS (10 ng/ml) for 24 h. Purity wasroutinely �95% as determined by fluorescence-activated cell sorter analysis.

ADAM10�/� MEFs were reconstituted with bovine hemagglutinin-taggedADAM10 cDNA in a retroviral expression vector based on pCXbsr (kindlyprovided by F. Fahrenholz, Institute of Biochemistry, Mainz, Germany) as pre-viously described (1, 38). COS-7 cells were transfected with Lipofectamine 2000(Gibco-Invitrogen) and analyzed 48 h posttransfection. Transfection efficiencywas confirmed by Western blotting with specific Abs.

The small interfering RNA (siRNA) method was used to knock downADAM10 expression. siRNA oligonucleotides targeting ADAM10 expressionwere purchased from Ambion, whereas transfection with scrambled oligonucle-otides was used as a control. Cells were transfected by jetSI-FluoF siRNAtransfection reagent (Eurogentec) with 50 nM siRNA. Forty-eight hours post-transfection, part of the cells was analyzed for specific depletion of ADAM10protein by Western blotting with anti-ADAM10 Abs. Transfection efficiency wasabout 80 to 90% as detected by fluorescence microscopy (data not shown).

Mice. Female mice of the inbred strains BALB/c, CH3, and C57BL/6 wereobtained from Charles River breeding laboratories. Axl�/�, Axl�/�, and Axl�/�

mice (41, 42) were bred in the Research Center Borstel under specific-pathogen-free conditions. Mice were bled from the tail vein, and sera were analyzed byenzyme-linked immunosorbent assay (ELISA).

Plasmid construction and deletion analysis. Axl (ARK) cDNA in the pRK5vector was kindly provided by P. Bellosta (European Institute of Oncology,Milan, Italy). Three Axl deletion mutants were created by removing partiallyoverlapping 14-aa sequences (438VSEPPPRAFSWPWW451, 442RAFSWPWWYVLLGA456, and 432QPLHHLVSEPPPRA446) in the membrane-proximal regionof murine Axl and designated Axl�1, Axl�2, and Axl�3, respectively. For thegeneration of the deletion mutants, an inverse PCR strategy was utilized. Pairs ofprimers (Axl�1 sense, 5-TATGTACTGCTGGGAGCACTTGTG-3; Axl�1 an-tisense, 5-CAGATGGTGGAGTGGCTGTCCTTG-3; Axl�2 sense, 5-AGGTGGGGGTTCACTCAC-3; Axl�2 antisense, 5-CTTGTGGCTGCCGCCTGCGTC-3; Axl�3 sense, 5-TCCTTGCCCTGGGCGCCAGGG-3; Axl�3 antisense,5-TTCTCGTGGCCTTGGTGGTAT-3) in the Axl coding sequence were de-signed in such a way that, after PCR amplification, the complete plasmid pRK5was obtained again, lacking only the bases located between the two primers. ThePCR products obtained were subsequently phosphorylated and ligated. Theidentities of the deletion-containing constructs were verified by standard DNAsequencing.

ELISA. Concentrations of Axl, Gas6, and Tyro3 in cell supernatants, lysates,and mouse sera were evaluated by DuoSet ELISA kits (R&D Systems) accordingto the manufacturer’s recommendations. Soluble Mer was detected with ratanti-mouse Mer and biotinylated goat anti-mouse Mer Abs (both from R&DSystems). In brief, a 96-well plate (Greiner) was coated and incubated overnightat 4°C with 1 �g/ml of rat anti-Mer Abs. Wells were blocked with 2% bovineserum albumin in PBS for 2 h. Samples (50 to 100 �l/well) were added to theplate, which was incubated overnight at 4°C. Serial dilutions of murine recom-binant Mer were used for standardization. Bound Mer was detected with bio-tinylated anti-Mer Abs at a concentration of 1 �g/ml, followed by incubation withstreptavidin-peroxidase. Axl/Gas6 heterocomplexes in mouse sera were detectedby a two-site ELISA. Plates were coated with monoclonal anti-Axl Abs, andAxl/Gas6 heterocomplexes in samples were detected with polyclonal biotinylatedgoat anti-mouse Gas6 Abs, followed by incubation with streptavidin-peroxidase.A chromogenic substrate (R&D Systems) was used for visualization, and thereaction was stopped after 20 min of incubation by addition of 1 N H2SO4.Optical density was determined at 450 nm with an ELISA reader (Dynatech).Anti-Axl or anti-Gas6 Abs did not show cross-reactivity for Gas6 or Axl, respec-tively.

Immunoprecipitation and Western blotting. NP-40 (0.5% final concentration)and a cocktail of protease inhibitors were added to cell supernatants or sera, andimmunoprecipitation with biotinylated anti-Axl Abs was performed for 2 h at4°C. Immunocomplexes were captured on streptavidin-agarose (Perbio Science).To analyze glycosylation, the samples after immunoprecipitation were treatedwith 250 mU of N-glycosidase F (Roche) for 3 h at 37°C according to themanufacturer’s instructions. Samples were resuspended in sodium dodecyl sul-fate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer (62.5 mMTris-HCl, pH 8.0, 1% glycerol, 2% SDS, 5% �-mercaptoethanol, 0.01% bromo-phenol blue), boiled for 5 min, and subjected to 10% SDS-PAGE. The resolved

VOL. 25, 2005 GENERATION OF SOLUBLE Axl BY ADAM10 9325

RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

proteins were transferred onto nitrocellulose (Bio-Rad). Blots were blocked for1 h in PBS containing 0.05% Tween 20 and 3% bovine serum albumin (Sigma-Aldrich). After incubations with primary and secondary Abs and washing withPBS–0.05% Tween 20, visualization of specific proteins was carried out by theenhanced-chemiluminescence method with enhanced-chemiluminescence West-ern blotting detection reagents (Amersham Pharmacia) according to the manu-facturer’s instructions. Cellular extracts were prepared and analyzed by Westernblotting as described elsewhere (19).

Reverse transcription (RT)-PCR. RNA was extracted from cells with TRIZOLreagent, and cDNA was synthesized from 5 �g of total RNA by using randomoligonucleotides as primers and a SuperScriptII kit (all reagents were fromGibco-Invitrogen). cDNA was amplified by a standard PCR procedure as de-scribed previously (19). The following primers were used: murine Axl (473 bp)sense, 5-CTCGGGGACCAGCCAGTGTTC-3; murine Axl antisense, 5-GGCCGGTCTCGAGGGTTCAGTT-3; �-actin (500 bp) sense, 5-GTGGGGCGCCCCAGGCACCA-3; �-actin antisense, 5-CTCCTTAATGTCACGCACGATTTC-3. All primers were purchased from Metabion. Amplification of the �-actinmessage was used to normalize the amount of cDNA used. A mock PCR(without cDNA) was included to exclude contamination in all experiments.

Flow cytometric analysis. Adherent cells were harvested from culture plateswith Accutase (PAA Laboratories). Axl, Mer, or Tyro3 expression was evaluatedby incubation of cells with biotinylated goat anti-mouse Axl, Mer, or Tyro3 Abs,respectively, followed by incubation with streptavidin-fluorescein isothiocyanate,and analyzed by flow cytometry with a FACScalibur (Becton Dickinson) andCellQuest software. Negative controls consisted of biotinylated, isotype-matchedAbs (BD PharMingen). The fluorescence signal of the labeled cells was calcu-lated as the median fluorescence intensity of the cell population.

Wound-healing assay. Plates were coated with 10 �g/ml of Axl-Fc or IL-3R–Fc(control) chimeric protein. Exponentially growing cells (2 106) were platedonto coated plates in complete growth medium. After 18 h, the monolayers ofcells were wounded by manual scratching with a pipette tip, washed with PBS,placed into complete growth medium, and photographed in phase-contrast witha Nikon microscope (Nikon Diaphot 300). Matched-pair marked wound regionswere photographed again after 6 and 18 h.

Data analysis. All experiments were performed in at least three independentassays. Data are summarized as means � standard deviations. Statistical analysisof the results was performed by Student’s t test for unpaired samples. A P valueof �0.05 was considered statistically significant.

RESULTS

Murine fibroblasts release sAxl. Murine L929 fibrosarcomacells abundantly express Axl mRNA and protein, as detectedby RT-PCR and Western blotting, whereas the amount of Axlis higher in a TNF-�-resistant derivative of the L929 cell linereferred to here as L929R (Fig. 1A and B). Membrane-boundAxl could also be detected on the cell surface of both fibroblastcell lines by flow cytometry (Fig. 1C) and confocal microscopy(data not shown). Therefore, we have chosen these two celllines as model systems for analysis of the mechanism(s) of sAxlgeneration and comparison. To investigate whether these cellsrelease sAxl into the culture medium, conditioned medium wascollected after 18 h of culture. Next, sAxl was immunoprecipi-tated from the conditioned medium with specific Abs andimmunoprecipitates were analyzed by Western blotting. Inparallel, the amount of Axl in the cell lysates was assessed.Indeed, L929 fibroblasts constitutively released sAxl into theculture medium, as shown by the ability of anti-Axl Abs toprecipitate a characteristic 65-kDa sAxl isoform (Fig. 1D),which is in accord with previous findings (23). Further, a full-length Axl protein (120 kDa) was detected in the cell lysates(Fig. 1D), whereas isotype-matched control Abs did not pre-cipitate Axl (data not shown). To assess whether sAxl is gly-cosylated, the immunoprecipitates from the conditioned me-dium were treated with N-glycosidase and the resultingproducts were analyzed by Western blotting. As shown in Fig.1E, the treatment of sAxl with N-glycosidase leads to a shift of

the specific band to a lower molecular mass position (about 60kDa), which presumably corresponds to a nonglycosylatedform of this molecule.

It has been reported that a variety of proteins may be re-leased from the cell surface after stimulation with PMA, apotent activator of protein kinase C, by triggering metallopro-teinase-dependent ectodomain sheddase machinery (6, 39).This pathway is defined as inducible shedding and appears tobe highly conserved among multiple cell types. Thus, we inves-tigated whether PMA treatment could enhance the release ofsAxl from murine L929 fibroblasts. To this end, cells weretreated with PMA (200 ng/ml) for different times or left un-treated and the supernatants were collected and tested byELISA. Time course experiments demonstrated that both cell

FIG. 1. Membrane-bound Axl and sAxl are present in L929 andL929R fibroblasts. (A) RT-PCR analysis of Axl expression in L929 andL929R cells. Amplification of the �-actin message was used to nor-malize the amount of cDNA. (B) L929 and L929R cells were lysed, andexpression of Axl was analyzed by Western blotting with specific Abs.Lysates from Axl�/� fibroblasts were used as a negative control. Ex-pression of �-actin was detected on the same membrane after strippingand served as a loading control. (C) Expression of Axl on the cellmembrane of L929 and L929R fibroblasts was assessed by flow cytom-etry analysis. (D) sAxl was detected in cell-conditioned medium (CM),and membrane-bound Axl was detected in lysates (L) from L929 andL929R cells by Western blotting after immunoprecipitation with spe-cific anti-Axl Abs. (E) Glycosylation pattern of sAxl. sAxl was immu-noprecipitated from cell-conditioned medium and treated with N-glycosidase or left untreated, as described in Materials and Methods.After treatment, protein lysates were analyzed by Western blottingwith anti-Axl Abs.

9326 BUDAGIAN ET AL. MOL. CELL. BIOL.

RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

lines release sAxl within the first 15 min, and PMA significantlyincreases the sAxl concentration in the supernatants (Fig. 2Aand B). The ability of PMA to enhance the release of sAxl wasmost prominent within 2 h of treatment (5.1- and 5.6-foldincreases in L929 and L929R cells, respectively). It should benoted that Axl expression was approximately 10-fold higher inL929R versus L929 cells at all indicated time points, resultingin the corresponding increase in the generation of sAxl (Fig.2A and B). Since the maximal difference between constitutiveand inducible sAxl generation in both L929 cell lines wasobserved after 2 h, most subsequent experimental measure-ments were performed at this time point. Notably, the appear-ance of sAxl was associated with reduced expression of thetransmembrane protein, as demonstrated by immunofluores-cent staining of membrane Axl and flow cytometry (Fig. 2C

and D). The rapid kinetics of this process suggests that thePMA-inducible release of sAxl likely occurs through enhancedprocessing of mature Axl rather than new biosynthesis of thisprotein and decreases when its cellular stores are exhausted.Interestingly, despite the fact that parental and resistant L929cells express two other receptors of this family, Tyro3 and Mer,at the mRNA and protein levels (see Fig. S1A in the supple-mental material), no soluble forms of these molecules could bedetected in the culture medium by respective ELISAs or im-munoprecipitation experiments (see Fig. S1B in the supple-mental material; also data not shown).

Reportedly, a number of ligands induce the release of theirown receptors into the culture medium. For example, suchligand-driven release of a soluble form of the receptor hasbeen demonstrated for IL-4 and TNF-� (21, 25). Thus, we next

FIG. 2. Regulation of Axl shedding. L929 (A) and L929R (B) cells were treated with 200 ng/ml of PMA for different times, and theconcentration of sAxl in cell-conditioned medium was evaluated by a specific ELISA. Untreated cells were used as controls. *, P � 0.05 versuscontrol samples. L929 (C) and L929R (D) cells were left untreated or treated with PMA for 2 h. Expression of membrane-bound Axl was evaluatedby flow cytometry. (E and F) Cells were treated with PMA (200 ng/ml), Gas6 (50 ng/ml), IL-15 (50 ng/ml), TNF-� (10 ng/ml), and LPS (10 ng/ml)for 2 h, and the concentration of sAxl in the culture medium was quantified by ELISA. Untreated (medium) cells were used as controls. **, P �0.01 versus control samples.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

wanted to test whether sAxl generation could be modulated bythe Axl ligand Gas6. In parallel, we also tested for the ability ofseveral other agents to induce sAxl release, including TNF-�,IL-15, and LPS. TNF-� can induce the production of solubleIL-15R� in both L929 and L929R cells, and this effect is notdependent on TNF-� cytotoxicity (18). Furthermore, IL-15 isable to upregulate Axl expression in L929 fibroblasts (V. Buda-gian et al., submitted for publication), whereas LPS treatmenttriggers the expression of chemokines by these cells (E. Bul-anova and S. Bulfone-Paus, unpublished data). Both fibroblastcell lines were treated with Gas6 (50 ng/ml), IL-15 (50 ng/ml),TNF-� (10 ng/ml), or LPS (50 ng/ml) for 2 h, and sAxl releasewas evaluated by ELISA. Interestingly, only TNF-� was able toinduce a weak increase in sAxl production in L929 cells,whereas Gas6, IL-15, and LPS had no effect (Fig. 2E). Con-versely, only Gas6 induced a modest increase in sAxl genera-tion in L929R cells (Fig. 2F), thereby indicating that sAxlrelease may be partially dependent on ligand binding, at leastin cells abundantly expressing Axl on the cell surface. Giventhat TNF-� was not able to enhance sAxl production in theresistant fibroblasts, the increase in sAxl generation in theTNF-�-treated parental cells presumably resulted from TNF-�-mediated cell death (20). Taken together, these experimentsdemonstrate that L929 and L929R fibroblasts constitutivelyrelease sAxl into the culture medium, and PMA stimulationfurther enhances this process, whereas Gas6 up-regulates sAxlgeneration only in fibroblasts highly expressing membrane-bound Axl.

Metalloproteinase inhibitors block constitutive and induc-ible Axl shedding. It has been reported that both human andmurine Axl proteins undergo proteolytic cleavage by an as yetunidentified protease(s) to yield soluble variants of this mole-cule (23, 50). Members of the metalloproteinase superfamilyhave been shown to be responsible for the cleavage of themajority of shed proteins (14). The proteolytic activity of me-talloproteinases can be blocked by broad-spectrum metallo-proteinase inhibitors such as batimastat or GM6001 (1). Totest whether the release of sAxl is mediated by metalloprotein-ases, cells were first incubated with GM6001 or its inactiveanalog as a control. These experiments demonstrated the abil-ity of GM6001 to suppress constitutive (2.8- and 3.5-fold, re-spectively) and PMA-induced (2.9- and 3.2-fold, respectively)Axl shedding in the L929 and L929R fibroblast cell lines, com-pared with a control compound (Fig. 3A). Conversely, theinhibitor increased the amount of Axl in cell lysates.

Given that sAxl generation is mediated by defined proteol-ysis, which is sensitive to GM6001, we wanted next to establishthe identity of the specific metalloproteinase(s) responsible forAxl cleavage. Taking into account that ADAM10 andADAM17/TACE have been implicated in the constitutive andPMA-inducible ectodomain shedding of a number of cell sur-face-expressed molecules (1, 14, 18, 54), we investigated theinvolvement of these two proteases in sAxl generation. For thispurpose, a hydroxamate-based metalloproteinase inhibitor,GW280264X, capable of inhibiting both ADAM10 and TACEwas used (1, 54). Cells were treated with 10 �M inhibitor for1 h, followed by PMA stimulation for another 2 h, and the cellsupernatants and cell lysates were tested by ELISA (Fig. 3B).The compound GW280264X significantly inhibited both con-stitutive and PMA-induced Axl shedding, as demonstrated by

a dramatic drop in the amount of sAxl in the cell-conditionedmedium from L929 cells (2.8- and 7.1-fold inhibition of con-stitutive and inducible shedding, respectively). As mentionedabove, the amount of Axl was significantly higher in L929Rcells, leading to the accordingly enhanced release of the solu-ble form. Notwithstanding, the dynamics of sAxl release wasrather similar in both cell lines, resulting in 4.4- and 7.2-foldinhibition of constitutive and PMA-induced Axl shedding, re-spectively, in the L929R cell line (Fig. 3B). However, smallamounts of sAxl were still detectable in the conditioned me-dium, indicating that the inhibitor could not completely abro-gate sAxl generation, and another releasing mechanism(s) mayalso be operative. At the same time, the action of this chemicalcompound resulted in an increase of Axl in cell lysates both inthe presence and in the absence of PMA (Fig. 3B). These datawere further confirmed by flow cytometry analysis, whichshowed an increase in the amount of Axl on the cell membranein the presence of GW280264X (Fig. 3C). The inhibitor had noeffect upon the total protein content in cell lysates or cellviability within 48 h, compared with cycloheximide (data notshown), indicating that it does not act through nonspecificinhibition of protein synthesis. Taken together, these experi-ments show the ability of GW280264X to inhibit constitutiveand PMA-inducible sAxl release, thereby suggesting that oneor both metalloproteinases of the ADAM family are essentialfor sAxl generation.

Cleavage of Axl is reduced in murine fibroblasts lackingADAM10 but is not affected in TACE-deficient cells. To gainfurther insights into the mechanism of constitutive and induc-ible sAxl release, MEFs generated from mouse embryos with atargeted deletion of ADAM10 or TACE were used (36, 54).The endogenous expression of Axl was rather similar inADAM10�/�, TACE�/�, and respective WT MEFs, as de-tected by RT-PCR and Western blotting (Fig. 4A and B). Theabsence of ADAM10 or TACE in corresponding cells wasalready confirmed by Western blotting with specific anti-ADAM10 or anti-TACE Abs (18). The cells were plated inequal numbers and stimulated with PMA or left untreated, andthe concentration of sAxl in the cell-conditioned medium wasassessed by ELISA after 4 h, because PMA-induced release ofsAxl from these fibroblasts was maximal at this time point(data not shown). These experiments demonstrated that bothconstitutive and PMA-inducible release of sAxl was dramati-cally lower in ADAM10�/� MEFs compared with their normalWT counterparts (Fig. 4C). Notwithstanding, Axl shedding wasnot completely abrogated in these cells, as ADAM10�/� MEFsconstitutively released minute amounts of sAxl, whereas PMAtreatment had little, if any, effect on this process (Fig. 4C).Constitutive and PMA-inducible Axl cleavage was diminishedapproximately 2.5- and 12.6-fold in ADAM10�/� fibroblasts,respectively, compared with cells from ADAM10�/� mice. Asexpected, the reconstitution of ADAM10 into ADAM10�/�

MEFs (ADAM10�/�Re) completely restored the constitutiveand inducible sAxl generation. Since Axl was not proteolyti-cally removed from the cell membrane in ADAM10-deficientfibroblasts, these cells exhibited increased expression of full-length Axl at the cell surface, as determined by staining withanti-Axl Abs and flow cytometry analysis (Fig. 4D). The re-quirement of ADAM10 for sAxl generation was further con-firmed in at least two independently derived ADAM10-defi-


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

cient MEFs (data not shown). No significant changes in sAxlproduction were observed in TACE�/� versus TACE�/�

MEFs. It should, however, be noted that the amount of con-stitutively released sAxl was slightly higher in fibroblasts fromTACE�/� and TACE�/� mice compared with ADAM10�/�

and ADAM10�/� MEFs, while PMA-inducible cleavage wassomewhat reduced (Fig. 4C). This fact may indicate that theabsence of TACE attenuates PMA-inducible shedding of Axlto a certain extent but cannot considerably affect this process.

To address the role of ADAM10 in more detail and toconfirm the aforementioned data, we performed siRNA-me-diated suppression of ADAM10 in both L929 cell lines. Cellswere transfected with siRNA oligonucleotides targetingADAM10 expression, and ADAM10 cellular content was as-sessed by RT-PCR and Western blotting. Both methodsshowed a decrease in the cellular amount of ADAM10 at themRNA (data not shown) and protein (Fig. 4E) levels, com-pared with cells transfected with scrambled control oligonucle-

FIG. 3. Effects of metalloproteinase inhibitors on Axl cleavage. L929 or L929R cells were incubated with 2.5 �M GM6001 or with its inactiveanalog as a control (A) or with 10 �M hydroxamate inhibitor GW280264X (B) for 1 h prior to treatment with 200 ng/ml of PMA for another 2 h.Untreated cells were used as controls. Cell-conditioned medium (CM) or cell lysates (L) were harvested and analyzed by ELISA for the presenceof sAxl or membrane-bound Axl, respectively. (C) Cells were treated with the above inhibitors, followed by PMA stimulation for another 2 h.Expression of membrane-bound Axl was evaluated by flow cytometry analysis. * and **, P � 0.01 versus control samples; neg., negative.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

otides. Next, the release of sAxl was tested by ELISA, whichdemonstrated that ADAM10 knockdown resulted in a signifi-cant drop in constitutive and inducible sAxl generation in bothfibroblast cell lines (Fig. 4F). Taken together, these resultsshow that ADAM10 plays a major role in constitutive andinducible sAxl release, although another protease-dependentor -independent mechanism(s) may also be operative.

The membrane-proximal sequence of Axl is important forshedding. To identify a potential ADAM10 cleavage site, weperformed a deletion analysis whereby three partially overlap-ping 14-aa stretches in the membrane-proximal region of Axlwere removed (Fig. 5A). Given that cleavage of human Axl was

mapped to the 14-aa stretch in the extracellular region (50),the first deletion mutant (Axl�1) corresponded to the homol-ogous sequence in mouse Axl. The resulting constructs (Axl�1,Axl�2, and Axl�3) were transiently transfected into COS-7cells, which have relatively high endogenous ADAM10 activity(38), and sAxl generation was assessed 48 h after transfectionby ELISA. Transfection with a full-length Axl construct (AxlWT) served as a control. Expression of control and deletionconstructs (Axl WT, Axl�1, Axl�2, and Axl�3) was compara-ble, as detected by Western blotting (Fig. 5B). As demon-strated in Fig. 5C, transfection with Axl�1 decreased constitutiveand PMA-inducible sAxl release about twofold. Spontaneous

FIG. 4. Analysis of sAxl release in ADAM10- or TACE-deficient MEFs. Expression of Axl in different MEFs was assessed by RT-PCR (A) andWestern blotting (B). Amplification of the �-actin message was used to normalize the amount of cDNA to be used. Detection of �-actin proteinwas used to prove equal loading of cell lysates. (C) MEFs were treated with 200 ng/ml of PMA for 4 h, and the sAxl concentration in the culturemedium was detected by ELISA. Untreated cells were used as a control. **, P � 0.05 versus ADAM10�/� cells; ***, P � 0.01 versus untreatedsamples. (D) Expression of membrane-bound Axl on ADAM10�/� and ADAM10�/� MEFs was evaluated by flow cytometry. (E and F) L929 andL929R cells were transfected with ADAM10 or scrambled (control) siRNA oligonucleotides, and expression of ADAM10 was assessed by Westernblotting; detection of �-actin is shown as a loading control (E). (F) Forty-eight hours after transfection, cells were treated with PMA for 2 h orleft untreated. Next, the concentration of sAxl in cell-conditioned medium was evaluated by ELISA. *, P � 0.05; **, P � 0.01 (versus cellstransfected with scrambled control oligonucleotides). The data represent at least three separate experiments with comparable results.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

and inducible generation of sAxl was considerably lower incells transfected with the Axl�2 (4.9- and 5.1-fold, respec-tively) and Axl�3 (7.1- and 8.2- fold, respectively) mutants.Taken together, these results indicate that deletions in themembrane-proximal region of Axl dramatically affect ADAM10-mediated sAxl generation.

DCs and tumor cells express membrane-bound Axl and re-lease sAxl into the culture medium. Normal and transformedmurine cell lines constitutively express Axl mRNA in vitro andin vivo. Given the presence of sAxl in biological fluids (23), wewished to determine whether mouse primary and transformedcells could release this molecule. To clarify this issue, we gen-erated DCs essentially as described previously (15). As illus-trated in Fig. 6A, flow cytometry analysis confirmed that DCs

from WT mice have ample expression of Axl on the cell mem-brane, whereas no membrane-associated Axl was detected incells from Axl�/� animals. Further, WT DCs also express Ty-ro3 and Mer, although the expression of Tyro3 on the cellmembrane is significantly lower (Fig. 6A). Next, WT DCs werestimulated with PMA for 4 h or left untreated and cell-condi-tioned medium was harvested for ELISA to assess the produc-tion of sAxl, Tyro3, and Mer. These experiments showed thatWT DCs constitutively released considerable amounts of sAxl( 2 ng/ml), even in the absence of PMA stimulation (Fig. 6B).Conversely, we were not able to detect Tyro3 and Mer in thesupernatants from these cells (Fig. 6B). As expected, treatmentwith PMA further elevated sAxl production. Concomitantly,PMA reduced the amount of Axl in the cell lysates. No detect-able levels of sAxl or membrane-bound full-length Axl werefound in the supernatants or cell lysates, respectively, of DCsfrom Axl�/� mice, which served as a control (Fig. 6B). Re-markably, the inhibitor GW280264X significantly reduced boththe constitutive and PMA-inducible cleavage of Axl in WTDCs (Fig. 6B).

Both transmembrane and soluble forms of Axl have beendetected by immunoprecipitation in cell lysates of tumorsgrowing in syngeneic animals, and a large amount of a char-acteristic 65-kDa form of sAxl was also found in tumor exu-dates (23).

Thus, the constitutive and inducible release of sAxl wasassessed in the cell-conditioned medium from two murine tu-mor cell lines, TS/A and J558L. To this end, both tumor celllines were stimulated with PMA for 4 h or left untreated andthe presence of sAxl in the cell supernatants was evaluated byELISA. Figure 6C shows that TS/A cells constitutively pro-duced large amounts of sAxl (0.8 to 1 ng/ml) in the culturemedium, whereas PMA enhanced the release of this molecule 2.5-fold. The PMA-inducible sAxl release was associatedwith a reduction in the amount of cell-associated protein (datanot shown). Conversely, sAxl production was considerablylower in J558L cells at the indicated time point and PMAtreatment had little effect on this process (Fig. 6C), which isexplained by the fact that these cells have low endogenous Axlexpression, as detected by semiquantitative RT-PCR and flowcytometry analysis (see Fig. S2 in the supplemental material).Incubation of cells with the metalloproteinase inhibitorGW280264X significantly reduced both constitutive and PMA-induced Axl shedding in TS/A cells (Fig. 6C). This effect wasalmost undetectable in J558L cells due to the low level of sAxlgeneration in these cells. Taken together, these results dem-onstrate that sAxl is released by select murine primary andtransformed cells.

sAxl is present in mouse serum. It has been reported thatsAxl could be detected by immunoprecipitation with specificAbs in mouse serum, plasma, and various mouse organs, suchas the brain, cerebellum, liver, and spleen (23). To confirmthese results and to analyze quantitatively the amount of sAxlin mouse serum, the sera from at least five animals per strainwere tested by ELISA. As shown in Fig. 7A, CH3, C57BL/6,and BALB/c mice display similar levels of circulating sAxl(range of 5 to 6 ng/ml). Next, the sAxl concentrations wereassessed in sera from Axl�/�, Axl�/�, and WT mice (Axl�/�).The amount of sAxl in sera from Axl�/� mice was comparableto that found in other strains, whereas animals heterozygous

FIG. 5. The membrane-proximal region of Axl is important forsAxl generation. (A) Axl deletion mutants were generated as describedin Materials and Methods. COS-7 cells were transiently transfectedwith full-length Axl (Axl WT) or with Axl deletion mutants (Axl�1,Axl�2, and Axl�3) lacking different 14-aa sequence in the membrane-proximal region of Axl. FNIII, fibronectin type III; TM, transmem-brane. (B) Efficiency of transfection was proved by Western blottingwith anti-Axl Abs, and detection of �-actin on the same blots is shownas a loading control. (C) After 48 h, the culture medium was replacedwith fresh culture medium and the cells were incubated in the presenceor absence of PMA for 1 h. Conditioned medium was collected, andthe concentration of sAxl was determined by ELISA. *, P � 0.05; **,P � 0.01 (versus Axl WT-transfected cells). The results are represen-tative of three independent experiments.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

for Axl (Axl�/�) had an approximately twofold lower concen-tration of sAxl (Fig. 7A). As expected, no sAxl was detectablein sera from Axl�/� mice. It should be noted that soluble formsof Tyro3 and Mer were not present in mouse serum, as de-tected by specific ELISAs (Fig. S1B in the supplemental ma-terial).

sAxl associates with Gas6 in mouse serum. The ability of therecombinant Axl ectodomain to bind exogenous Gas6, therebypreventing Gas6-specific interactions with the membrane-bound full-length receptor and Axl-mediated downstream sig-naling, has already been documented (23). We were, however,not able to detect Gas6 in mouse serum by Gas6-specificELISA with anti-Gas6 monoclonal and anti-Gas6 polyclonalAbs as coating and detection Abs, respectively (data notshown). Given that sAxl is present in a relatively high concen-tration in mouse serum, we asked whether some amount ofendogenous sAxl could exist in association with Gas6, thusaffecting the half-life and biological properties of the ligand

and causing difficulties in Gas6 detection. To test this sugges-tion, mouse serum samples were assessed by a two-site ELISAfor sAxl-Gas6 interactions. In a direct assay, 96-well plateswere covered with monoclonal coating Abs which recognizethe extracellular domain of Axl. After incubation with serumsamples, polyclonal detection Abs targeted against Gas6 wereadded. In an inverted setup, monoclonal anti-Gas6 Abs servedas coating Abs and polyclonal anti-Axl Abs served as detectionAbs. The direct two-site ELISA showed that a significantamount of endogenous Gas6 is constitutively associated withsAxl in mouse serum, whereas the inverted ELISA did notgenerate any signal (data not shown), presumably due to theinability of anti-Gas6 monoclonal Abs to bind Gas6. As dem-onstrated in Fig. 7B, circulating levels of Gas6 in mouse serumlay in the range of about 3 ng/ml. Given that monoclonalcoating Abs designed against the Axl ectodomain capture sAxlin mouse serum, whereas monoclonal coating Abs againstGas6 do not bind Gas6, it is likely that the Ab-binding site on

FIG. 6. Detection of sAxl, Mer, and Tyro3 in culture medium from DCs and tumor cell lines. (A) DCs were generated as described in Materialsand Methods, and expression of Axl, Mer, and Tyro3 on the cell membrane of WT and Axl�/� DCs was evaluated by flow cytometry analysis.(B) DCs were pretreated with GW280264X for 30 min prior to stimulation with PMA for another 4 h or left untreated. Unstimulated and Axl�/�

DCs served as control cells. The concentration of sAxl or cell-associated Axl in cell-conditioned medium (CM) or cell lysates (L), respectively, andthe concentrations of Mer and Tyro3 in cell-conditioned medium were analyzed by ELISA. (C) TS/A and J558 cells were incubated in the presenceor absence of PMA for 4 h. Culture medium was collected, and the concentration of sAxl was evaluated by ELISA. *, P � 0.05 versus untreatedcells; **, P � 0.05 versus control samples.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Gas6 is masked by sAxl; therefore, all Gas6 molecules inmouse serum are presumably bound to sAxl. This was furtherconfirmed by the inability of a specific ELISA to detect Gas6 inserum from Axl�/� mice. Conversely, the serum levels of Gas6appear to correlate with the sAxl concentration (about 5 ng/ml) (Fig. 7B). The ability of sAxl to bind Gas6 was also con-firmed by coimmunoprecipitation experiments and Westernblotting (Fig. 7C). Remarkably, Gas6 and Axl were undetect-able by respective specific ELISAs in serum from Axl�/� ani-mals (Fig. 7B). These results were further corroborated by

Western blotting analysis after immunoprecipitation from se-rum with specific anti-Axl or anti-Gas6 Abs, respectively (datanot shown). Given the absence of Gas6 in serum from Axl�/�

animals, as detected by specific and two-site ELISAs and West-ern blotting analysis, we wished to determine whether Gas6requires the expression of Axl to be secreted. To this end, weutilized COS-7 cells, which do not express endogenous Gas6and Axl. These cells were cotransfected with constructs codingfor Gas6 and Axl, whereas transfection of Gas6 with the plate-let-derived growth factor receptor served as a negative control.Remarkably, these experiments demonstrated that cotransfec-tion of Gas6 with Axl dramatically enhanced Gas6 productionin COS-7 cells ( 8 ng/ml) (see Fig. S3 in the supplementalmaterial). Interestingly, Gas6 was detected predominantly inassociation with sAxl by two-site ELISA (but not by an ELISAspecific for Gas6) and correlated with the sAxl concentration( 7 ng/ml). These results indicate that expression of Axl isrequired for the secretion of Gas6 and might play an importantrole in cellular mechanisms regulating Gas6 production.

Furthermore, two other members of this family, Tyro3 andMer, were not upregulated in DCs and MEFs from Axl�/�

animals, as assessed by RT-PCR and Western blotting (datanot shown) or flow cytometry analysis (Fig. 6A and data notshown). No associations between Gas6 and Tyro3 or Mer wereobserved in respective two-site ELISAs (data not shown),which was not surprising given the lack of soluble forms ofthese receptors in mouse serum and cell-conditioned medium.This fact strongly argues that the function of the members ofthis RTK family is not redundant.

Axl/Gas6 complexes were also detected in the conditionedmedium from L929R cells (Fig. 7B). However, we failed todetect these complexes in conditioned medium from L929 andTS/A cells, as well as DCs and MEFs from WT and Axl�/�

mice, which might be explained by the fact that these cells donot express detectable Gas6 at the mRNA and protein levels(data not shown) and thus cannot serve as a source of thisprotein. Taken together, these experiments show that sAxl canform heterocomplexes with Gas6 in mouse serum, which mayhave particular relevance for the biology of Gas6 and theability of this ligand to mediate Gas6-specific responses.

Immobilized Axl-Fc chimeric protein promotes cell migra-tion and induces phosphorylation of Axl and PI3K. The extra-cellular portion of Axl contains features which are reminiscentof cell adhesion molecules (67). It has been shown that Axlmediates cell aggregation and adhesion by homotypic-ho-mophilic association of the extracellular domains of this mol-ecule (9). Given that the extracellular domain of Axl can func-tion as a specific substrate for adhesion of Axl-expressing cells(9), we next assessed in a wound-healing scratch assay whetherthe chimeric molecule Axl-Fc could affect the migratory prop-erties of murine fibroblasts. Axl-Fc comprises the extracellulardomain of mouse Axl (26) fused to the carboxy-terminal Fcregion of human immunoglobulin G1 via a polypeptide linker.It seemed to us theoretically reasonable that the presence ofthe extracellular portion of Axl capable of homophilic associ-ations might render this protein capable of mimicking effects ofendogenous sAxl, thereby providing a tool to assess its func-tion. The plates were coated with Axl-Fc or IL-3R–Fc andseeded with equal numbers of growing cells. After 18 h ofculture, the cells were cleared within a defined area by scratch-

FIG. 7. sAxl is present in mouse serum and associates with Gas6.(A) Sera from C57BL/6, CH3, BALB/c, Axl�/�, Axl�/�, and Axl�/�

mice were tested for the presence of sAxl by ELISA. Each bar repre-sents the mean of at least five animals per strain (*, P � 0.05; **, P �0.01). (B) The concentration of Gas6 in mouse sera was determined bya two-site ELISA. The concentration of sAxl or Gas6 was determinedby a specific ELISA for Axl or Gas6, respectively, and is shown forcomparison. (C) Axl was precipitated from mouse serum with specificAbs. Immunocomplexes were subjected to SDS-PAGE and transferredonto nitrocellulose membrane, and Gas6 was detected by probing themembrane with specific Abs. Detection of Axl on the same blots servedas a loading control. IgG, immunoglobulin G.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

ing with a pipette tip, washed, placed into complete growthmedium, photographed in phase-contrast (Fig. 8A, 0 h), andallowed to migrate into the cleared area, whereas cells culturedin the wells coated with IL-3R–Fc were used as controls. Infact, L929R fibroblasts growing in the Axl-Fc-coated wells hadseparated from the monolayer at the wound edges, formingvisible cell aggregates, and significant numbers of these cellsstarted to migrate into the cleared area within 6 to 18 h,compared with cells in untreated wells (data not shown) orcontrol wells coated with IL-3R–Fc (Fig. 8). A similar effectwas observed in MEFs from WT mice (Fig. 9A, right side), andin parental L929 cells, but the migratory properties of parental

cells were somewhat diminished (data not shown). It should bementioned that Axl-Fc had no effect on the growth and migra-tory characteristics of murine fibroblasts when added in a sol-uble form, and additional cross-linking with anti-Axl Abs alsohad no effect (data not shown). Thus, these results show thatonly immobilization of Axl-Fc on plastic renders this moleculebiologically active to efficiently promote cell migration.

Reportedly, Axl-mediated cell aggregation results in a de-gree of receptor activation (23). Given the observed ability ofimmobilized Axl-Fc to mediate cell migration and the ability ofthe extracellular domain of this RTK to associate throughhomotypic interactions, we wished to determine whether thischimeric molecule could bind cell-associated full-length Axland induce Axl phosphorylation and subsequent Axl-mediatedactivation of downstream targets, such as PI3K (32). To testthis hypothesis, we assessed the phosphorylation state of trans-membrane Axl and PI3K. Cells were incubated for differenttimes in plates coated with Axl-Fc, collected, and lysed, andAxl and PI3K were immunoprecipitated from cell lysates withspecific Abs. Then, the immunoprecipitates were subjected toSDS-PAGE and resolved proteins were transferred onto nitro-cellulose membrane and probed with anti-pTyr Abs. Theseexperiments revealed that incubation in Axl-Fc-coated wellsfor 15 min was sufficient to induce the phosphorylation of Axland PI3K to levels comparable to those observed upon Gas6stimulation (Fig. 8B). No changes in the phosphorylation pat-tern of these proteins were detected in cells incubated in con-trol wells left uncoated or coated with IL-3R–Fc. Taken to-gether, these results demonstrate that, upon immobilization,Axl-Fc chimeric protein induces the phosphorylation of cell-bound Axl and its downstream target PI3K.

Remarkably, the migratory properties of MEFs from Axl�/�

mice were dramatically reduced compared with those of con-trol WT cells (Fig. 9A). This was already detectable after 6 h ofincubation, whereas at 18 h total wound closure was observedin Axl-Fc-coated wells seeded with MEFs from WT animals. Incontrast, the migratory characteristics of Axl�/� fibroblastswere almost undistinguishable in Axl-Fc-coated versus IL-3R–Fc-coated (control) wells, in which no wound closure was ob-served after 18 h of incubation. Moreover, the phosphorylationof transmembrane Axl and PI3K was not observed in MEFsfrom Axl�/� mice (Fig. 9B), thereby providing strong supportfor the idea that the phosphorylation of these molecules de-pends on, and results from, the direct binding of immobilizedAxl-Fc to membrane-bound Axl. Notably, the addition ofsAxl-Fc in different concentrations (0.01 to 10 �g/ml) did notaffect the ability of immobilized Axl-Fc to induce cell migrationand protein phosphorylation in L929R cells and MEFs fromWT mice (see Fig. S4 in the supplemental material; also datanot shown).

Because L929R cells express and release Gas6, whereassAxl-Fc binds Gas6 and neutralizes its activity (23, 69, 70),these experiments also argue against the possibility that en-dogenous Gas6 participates in the observed cell migration andphosphorylation events by forming a complex with immobi-lized Axl-Fc. In addition, immobilized Axl-Fc preserved theability to mediate cell migration and the phosphorylation ofmembrane Axl and PI3K in cells treated with warfarin, whichselectively inhibits posttranslational vitamin K-dependent�-carboxylation of Gas6, which is essential for its receptor-

FIG. 8. Immobilized Axl-Fc chimeric protein promotes cell migra-tion and induces the phosphorylation of Axl and PI3K. (A) L929R cellswere plated onto Axl-Fc- or IL-3R–Fc (control)-coated six-well platesand allowed to grow in the presence of 10% FCS. After 18 h, a woundwas created by scratching with a pipette tip (0 h). The cells werewashed with PBS and incubated further to allow migration into thewounded area. Phase-contrast images of matched pairs of markedwound regions were taken 6 and 18 h later to assess cell migration.(B) L929R cells were serum starved for 4 h and incubated in Axl-Fc-coated wells for 15 min. Incubation in IL-3R–Fc (control)-coated wellsor stimulation with 100 ng/ml of Gas6 was used as a negative orpositive control, respectively. Cells were lysed, and Axl or PI3K wasimmunoprecipitated from the lysates with specific Abs. Precipitateswere subjected to 10% SDS-PAGE and analyzed with anti-pTyr Abs.Detection of Axl or PI3K on the same blots was used as a loadingcontrol. The picture is representative of three independent experi-ments, all of which yielded highly comparable results. IP, immunopre-cipitation; WB, Western blotting.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

binding and growth-potentiating activities (68–70) (see Fig. S4in the supplemental material). These results indicate that en-dogenous Gas6 is presumably not essential for signaling- andmigration-inducing effects of immobilized Axl-Fc, at least un-der the experimental conditions used in this study.

DISCUSSION

In this study, we showed that a soluble form of Axl RTK isgenerated through proteolytic cleavage by a disintegrin-likemetalloproteinase, ADAM10. In addition, experimental evi-dence is provided that supports a role for immobilized sAxl inpromoting cell migration and activation of membrane-boundfull-length Axl and its downstream target, PI3K. Apart fromthe identification of ADAM10 as a major proteinase respon-sible for Axl ectodomain shedding, we also demonstrated thatsAxl associates with endogenous Gas6 in mouse serum. Adynamic equilibrium between sAxl and Gas6 levels in biolog-

ical fluids might play an important regulatory role and dramat-ically affect Gas6 function.

Murine primary and transformed cells spontaneously releasesAxl into the culture medium, whereas PMA stimulation fur-ther enhances this process. Interestingly, murine DCs produceconsiderable amounts of sAxl and can serve as a source of thissoluble protein. Further experiments are required to under-stand what role Axl and its soluble form could play in thecomplex biology of these antigen-presenting cells. Despite thefact that Axl cleavage was inhibited by the broad-spectrummetalloproteinase inhibitor GM6001 and an inhibitor of bothADAM10 and TACE, GW280264X, decisive evidence for themajor role of ADAM10 was provided by experiments withfibroblasts derived from mice with a homozygous disruption ofthe ADAM10-encoding gene. Both spontaneous and induciblesAxl release was dramatically reduced in these cells comparedwith shedding by corresponding WT fibroblasts, whereas re-constitution of active ADAM10 expression restored Axl cleav-

FIG. 9. Immobilized Axl-Fc chimeric protein induces cell migration and phosphorylation of Axl and PI3K in WT but not Axl�/� MEFs.(A) MEFs were plated onto Axl-Fc- or IL-3R–Fc (control)-coated six-well plates. A confluent cell monolayer was wounded by scratching with apipette tip (0 h). The cells were washed with PBS and incubated further to allow migration into the wounded area. Phase-contrast images ofmatched pairs of marked wound regions were taken 6 and 18 h later to assess cell migration. (B) MEFs were serum starved for 4 h and incubatedin Axl-Fc-coated wells for 15 min. Incubation in IL-3R–Fc-coated (control) wells or stimulation with 100 ng/ml of Gas6 was used as a negative orpositive control, respectively. Cells were lysed, and Axl or PI3K was immunoprecipitated from the lysates with specific Abs. Precipitates weresubjected to 10% SDS-PAGE and analyzed with anti-pTyr Abs. Detection of Axl or PI3K on the same blots was used as a loading control. IP,immunoprecipitation; WB, Western blotting.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

age. Conversely, Tyro3 and Mer were not detectable in mouseserum and cell-conditioned medium, either alone or in associ-ation with Gas6, indicating a unique role for Axl among thefamily members.

For the members of the ADAM family, it has been sug-gested that structural and kinetic characteristics rather thanminimal consensus shedding sequences and the identity of theamino acids constituting them regulate the access of the pro-tease to the cleavage site (63, 66). This underlines the com-plexity of ADAM-substrate interactions and indicates thatthere may not exist a minimal consensus shedding sequencewithin the cleavage site but rather additional structural re-quirements distal from the cleavage site. The most consistentfeature indicative of cleavage sites among ADAM substrates isthat they usually reside in a stalk region between the mem-brane and an initial globular extracellular subdomain (57). Thefact that sAxl generation was dramatically lower in cells trans-fected with Axl�2 and Axl�3 deletion mutants, compared withAxl�1, strongly supports the idea that structural requirementsplay a major role in regulating the access of ADAM10 to thecleavage site of Axl.

Axl RTK is thus a novel substrate for ADAM10, althoughthe fact that some amount of sAxl is still present in the super-natants of ADAM10�/� cells argues that ADAM10-mediatedproteolysis is not the only mechanism involved in sAxl gener-ation. Notably, ADAM10 activity has been implicated in theconstitutive shedding of a number of substrates, whereas in-ducible shedding is mainly attributed to TACE (14, 55). Incontrast, ADAM10-deficient fibroblasts exhibited a dramaticdefect in both constitutive and inducible proteolytic cleavageof Axl. Currently, we perform experiments to identify whetherother members of the ADAM family may play an auxiliary rolein the constitutive and/or inducible shedding of this RTK.However, a minor involvement of other protease-independentmechanisms cannot be excluded.

Ectodomain cleavage and shedding of N-glycosylated sAxlmight play a particular role in distinct physiologic and/orpathological conditions and alter, in the first place, Gas6 actionon the cells expressing full-length membrane Axl, as well asTyro3 and Mer, by competing for Gas6 with the cellular re-ceptors and inhibiting its activity. Conversely, sAxl may in-crease the bioavailability of Gas6 by prolonging its half-life andslowing ligand release from the heterocomplex to providephysiologically relevant levels of Gas6 in tissues, thereby re-sulting in local or systemic effects of this protein and influenc-ing the nature and/or duration of the signaling event. It is alsotheoretically conceivable that the sAxl/Gas6 heterocomplexcould bind and activate the membrane-tethered receptor. Sup-port for this suggestion is provided by the observation thatGas6 can mediate cell aggregation by a heterotypic intercellu-lar mechanism, whereby cell-bound Gas6 interacts with Axl onan adjacent cell (45). This mode of action is in agreement withthe reported ability of soluble IL-6R to potentiate IL-6 signal-ing by binding to cell membrane-anchored gp130 (40). In ad-dition, the sAxl/Gas6 complex may also be capable of interact-ing with transmembrane Axl through homotypic associationsbetween Axl extracellular domains. Although initial experi-ments do not support a role for endogenous Gas6 in the ob-served ability of immobilized Axl-Fc to induce the phosphor-ylation of transmembrane Axl and PI3K and cell migration,

additional studies are required to address this question in moredetail.

Despite the fact that the exogenously added Axl ectodomaininhibits the action of Gas6 (23), it remains to be elucidatedhow relatively high or low circulating levels of endogenous sAxlmay affect the function of Gas6. Because sAxl prevented rec-ognition of Gas6 by monoclonal anti-Gas6 Abs in an invertedELISA, it appears likely that sAxl is present in excess overGas6 and all Gas6 molecules in mouse serum are presumablybound to sAxl. This may indicate that either these two mole-cules exist in a dynamic equilibrium or some proportion of sAxlis not associated with Gas6. Given that sAxl may also hide acertain Gas6 antigenic epitope(s) from polyclonal detectionAbs, the results from the direct two-site ELISA should beinterpreted with caution in regard to the actual endogenousGas6 concentration in mouse serum.

Exogenously administered soluble IL-4R can either poten-tiate or inhibit the activity of IL-4 in vivo, depending on therelative molar ratios of sIL-4R to IL-4; low ratios promoted thecytokine activity, while a large excess of sIL-4 over IL-4 inhib-ited responses to IL-4 (56). Similarly, low levels of soluble TNFreceptor presumably enhance TNF signaling, perhaps by sta-bilizing the ligand, whereas higher concentrations inhibit TNFactivity (37). In agreement with this model, high sAxl concen-trations may prevent Gas6 interaction with the membrane-bound receptors and inhibit Gas6-specific signaling, whereaslow sAxl levels could potentiate Gas6 activity, but further stud-ies are required to confirm or reject this suggestion.

Extracellular domain architectures of Axl family membersresemble adhesion molecules of the cadherin and immuno-globulin superfamily and receptor-type protein tyrosine phos-phatases, which were shown to promote cell adhesion througha homophilic mechanism (9). Homotypic associations involvingthe extracellular domains of murine Axl were also shown tomediate cell aggregation and enhance receptor phosphoryla-tion, implicating such a mechanism in the activation of thisgroup of RTKs and suggesting an important role in signalingcell adhesion (9). The release of the cadherin ectodomaincontaining homophilic binding sites is functionally of majorimportance for the regulation of cell adhesion and cell migra-tion (52, 54). Thus, the ability of immobilized Axl-Fc chimericprotein to enhance cell migration and to induce, presumablythrough homophilic interactions, the phosphorylation of trans-membrane Axl and PI3K could mimic similar effects of endog-enous sAxl. Transmembrane Axl may function as a substrateadhesion molecule, mediating cell binding to proteolyticallyprocessed, immobilized Axl as a component of the extracellu-lar matrix.

Remarkably, the adhesion between ADAM10�/� fibroblastsis reportedly much stronger than that of their WT counterpartsdue to N-cadherin-mediated cell-cell adhesion (54). Given thatboth N-cadherin and Axl are substrates for ADAM10, theincrease in the adhesive properties of these fibroblasts couldalso be mediated by detainment of membrane Axl. Further-more, ADAM10-dependent cleavage of N-cadherin regulates�-catenin nuclear signaling (54). Gas6 and Axl were alsoshown, through stabilization of the cytoplasmic �-catenin pool,to have a possible role in mammary development and/or neo-plastic transformation (33). Thus, Axl and N-cadherin mayshare not only structural but also functional similarities.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Contrary to the role of protein S as a natural anticoagulationfactor, Gas6 stimulates platelets, and Gas6�/� mice have aplatelet aggregation defect that confers protection againstthrombosis (4, 5). Gas6 and its receptors (Axl, Tyro3, and Mer)play an important role during thrombus formation through theability to amplify signaling via integrin �IIb�3 (5). Gas6 stim-ulates growth of cardiac fibroblasts and induces Axl-mediatedchemotaxis of vascular smooth muscle cells (30, 62). Upregu-lated expression of Gas6 and Axl parallels neointima forma-tion after balloon injury of the rat carotid artery (46). Thesefacts support the important role of the Gas6/Axl system as anovel regulator of vascular cell function that may have animpact on the response of blood vessels to injury, therebycontributing to the development of atherosclerosis and/or re-stenosis (46).

Given that recombinant sAxl is able to bind Gas6, therebyprotecting mice from life-threatening thrombosis (5), it is rea-sonable to suggest that the high levels of circulating endoge-nous sAxl observed in mouse serum in vivo may be physiolog-ically relevant to maintain the finely tuned balance betweenhemostatic and antithrombotic mechanisms. The bioavailabil-ity of sAxl could have a regulatory role and provide protectionagainst excessive Gas6 activity. It remains to be elucidatedwhether high or low levels of sAxl can correlate with distinctpathological conditions and serve as a prognostic marker forcertain types of diseases.

Remarkably, Gas6 was not detected in serum from Axl�/�

mice, thereby indicating the existence of a certain compensa-tory and/or protective mechanism(s) and suggesting that Gas6might be able to elicit harmful effects if present in a free formin the serum. Results from overexpression studies indicate thatAxl is required for the secretion of Gas6 and might play animportant role in cellular mechanisms regulating Gas6 expres-sion. Further studies are necessary to test whether the amountof Gas6 may be elevated in situ in distinct pathophysiologicsituations, such as the vascular injury repair process andthrombus formation, and how high circulating levels of en-dogenous sAxl could affect the function of Gas6 under thesecircumstances in vivo. Furthermore, the feasibility of usingrecombinant exogenous sAxl as a therapeutic agent againstlife-threatening thrombosis should be tested.

Many studies have documented the ability of Gas6 to sup-port cell growth and survival (2, 10, 31). Gas6 inhibits granu-locyte adhesion to endothelial cells promoted by several fac-tors (7) and mediates cell adhesion to phosphatidylserine (49),while Mer is critical for the engulfment and efficient clearanceof apoptotic cells (59). Gas6 induces through Axl STAT3-dependent mesangial cell proliferation in experimental glo-merulonephritis (68–70), and both are implicated in glomeru-lar hypertrophy at the early stages of diabetic nephropathy(47). Gas6 and Axl are also involved in the pathogenesis ofrheumatoid arthritis (51). RTKs of this family play an impor-tant immunoregulatory role (42) and are essential regulators ofmammalian spermatogenesis (41). Thus, a potential ability ofsAxl to serve as a natural antagonist of Gas6 could have clinicalrelevance.

Also of importance is the fact that the extracellular domainof an RTK responsible for ligand binding reportedly plays aregulatory role in controlling receptor kinase activity, whereasdeletion or truncation of this part relieves negative constraints

that regulate the kinase function (50). A number of studieshave provided compelling evidence for the significance of theectodomain in regulating RTK activity by showing that in vitrodeletion of this part renders many of these receptors trans-forming in the absence of ligand, including the epidermalgrowth factor receptor, the insulin receptor, and erbB2 (8, 22,65). In accord with the fact that a number of ligands canenhance shedding of their own receptors (21, 25), Gas6 in-duced a modest up-regulation of sAxl release in resistant L929fibroblasts abundantly expressing Axl on the cell surface. Thus,the proteolytic processing of Axl by ADAM10 may, at leastpartially, depend on, and result from, ligand binding. Impor-tantly, the proteolytic cleavage of Axl leaves a membrane-bound carboxy-terminal fragment which is a substrate for reg-ulated intramembrane proteolysis (RIP). The proteolyticprocessing and shedding of the first extracellular domain con-stitute a critical step in the RIP of several substrates, includingNotch, amyloid precursor protein, and CD44 (28). Initial re-sults from our laboratory suggest that Axl shedding also en-hances a presenelin/�-secretase-mediated RIP of the remain-ing cytoplasmic portion of Axl, which subsequently translocatesto the nucleus (E. Bulanova et al., unpublished data).

In summary, our data demonstrate that Axl is constitutivelyconverted into its soluble form by proteolytic cleavage andimplicate ADAM10 as the metalloproteinase responsible forthis RTK shedding. Rather high levels of sAxl are present inmouse serum and cell-conditioned medium from murine pri-mary and transformed cells. Given that sAxl can form hetero-complexes with Gas6 and the ability of immobilized Axl-Fc tomediate signaling events and promote cell migration, the bio-availability of sAxl might have broad implications in diversephysiologic processes or their pathological deviations mediatedby Axl itself, as well as by other members of this RTK family.

ACKNOWLEDGMENTS

We are grateful to Stefan Rose-John for TACE�/� MEFs, DieterAdam for immortalization of Axl�/� MEFs and for the platelet-de-rived growth factor receptor expression vector, Falk Fahrenholz forthe bovine ADAM10 expression vector, Paola Bellosta for ARKcDNA, and Helmut Brade for LPS. We thank Martina Hein andManuel Fohlmeister for excellent technical assistance.

This work was supported by Deutsche Forschungsgemeinschaftgrant SFB415 (A10 to S.B.P. and B9 to P.S.).

REFERENCES

1. Abel, S., C. Hundhausen, R. Mentlein, A. Schulte, T. A. Berkhout, N. Broad-way, D. Hartmann, R. Sedlacek, S. Dietrich, B. Muetze, B. Schuster, K.-J.Kallen, P. Saftig, S. Rose-John, and A. Ludwig. 2004. The transmembraneCXC-chemokine ligand 16 is induced by IFN-� and TNF-� and shed by theactivity of the disintegrin-like metalloproteinase ADAM10. J. Immunol.172:6362–6372.

2. Allen, M. P., C. Zeng, K. Schneider, X. Xiong, M. K. Meintzer, P. Bellosta,C. Basilico, B. Varnum, K. A. Heidenreich, and M. E. Wiermann. 1999.Growth arrest-specific gene 6 (Gas6)/adhesion related kinase (Ark) signalingpromotes gonadotropin-releasing hormone neuronal survival via extracellu-lar signal-regulated kinase (ERK) and Akt. Mol. Endocrinol. 13:191–201.

3. Althoff, K., P. Reddy, N. Voltz, S. Rose-John, and J. Mullberg. 2000. Shed-ding of interleukin-6 receptor and tumor necrosis factor alpha. Eur. J. Bio-chem. 267:2624–2631.

4. Angelillo-Scherrer, A., P. G. de Frutos, C. Aparicio, E. Melis, P. Savi, F.Lupu, J. Arnout, M. Dewerchin, M. Hoylaerts, J. Herbert, D. Collen, B.Dahlback, and P. Carmeliet. 2001. Deficiency or inhibition of Gas6 causesplatelet dysfunction and protects mice against thrombosis. Nat. Med. 7:215–221.

5. Angelillo-Scherrer, A., L. Burnier, N. Flores, P. Savi, M. DeMol, P. Schaef-fer, J. M. Herbert, G. Lemke, S. P. Goff, G. K. Matsushima, H. S. Earp, C.Vesin, M. F. Hoylaerts, S. Plaisance, D. Collen, E. M. Conway, B. Wehrle-


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

Haller, and P. Carmeliet. 2005. Role of Gas6 receptors in platelet signalingduring thrombus stabilization and implications for antithrombotic therapy.J. Clin. Investig. 115:237–246.

6. Arribas, J., and J. Massague. 1995. Transforming growth factor-alpha andbeta-amyloid precursor protein share a secretory mechanism. J. Cell Biol.128:433–441.

7. Avanzi, G. C., M. Gallicchio, F. Bottarel, L. Gammaitoni, G. Cavalloni, D.Buonfiglio, M. Bragardo, G. Bellomo, E. Albano, R. Fantozzi, G. Garbarino,B. Varnum, M. Aglietta, G. Saglio, U. Dianzani, and C. Dianzani. 1998.GAS6 inhibits granulocyte adhesion to endothelial cells. Blood 91:2334–2340.

8. Bargmann, C. I., and R. A. Weinberg. 1988. Oncogenic activation of theneu-encoded receptor protein by point mutation and deletion. EMBO J.7:2043–2052.

9. Bellosta, P., M. Costa, D. A. Lin, and C. Basilico. 1995. The receptor tyrosinekinase ARK mediates cell aggregation by homophilic binding. Mol. Cell.Biol. 15:614–625.

10. Bellosta, P., Q. Zhang, S. P. Goff, and C. Basilico. 1997. Signaling throughthe Ark tyrosine kinase receptor protects from apoptosis in the absence ofgrowth stimulation. Oncogene 15:2387–2397.

11. Biscardi, J. S., F. Denhez, G. F. Buehler, D. A. Chesnutt, S. C. Baragona,J. P. O’Bryan, C. J. Der, J. J. Fiordalisi, D. W. Fults, and P. F. Maness. 1996.Rek, a gene expressed in retina and brain, encodes a receptor tyrosine kinaseof the Axl/Tyro3 family. J. Biol. Chem. 271:29049–29059.

12. Black, R. A., C. T. Rauch, C. J. Kozlosky, J. J. Peschon, J. L. Slack, M. F.Wolfson, B. J. Castner, K. L. Stocking, P. Reddy, S. Srinivasan, M. Gerhart,R. Davis, J. N. Fitzner, R. S. Johnson, R. J. Paxton, C. J. March, and D. P.Cerretti. 1997. A metalloproteinase disintegrin that releases tumor-necrosisfactor-� from cells. Nature 385:729–733.

13. Blobel, C. P. 1997. Metalloprotease-disintegrins: links to cell adhesion andcleavage of TNF � and Notch. Cell 90:589–592.

14. Blobel, C. P. 2005. ADAMs: key components in EGFR signaling and devel-opment. Nat. Rev. 6:32–43.

15. Brandt, K., R. Ruckert, D. C. Foster, and S. Bulfone-Paus. 2003. Interleu-kin-21 inhibits dendritic cell activation and maturation. Blood 102:4090–4098.

16. Brunelleschi, S., L. Penengo, M. M. Santoro, and G. Gaudino. 2002. Recep-tor tyrosine kinases as target for anti-cancer therapy. Curr. Pharm. Des.8:1959–1972.

17. Budagian, V., P. Nanni, P.-L. Lollini, P. Musiani, E. Di Carlo, E. Bulanova,R. Paus, and S. Bulfone-Paus. 2002. Enhanced inhibition of tumor growthand metastasis, and induction of antitumor immunity by IL-2–IgG2b fusionprotein. Scand. J. Immunol. 55:484–492.

18. Budagian, V., E. Bulanova, Z. Orinska, A. Ludwig, S. Rose-John, P. Saftig,E. C. Borden, and S. Bulfone-Paus. 2004. Natural soluble interleukin-15R�is generated by cleavage that involves the tumor necrosis factor-alpha-con-verting enzyme (TACE/ADAM17). J. Biol. Chem. 279:40368–40375.

19. Bulanova, E., V. Budagian, T. Pohl, H. Krause, H. Durkop, R. Paus, and S.Bulfone-Paus. 2001. The IL-15R� chain signals through association with Sykin human B cells. J. Immunol. 167:6292–6302.

20. Bulfone-Paus, S., E. Bulanova, T. Pohl, V. Budagian, H. Durkop, R. Ruckert,U. Kunzendorf, R. Paus, and H. Krause. 1999. Death deflected: IL-15 in-hibits TNF-�-mediated apoptosis in fibroblasts by TRAF2 recruitment to theIL-15R� chain. FASEB. J. 13:1575–1585.

21. Chilton, P. M., and R. Fernandez-Botran. 1993. Production of soluble IL-4receptors by murine spleen cells is regulated by T cell activation and IL-4.J. Immunol. 151:5907–5917.

22. Christianson, T. A., J. K. Doherty, Y. J. Lin, E. E. Ramsey, R. Holmes, E. J.Keenan, and G. M. Clinton. 1998. NH2-terminally truncated HER-2/neuprotein: relationship with shedding of the extracellular domain and withprognostic factors in breast cancer. Cancer Res. 58:5123–5129.

23. Costa, M., P. Bellosta, and C. Basilico. 1996. Cleavage and release of asoluble form of the receptor tyrosine kinase ARK in vitro and in vivo. J. Cell.Physiol. 168:737–744.

24. Demarchi, F., R. Verardo, B. Varnum, C. Brancolini, and C. Schneider.2001. Gas6 anti-apoptotic signaling requires NF-�B activation. J. Biol.Chem. 278:31738–31744.

25. Dri, P., C. Gasparini, R. Menegazzi, R. Cramer, L. Alberi, G. Presani, S.Garbisa, and P. Patriarca. 2000. TNF-induced shedding of TNF receptors inhuman polymorphonuclear leukocytes: role of the 55-kDa TNF receptor andinvolvement of a membrane-bound and non-matrix metalloproteinase. J. Im-munol. 165:2165–2172.

26. Faust, M., C. Ebensperger, A. S. Schulz, L. Schleithoff, H. Hameister, C. R.Bartram, and J. W. Janssen. 1992. The murine ufo receptor: molecularcloning, chromosomal localization and in situ expression analysis. Oncogene7:1287–1293.

27. Fernandez-Bortran, R. 2000. Soluble cytokine receptors: novel immunother-apeutic agents. Expert Opin. Investig. Drugs 9:497–514.

28. Fortini, M. E. 2002. �-Secretase-mediated proteolysis in cell-surface-recep-tor signaling. Nat. Rev. 3:673–684.

29. Fridell, Y. C., Y. Jin, L. A. Quilliam, A. Buchert, P. McCloskey, G. Spizz, B.Varnum, C. Der, and E. T. Liu. 1996. Differential activation of the Ras/

extracellular-signal-regulated protein kinase pathway is responsible for thebiological consequences induces by the Axl receptor tyrosine kinase. Mol.Cell. Biol. 16:135–145.

30. Fridell, Y. C., J. Villa, E. C. Attar, and E. T. Liu. 1998. GAS6 inducesAxl-mediated chemotaxis of vascular smooth muscle cells. J. Biol. Chem.273:7123–7126.

31. Goruppi, S., E. Ruaro, and C. Schneider. 1996. Gas6, the ligand of Axltyrosine kinase receptor, has mitogenic and survival activities for serumstarved NIH3T3 fibroblasts. Oncogene 12:471–480.

32. Goruppi, S., E. Ruaro, B. Varnum, and C. Schneider. 1997. Requirement ofphosphatidylinositol 3-kinase-dependent pathway and Src for Gas6-Axl mi-togenic and survival activities in NIH 3T3 fibroblasts. Mol. Cell. Biol. 17:4442–4453.

33. Goruppi, S., C. Chiaruttini, M. E. Ruaro, B. Varnum, and C. Schneider.2001. Gas6 induces growth, �-catenin stabilization, and T-cell factor tran-scriptional activation in contact-inhibited C57 mammary cells. Mol. Cell.Biol. 21:902–915.

34. Graham, D. K., T. L. Dawson, D. L. Mullaney, H. R. Snodgrass, and H. S.Earp. 1994. Cloning and mRNA expression analysis of a novel human pro-tooncogene, c-mer. Cell Growth Differ. 5:647–657.

35. Guttridge, K. L., J. C. Luft, T. L. Dawson, E. Kozlowska, N. P. Mahajan, B.Varnum, and H. S. Earp. 2002. Mer receptor tyrosine kinase signaling.J. Biol. Chem. 277:24057–24088.

36. Hartmann, D., B. De Strooper, L. Serneels, K. Craessaerts, A. Herreman, W.Annaert, L. Umans, T. Lubke, A. L. Illert, K. von Figura, and P. Saftig. 2002.The disintegrin/metalloproteinase ADAM10 is essential for Notch signalingbut not for �-secretase activity in fibroblasts. Hum. Mol. Genet. 11:2615–2624.

37. Heaney, M. L., and D. W. Golde. 1996. Soluble cytokine receptors. Blood87:847–857.

38. Hundhausen, C., D. Misztela, T. A. Berkhout, N. Broadway, P. Saftig, K.Reiss, D. Hartmann, F. Fahrenholz, R. Postina, V. Matthews, K. J. Kallen,S. Rose-John, and A. Ludwig. 2003. The disintegrin-like metalloproteinaseADAM10 is involved in constitutive cleavage of CX3CL1 (fractalkine) andregulates CX3CL1-mediated cell-cell adhesion. Blood 102:1186–1195.

39. Hwang, C., M. Gatanaga, G. A. Granger, and T. Gatanaga. 1993. Mechanismof release of soluble forms of tumor necrosis factor/lymphotoxin receptors byphorbol myristate acetate-stimulated human THP-1 cells in vitro. J. Immu-nol. 151:5631–5638.

40. Jones, S. A., and S. Rose-John. 2002. The role of soluble receptors incytokine biology: the agonistic properties of the sIL-6R/IL-6 complex. Bio-chim. Biophys. Acta 1592:251–263.

41. Lu, Q., M. Gore, Q. Zhang, T. Camenisch, S. Boast, F. Casagranda, C. Lai,M. K. Skinner, R. Klein, G. K. Matsushima, H. S. Earp, S. P. Goff, and G.Lemke. 1999. Tyro-3 family receptors are essential regulators of mammalianspermatogenesis. Nature 398:723–728.

42. Lu, Q., and G. Lemke. 2001. Homeostatic regulation of the immune systemby receptor tyrosine kinases of the Tyro 3 family. Science 293:306–311.

43. Manfioletti, G., C. Brancolini, G. Avanzi, and C. Schneider. 1993. Theprotein encoded by a growth arrest-specific gene (gas6) is a new member ofthe vitamin K-dependent proteins related to protein S, a negative coregula-tor in the blood coagulation cascade. Mol. Cell. Biol. 3:4976–4985.

44. Mark, M. R., D. T. Scadden, Z. Wang, Q. Gu, A. Goddard, and P. J.Godowski. 1994. Rse, a novel receptor-type tyrosine kinase with homology toAxl/Ufo, is expressed at high levels in the brain. J. Biol. Chem. 269:10720–10728.

45. McCloskey, P., Y.-W. Fridell, E. Attar, J. Villa, Y. Jin, B. Varnum, and E. T.Liu. 1997. GAS6 mediates adhesion of cells expressing the receptor tyrosinekinase Axl. J. Biol. Chem. 272:23285–23291.

46. Melaragno, M. G., Y. C. Fridell, and B. C. Berk. 1999. The Gas6/Axl system.A novel regulator of vascular cell function. Trends Cardiovasc. Med. 9:250–253.

47. Nagai, K., H. Arai, M. Yanagita, T. Matsubara, H. Kanamori, T. Nakano, N.Iehara, A. Fukatsu, T. Kita, and T. Doi. 2003. Growth arrest-specific gene 6is involved in glomerular hypertrophy in the early stage of diabetic nephrop-athy. J. Biol. Chem. 278:18229–18234.

48. Nagata, K., K. Ohashi, T. Nakano, H. Arita, C. Zong, H. Hanafusa, and K.Mizuno. 1996. Identification of the product of growth arrest-specific gene 6 asa common ligand for Axl, Sky, and Mer receptor tyrosine kinases. J. Biol.Chem. 271:30022–30027.

49. Nakano, T., Y. Ishimoto, J. Kishino, M. Umeda, K. Inoue, K. Nagata, K.Ohashi, K. Mizuno, and H. Arita. 1997. Cell adhesion to phosphatidylserinemediated by a product of growth arrest-specific gene 6. J. Biol. Chem.272:29411–29414.

50. O’Bryan, J. P., Y.-W. Fridell, R. Koski, B. Varnum, and E. T. Liu. 1995. Thetransforming receptor tyrosine kinase, Axl, is post-translationally regulatedby proteolytic cleavage. J. Biol. Chem. 270:551–557.

51. O’Donnell, K., I. C. Harkes, L. Dougherty, and I. P. Wicks. 1999. Expressionof receptor tyrosine kinase Axl and its ligand Gas6 in rheumatoid arthritis.Am. J. Pathol. 154:1171–1180.

52. Paradies, N. E., and G. B. Grunwald. 1993. Purification and characterizationof NCAD90, a soluble endogenous form of N-cadherin, which is generated


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

by proteolysis during retinal development and retains adhesive and neurite-promoting function. J. Neurosci. Res. 36:33–45.

53. Peschon, J. J., J. L. Slack, P. Reddy, K. L. Stocking, S. W. Sunnarborg, D. C.Lee, W. E. Russell, B. J. Castner, R. S. Johnson, J. N. Fitzner, R. W. Boyce,N. Nelson, C. J. Kozlosky, M. F. Wolfson, C. T. Rauch, D. P. Cerretti, R. J.Paxton, C. J. March, and R. A. Black. 1998. An essential role for ectodomainshedding in mammalian development. Science 282:1281–1284.

54. Reiss, K., T. Maretzky, A. Ludwig, T. Tousseyn, B. De Strooper, D. Hart-mann, and P. Saftig. 2005. ADAM10 cleavage of N-cadherin and regulationof cell-cell adhesion and �-catenin nuclear signaling. EMBO. J. 24:742–752.

55. Sahin, U., G. Weskamp, K. Kelly, H.-M. Zhou, S. Higashiyama, J. Peschon,D. Hartmann, P. Saftig, and C. P. Blobel. 2004. Distinct roles for ADAM10and ADAM17 in ectodomain shedding of six EGFR ligands. J. Cell Biol.164:769–779.

56. Sato, T. A., M. B. Widmer, F. D. Finkelman, H. Madani, C. A. Jacobs, K. H.Grabstein, and C. R. Maliszewski. 1993. Recombinant soluble murine IL-4receptor can inhibit or enhance IgE responses in vivo. J. Immunol. 150:2717–2723.

57. Schlondorf, J., and C. P. Blobel. 1999. Metalloproteinase-disintegrins: mod-ular proteins capable of promoting cell-cell interactions and triggering sig-nals by protein-ectodomain shedding. J. Cell Sci. 112:3603–3617.

58. Schuermann, M. 1990. An expression vector system for stable expression ofoncogenes. Nucleic Acids Res. 18:4945–4946.

59. Scott, R. S., E. J. McMahon, S. M. Pop, E. A. Reap, R. Caricchio, P. L.Cohen, H. P. Earp, and G. K. Matsushima. 2001. Phagocytosis and clearanceof apoptotic cells is mediated by MER. Nature 411:207–211.

60. Six, E., D. Ndiaye, Y. Laabi, C. Brou, N. Gupta-Rossi, A. Israel, and F.Logeat. 2003. The Notch ligand Delta1 is sequentially cleaved by an ADAMprotease and gamma-secretase. Proc. Natl. Acad. Sci. USA 100:7638–7643.

61. Slack, B. E., L. K. Ma, and C. C. Seah. 2001. Constitutive shedding of theamyloid precursor protein ectodomain is up-regulated by tumor necrosisfactor-alpha converting enzyme. Biochem. J. 357:787–794.

62. Stenhoff, J., B. Dahlback, and S. Hafizi. 2004. Vitamin K-dependent Gas6

activates ERK kinase and stimulates growth of cardiac fibroblasts. Biochem.Biophys. Res. Commun. 319:871–878.

63. Sunnarborg, S. W., C. L. Hinkle, M. Srevenson, W. E. Russel, C. S. Raska,J. J. Reschon, B. J. Castner, M. J. Gerhart, R. J. Raxton, R. A. Black, andD. C. Lee. 2002. Tumor necrosis factor-alpha converting enzyme (TACE)regulates epidermal growth factor receptor ligand availability. J. Biol. Chem.277:12838–12845.

64. Varnum, B. C., C. Young, G. Elliott, A. Garcia, T. D. Bartley, Y. W. Fridell,R. W. Hunt, G. Trail, C. Clogston, R. J. Toso, D. Yanagihara, L. Bennett,Sylber, L. A. Merewether, A. Tseng, E. Escobar, E. T. Liu, and H. K.Yamane. 1995. Axl receptor tyrosine kinase stimulated by the vitamin K-dependent protein encoded by growth-arrest-specific gene 6. Nature 373:623–626.

65. Wang, L. H., B. Lin, S. M. Jong, D. Dixon, L. Ellis, R. A. Roth, and W. J.Rutter. 1987. Activation of transforming potential of the human insulinreceptor gene. Proc. Natl. Acad. Sci. USA 84:5715–5729.

66. Wang, X., K. He, M. Gerhart, Y. Huang, J. Jiang, R. J. Paxton, S. Yang, C.Lu, R. K. Menon, R. A. Black, G. Baumann, and S. J. Frank. 2002. Metal-loprotease-mediated GH receptor proteolysis and GHBP shedding. Deter-mination of extracellular domain stem region cleavage site. J. Biol. Chem.277:50510–50519.

67. Williams, A. F., and A. N. Barclay. 1988. The immunoglobulins superfamily-domains for cell surface recognition. Annu. Rev. Immunol. 6:381–405.

68. Yanagita, M., K. Ishii, H. Ozaki, H. Arai, T. Nakano, K. Ohashi, K. Mizuno,T. Kita, and T. Doi. 1999. Mechanism of inhibitory effect of warfarin onmesangial cell proliferation. J. Am. Soc. Nephrol. 10:2503–2509.

69. Yanagita, M., H. Arai, K. Ishii, T. Nakano, K. Ohashi, K. Mizuno, B.Varnum, A. Fukatsu, T. Doi, and T. Kita. 2001. Gas6 regulates mesangial cellproliferation through Axl in experimental glomerulonephritis. Am. J. Pathol.158:1423–1432.

70. Yanagita, M., H. Arai, T. Nakano, K. Ohashi, K. Mizuno, A. Fukatsu, T. Doi,and T. Kita. 2001. Gas6 induces mesangial cell proliferation via latent tran-scription factor STAT3. J. Biol. Chem. 276:42364–42369.


RETRACTED


cb.asm.org/

Dow

nloaded from

http://mcb.asm.org/

MOLECULAR AND CELLULAR BIOLOGY, Mar. 2011, p. 1330 Vol. 31, No. 60270-7306/11/$12.00 doi:10.1128/MCB.05019-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

RETRACTION

Soluble Axl Is Generated by ADAM10-Dependent Cleavage andAssociates with Gas6 in Mouse Serum

Vadim Budagian, Elena Bulanova, Zane Orinska, Erwin Duitman, Katja Brandt, Andreas Ludwig,Dieter Hartmann, Greg Lemke, Paul Saftig, and Silvia Bulfone-Paus

Department of Immunology and Cell Biology, Research Center Borstel, D-23845 Borstel, and Institute of Biochemistry,Christian Albrechts University, D-24118 Kiel, Germany; Department of Human Genetics, KU Leuven, and

Flanders Interuniversity Institute for Biotechnology (VIB-4), B-3000 Leuven, Belgium; andMolecular Neurobiology Laboratory, Salk Institute for Biological Studies,

San Diego, California 92186

Volume 25, no. 21, p. 9324–9339, 2005. The publisher hereby retracts this article due to evidence of data manipulation in Fig.2C, 4B, and 9, a clear violation of ASM’s ethical standards.

1330

Soluble Axl Is Generated by ADAM10-Dependent Cleavage and ...

Documents

Transcript of Soluble Axl Is Generated by ADAM10-Dependent Cleavage and ...