Role of Axl in Tumor Growth, Metastasis and Angiogenesis

Oncogene (2009) 28, 3442–3455 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE Axl as a potential therapeutic target in cancer: role of Axl in tumor growth, metastasis and angiogenesis

YLi1,7,XYe1,7, C Tan2, J-A Hongo2, J Zha3, J Liu4, D Kallop5, MJC Ludlam6 and L Pei1

1Department of Molecular Oncology, Genentech Inc., South San Francisco, CA, USA; 2Department of Antibody Engineering, Genentech Inc., South San Francisco, CA, USA; 3Department of Pathology, Genentech Inc., South San Francisco, CA, USA; 4Department of Bioinformatics, Genentech Inc., South San Francisco, CA, USA; 5Department of Tumor Biology and Angiogenesis, Genentech Inc., South San Francisco, CA, USA and 6Department of Cell Regulation, Genentech Inc., South San Francisco, CA, USA

Dysregulation of Axl and its ligand growth arrest-specific 6 two immunoglobin-like domains and dual fibronectin is implicated in the pathogenesis of several human cancers. type III repeats in the extracellular region and a In this study, we have used RNAi and monoclonal cytoplasmic kinase domain. The ligands for TAM antibodies to assess further the oncogenic potential of Axl. receptors are Gas6 (growth arrest-specific 6) and protein Here we show that Axl knockdown reduces growth of lung S, two vitamin K-dependent proteins that exhibit 43% and breast cancer xenograft tumors. Inhibition of Axl amino-acid sequence identity and share similar domain expression attenuates breast cancer cell migration and structures (Stitt et al., 1995; Varnum et al., 1995). Each inhibits metastasis to the lung in an orthotopic model, protein has a N-terminal Gla domain containing 11 providing the first in vivo evidence that links Axl directly to g-carboxyglutamic acid residues, followed by four cancer metastasis. Axl knockdown in endothelial cells epidermal growth factor-like modules, and a C-terminal impaired tube formation and this effect was additive with sex hormone-binding globulin (SHBG)-like structure anti-vascular endothelial growth factor (VEGF). Further consisting of two tandem laminin G domains. The analysis demonstrated that Axl regulates endothelial cell SHBG domain is both necessary and sufficient for TAM functions by modulation of signaling through angio- receptor binding and activation, whereas the Gla poietin/Tie2 and Dickkopf (DKK3) pathways. We have domain binds the negatively charged membrane phos- developed and characterized Axl monoclonal antibodies that pholipids and plays an important role in TAM-mediated attenuate non-small cell lung carcinoma xenograft growth phagocytosis of apoptotic cells (Hasanbasic et al., 2005; by downregulation of receptor expression, reducing tumor Sasaki et al., 2006). Axl can also exist as a soluble cell proliferation and inducing apoptosis. Our data demon- molecule, which is generated by ADAM10 (activity of strate that Axl plays multiple roles in tumorigenesis and disintergrin-like metalloproteinase 10)-mediated proteo- that therapeutic antibodies against Axl may block Axl lysis and is associated with Gas6 in mouse serum functions not only in malignant tumor cells but also in the (Budagian et al., 2005a). The biological function of sAxl tumor stroma. The additive effect of Axl inhibition with has not been characterized. anti-VEGF suggests that blocking Axl function could be an TAM activation and signaling has been implicated in effective approach for enhancing antiangiogenic therapy. multiple cellular responses including cell survival, Oncogene (2009) 28, 3442–3455; doi:10.1038/onc.2009.212; proliferation, migration and adhesion (Hafizi and published online 27 July 2009 Dahlba¨ck, 2006). TAM receptor signaling has been shown to regulate vascular smooth muscle homeostasis Keywords: RTK; tumorigenesis; metastasis; angiogen- (Korshunov et al., 2006, 2007), platelet function, esis; RNAi; therapeutic antibodies thrombus stabilization (Angelillo-Scherrer et al., 2001; Gould et al., 2005) and erythropoiesis (Angelillo- Scherrer et al., 2008). TAM receptors are also implicated in the control of oligodendrocyte cell survival (Shankar Introduction et al., 2006) and in the regulation of osteoclast function (Katagiri et al., 2001). Recent studies in knockout mice Axl belongs to the subfamily of receptor tyrosine have revealed that TAM receptors play pivotal roles in kinases (RTKs) that also includes Tyro3 and Mer innate immunity (Lemke and Rothlin, 2008). TAM (TAM; Lai and Lemke, 1991; O’Bryan et al., 1991). The inhibits inflammation in macrophages and dendritic TAM receptors are characterized by a combination of cells (Sharif et al., 2006; Rothlin et al., 2007), promotes the phagocytosis of apoptotic cells (Lu et al., 1999; Prasad et al., 2006) and stimulates the differentiation of Correspondence: Dr L Pei, Department of Molecular Oncology, natural killer cells (Caraux et al., 2006). In many of these Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USA. instances, the primary downstream TAM signaling E-mail: lpei@gene.com 7These authors contributed equally. pathway appears to be PI3K/AKT pathway (Angel- Received 4 February 2009; revised 3 June 2009; accepted 19 June 2009; illo-Scherrer et al., 2001; Keating et al., 2006; Shankar published online 27 July 2009 et al., 2006); however, the Janus kinase-STAT pathway Multiple functions of Axl in tumorigenesis YLiet al 3443 is essential for TAM-mediated immune responses tested, including GBM, prostate, pancreatic, colon, (Rothlin et al., 2007). In addition, cooperative interac- NSCLC and breast cancers (SH-SY5Y, a neuroblasto- tion between TAM receptor and cytokine receptor ma cell line which is known not to express Axl was used signaling network is required for many TAM-regulated as a negative control; Figure 1b). In addition, the ligand biological functions (Budagian et al., 2005b; Rothlin Gas6 was coexpressed in these cancer cell lines et al., 2007). (Figure 1b). To determine whether Axl is constitutively Axl was originally cloned from patients with chronic activated, we measured the level of phosphorylated Axl myelogenous leukemia and, when overexpressed, it by enzyme-linked immunosorbent assay (ELISA). exhibits transforming potential (Janssen et al., 1991; All cancer cell lines tested had detectable levels of O’Bryan et al., 1991). Axl overexpression has been phosphorylated Axl (Figure 1c). Treatment of cells with reported in a variety of human cancers (Craven et al., exogenously added Gas6 could significantly induce 1995; Ito et al., 1999; Berclaz et al., 2001; Sun et al., further Axl phosphorylation in A172, Calu1 and 2004; Shieh et al., 2005), and is associated with H1299 cells, but not in other cell lines. These results invasiveness and metastasis in lung (Shieh et al., 2005), suggest that Axl is constitutively activated to various prostate (Sainaghi et al., 2005), breast (Meric et al., levels in tumor cell lines, probably via autocrine action 2002), and gastric cancers (Wu et al., 2002) as well as in of endogenous Gas6. renal cell carcinoma (Chung et al., 2003) and glioblastoma (Hutterer et al., 2008). A recent study showed that Axl knockdown in cancer cell lines reduces cell viability Axl overexpression via a ‘tyrosine kinase switch’ leads to and attenuates their migration resistance to imatinib in gastrointestinal stromal tumors To examine the functional significance of Axl expression (Mahadevan et al., 2007). Axl expression is induced by in tumor growth, migration and metastasis, we estab- chemotherapy drugs and overexpression of Axl confers lished stable cell lines, in which the expression of Axl drug resistance in acute myeloid leukemia (Hong et al., could be knocked down by doxycycline-inducible 2008). Axl has also been shown to regulate endothelial AxlshRNA. We initially tested six Axl siRNA sequences cell migration and tube formation (Holland et al., 2005). for their ability to inhibit Axl expression in human These findings suggest that Axl may be involved in the umbilical vein endothelial cells (HUVECs). All six Axl- regulation of multiple aspects of tumorigenesis. specific siRNAs reduced Axl protein levels by more than In this study, we have used RNAi and monoclonal 90% as measured by fluorescence-activated cell sorting antibodies to further examine the oncogenic potential of (FACS) 48 h posttransfection (Supplementary Figures 1A Axl. We show that Axl expression promotes non-small andB),whereasthecontrolsiRNAhadnoeffectonAxl cell lung carcinoma (NSCLC) and breast cancer protein. We then used short hairpin RNA (shRNA) based xenograft tumor growth. Inhibition of Axl expression on one of the Axl siRNAs (Axl-4) to generated stable, attenuated MDA-MB-231 migration and inhibited lung inducible Axl knockdown cell lines. These cell lines were metastasis of these cancer cells in an orthotopic model. generated as pools to circumvent clonal variation. Axl knockdown in endothelial cells impaired tube Treatment of cancer cell lines stably expressing formation and this effect is additive with anti-vascular AxlshRNA with doxycycline (Dox) for 72 h resulted in endothelial growth factor (VEGF). Further, we ob- almost complete inhibition of Axl protein expression tained evidence that Axl regulates endothelial cell (Figure 2a). Axl knockdown was accompanied by functions by modulation of signaling through angio- inhibition of downstream signaling events. Although poietin/Tie2 and Dickkopf-homologue 3 (DKK3) path- Gas6 induced Akt phosphorylation in parental H1299 ways. In addition, we have developed and characterized and H1299AxlshRNA cells without Dox treatment, Axl monoclonal antibodies that block Gas6 binding, Gas6-induced Akt phosphorylation was completely downregulate Axl expression and phosphorylation, abolished in cells with Axl knockdown, similar to cells inhibit Gas6-depedent cell proliferation and attenuate treated with soluble receptor AxlFc (Figure 2b). A549 xenograft tumor growth. To test the effect of Axl knockdown on cancer cells in vitro, we performed cell proliferation and migration assays. Induction of Axl knockdown by Dox reduced cell viability in A549 (50%) and MDA-MB-231 (33%; Results Figure 2c). Although Axl knockdown in H1299 cells did not affect cell viability in the absence of Gas6, it abolished Axl is frequently overexpressed in cancer cell lines Gas6-mediated increase in viability of cells cultured in low We used microarray analysis to screen a large panel of serum (Figure 2c). Axl knockdown significantly attenu- cancer cell lines representing various tumor types for ated the motility MDA-MB-231 (40%) and A549 (48%) Axl mRNA expression. Axl exhibited a wide range of cells (Figure 2d). These results suggest that Gas6-Axl expression, from barely detectable to extremely high signaling promotes cancer cell growth and migration to levels, even within the same tissue type (Figure 1a). Axl various degrees depends on the cellular context. was overexpressed in a variety of cancer cell lines, with highest frequency in glioblastoma multiform (GBM) and pancreatic cancer. We used immunoblotting to Axl knockdown reduces tumor growth in xenograft models determine Axl protein levels in selected cancer cell lines. To evaluate whether Axl expression is required for Axl was highly expressed in all the cancer cell lines cancer cell growth in vivo, we tested the effect of Axl

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3444 15000

10000

5000 Affymetrix MAS5 intensity

0 (34) colon (23) SCLC (43) breast (32) glioma (9) prostate (11) ovarian (51) NSCLC (24) pancreatic (17) melanoma

KD 600 150 Axl 500

400 75 Gas6 Gas6 (-) 300 Gas6 (+) β 37 -Actin 200

100 U87 Phospho-Axl (pg/mg total protein) RKO A172 A549 Calu1 H1299 DU145 0 HCC366 SHSY5Y Pan0327 U87 A172 RKO Calu1 A549 DU145 H1299 HCC366SHSY5Y

MDA-MB-231 Pan0327 MDAMB231 Figure 1 Axl expression in cancer cell lines. (a) Axl mRNA levels in cancer cell lines. mRNA levels of Axl in cancer cell lines were compiled from several microarray experiments on Affymetrix GeneChip Human Genome U133 Plus 2.0 array, representing 244 samples (sample number for each tumor type is indicated). Probe set 202686_s_at was chosen to represent the expression of Axl. (b) Western blotting analysis of Axl and growth arrest-speciﬁc 6 (Gas6) in selected cancer cell lines. U87 and A172: glioma; DU145: prostate; Pan0327: pancreas; RKO: colon; MDA-MB-231: breast; Calu1, A549, H1299 and HCC366: non-small cell lung carcinoma (NSCLC); SHSY5Y: neuroblastoma. (c) Phosphorylated Axl in cancer cell lines. Cells were starved overnight in serum free medium and treated with Gas6 (200 ng/ml) for 30 min. Phospho-Axl was measured by enzyme-linked immunosorbent assay (ELISA).

knockdown on the growth of NSCLC A549 and H1299, Dox-treated groups. Axl knockdown signiﬁcantly and breast cancer MDA-MB-231 xenografts in mice. reduced growth of established MDA-MB-231 tumors Cells stably expressing AxlshRNA were implanted (33% inhibition; Figure 3c). Similar results were subcutaneously in nude (A549 and H1299) or SCID obtained in A549 established tumor models (data not (MDA-MB-231) mice, respectively. For the A549 and shown). The effective knockdown of Axl was veriﬁed by H1299 models, mice were randomized into treatment western blotting analysis of tumor lysates at the end of and control groups after implantation and administered the study (Figure 3). These results suggest that Axl with Dox or sucrose, respectively, in drinking water. signaling contributes to the in vivo growth of A549, As compared with untreated mice, Dox-induced Axl H1299 and MDA-MB-231 tumor xenografts. silencing resulted in a markedly impaired ability of A549 tumors to grow (Figure 3a). Dox-induced Axl knockdown in H1299 cells resulted in a moderate (25%) Axl knockdown inhibits metastasis of MDA-MB-231 inhibition of tumor growth in nude mice (Figure 3b). breast cancer cells to the lung For the MDA-MB-231 model, xenograft tumors were It has been demonstrated that Axl promotes motility allowed to grow to an average size of 200 mm3; mice and invasiveness of cancer cells in vitro (Figure 2d; Tai were then randomized and divided into untreated or et al., 2008; Zhang et al., 2008); however, the role of Axl

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3445

Axl Axl Axl AxlFc - - + - - + - - + CCshRNA shRNA CshRNA Gas6 - + + - + + - + + 75 Dox -++ - + -++ - - + - pAkt 150 50 Axl 100 Akt β 37 -actin 50 A549 H1299 MDA-MB-231 Dox (-) Dox(+) H1299 H1299AxlShRNA

600000 120 500000 100

80 400000

60 Dox (-) Dox (+) 300000 40 Viability (%) 200000

20 RLU (cell viability)

0 100000 A549 H1299 MDA-MB-231 AxlshRNA AxlshRNA AxlshRNA 0 Gas6 -++ - (200 ng/ml) Dox (-) Dox (+) H1299AxlshRNA 2000 1800 1600 1400 1200 Dox (-) 1000 Dox(+) 800 600 RFU (migration) 400 200 0 MDA-MB231 A549 H1299 AxlshRNA AxlshRNA AxlshRNA Figure 2 Axl knockdwn in cancer cell lines reduces cell viability and attenuates migration. (a) Inducible Axl knockdown in cancer cell lines (western blotting analysis). AxlshRNA expression was induced by treatment of cells with doxycycline for 72 h. (b) Axl knockdown abolished downstream signaling. Cells were serum starved for 24 h, preincubated with 100 ng/ml AxlFc for 2 h and treated with growth arrest-specific 6 (Gas6) for 30 min. Western blotting analysis was performed for phospho-Akt and Akt. (c) Effect of Axl knockdown on cell viability. Cells were cultured either in 10% serum (left panel), or in 1% serum (H1299, right panel) with or without Gas6. AxlshRNA was induced by treatment of cell with doxycycline (Dox) for 72 h, and the cell viability was measured using CellTiter Glo assay (n ¼ 8). (d) Axl knockdown attenuates cell migration. The migration assay was performed as described (Supplementary materials and methods). Error bars represent standard deviation (n ¼ 8). in cancer metastasis remains to be determined. We removal of the primary tumors. Five weeks after approached this question by utilizing an orthotopic removal of the primary tumors, large metastatic foci model and monitoring tumor metastasis by biolumines- were detected in the lungs of three out of five mice in the cence imaging. We generated a stable MDA-MB-231 control group, whereas there was no sign of lung cell line that expresses both a Dox-inducible AxlshRNA metastasis in the Dox-treated animals (n ¼ 5; Figure 4b, and a luciferase reporter gene. This cell line displayed group 1). The presence of tumor foci in the lungs was high levels of bioluminescence as determined by an confirmed by hematoxylin and eosin (H&E) staining in vitro luciferase assay (data not shown). (Figure 4c), and by c-Met (a protein known to be Figure 4a outlines the experimental design for our expressed in MDA-MB-231 cells) immunohistochemis- study. At first we asked whether Axl expression affects try to definitively mark human tumor cells (not shown). dissemination of breast cancer cells to the distant organ These results suggest that Axl promotes breast cancer sites. Doxycycline treatment was started the day after cell metastasize to the lung. cell implantation. Primary tumors were well established We then asked whether Axl expression is required for 2 weeks after implantation. The primary tumors were metastatic growth once cancer cells have colonized the removed when they reached the size of 1000 mm3, and lung. To address this question, we allowed primary metastasis to the lung was monitored for 5 weeks after tumor size to reach 1000 mm3, removed the primary

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3446 A549shAxl 400 Dox (-) Dox (+) Dox (-) Tumor ID 435 436 437 439 440 424 425 426 427 428 Dox (+) 300 Axl 200

100 actin Tumor volume (mm^3) 0 06010 20 30 40 50 Days

H1299shAxl 2000 Dox (-) Dox (-) Dox (+) 1600 Dox (+) Tumor ID 872 873 875 876 877 856 857 861 862 863 1200 Axl 800 actin 400

Tumor volume (mm^3) 0 0244 8 12 16 20 Days

MDA-MB-231shAxl

3000 Dox (-) Dox (+) Dox (-) 2500 Dox (+) Tumor ID 307 308 310 320 327 301 304 309 313 316 2000 Axl 1500 1000 500 actin

Tumor volume (mm^3) 0 0405 10 15 20 25 30 35 Days Figure 3 Axl knockdown reduces tumor growth in xenograft models. The graphs on the left show tumor growth curves and the panels on the right depict western blotting analysis of Axl expression in tumor lysates. (a) A549AxlshRNA. (b) H1299AxlshRNA. (c) MDA- MB-231AxlshRNA. The graphs are representative of three independent experiments. Error bars represent standard error of the mean (n ¼ 10 for each group in each experiment).

tumors and then started treatment with Dox. Mice were in HUVECs did not significantly affect cell growth monitored for 5 weeks after removal of primary tumors. (Supplementary Figure 1C). Both the control and the Dox-treated mice had We next used a primary HUVEC/PASMC (pulmon- metastatic foci (3 out of 5 in each group; Figure 4b, ary artery smooth muscle cell) coculture branching tube group 2). These results suggest that Axl is not critical for formation assay to assess the importance of Axl in metastatic loci growth once cancer cells established angiogenesis. Tube formation as quantified by tube themselves in the lung. length was reduced B40% in HUVECs transfected with Axl siRNA compared with cells transfected with the control siRNA (Figure 5b). These results suggest that Axl knockdown in HUVECs impairs endothelial tubule Axl is involved in the regulation of endothelial tube formation morphogenesis. Previous study has shown that Axl is overexpressed not only in tumors but also in the surrounding vascular cells (Hutterer et al., 2008). To evaluate further the potential Axl modulates expression of angiogenic factors role of Axl in the regulation of angiogenesis, we tested Dickkopf-homologue 3 and angiopoietin-2 the effect of Axl knockdown on endothelial tube To gain insight into the mechanisms involved in the formation. FACS analysis demonstrated expression of regulation of angiogenesis by Axl, we performed gene Axl on the surface of HUVECs (Figure 5a). Forty-eight profiling in HUVECs transfected with Axl or control hours posttransfection, Axl protein level was down- siRNAs. Figure 6a lists the top eight genes that were upregulated more than 90% in HUVECs transfected with or downregulated in HUVECs after Axl knockdown. Axl-specific siRNA, whereas the control luciferase Two of the genes, angiopoietin-2 (Ang-2) and DKK3, siRNA (GL2) had no effect (Figure 5a). Axl knockdown are known regulators of angiogenesis. We confirmed the

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3447

Figure 4 Axl knockdown inhibits metastasis of MDA-MB-231 breast cancer cells to the lung. (a) Schematic presentation of the experimental design. (b) Bioluminescence imaging of the lung 5 weeks after removal of the primary tumors. (c) Hematoxylin and eosin (H&E) staining of the lung 5 weeks after removal of the primary tumors. The circle indicate the tumor mass. This Figure is representative of three independent experiments (n ¼ 5 for each group in each experiment). microarray results by quantitative RT–PCR for DKK3 Axl knockdown has additive effect with anti-VEGF and Ang-2 (Figure 6b). In addition, the amount of the As VEGF is a major factor that regulates endothelial DKK3 and Ang-2 proteins in HUVEC lysates and in the cell functions, we tested the effect of Axl knockdown culture medium also changed following Axl knockdown together with anti-VEGF monoclonal antibody in tube as measured by ELISA (Supplementary Figure 2). formation assay. Treatment of cells with anti-VEGF To understand the relationship between Axl and DKK3 alone resulted in dose-dependent reduction in tube or Ang-2, we tested the effect of DKK3 and Ang-2 siRNA formation (not shown), Axl knockdown together with knockdown on Axl expression. Knockdown of neither anti-VEGF resulted in enhanced inhibition of tube DKK3 nor Ang-2 affected Axl protein level (Supplemen- formation (Figure 7), similar results were seen with tary Figure 2). In addition, knockdown of DKK3 had no DKK3 knockdown (Figure 7). These results suggest that effect on Ang-2, nor did Ang-2 knockdown affect DKK3 downregulation of Axl expression has an additive effect expression (Supplementary Figure 2). We next tested the with anti-VEGF to inhibit endothelial tube formation. effect of DKK3 and Ang-2 siRNAs on endothelial tube formation. Knockdown of DKK3 by RNAi reduced tube formation to a similar extent as did Axl knockdown, Generation of monoclonal antibodies that block whereas Ang-2 siRNA alone had no signiﬁcant effect on Axl function tubule formation (Figure 6c). These results suggest that Our validation data using RNAi suggested that Axl both DKK3 and Ang-2 are downstream targets of Axl is involved in tumor growth, metastasis as well as in and that their expression may be regulated by different the regulation of angiogenesis. We therefore set signaling pathways. out to develop monoclonal antibodies against Axl.

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3448 100 No 1st Ab 250

80 Axl-4 200 GL-2

60 150

MFI 100

% of Max 40 50 20 0

D-3 D-4 0 Axl-4 100 101 102 103 104 siRNA FL2-H: FL2-H PE untransfectedControl (GL2)

control

50000 45000 40000 35000 30000 25000 20000 15000

Axl-4 total tube length 10000 5000 0

D-3 D-4 Axl-4

untransfectedControl (GL2) siRNA

Figure 5 Axl knockdown in human umbilical vein endothelial cells (HUVECs) impairs endothelial tubule formation. (a) Fluorescence-activated cell sorting (FACS) analysis of Axl expression on HUVEC cell surface 96 h posttransfection with control (GL2) or Axl siRNA. (b) Axl siRNAs reduce endothethial tube formation. Left panel: tubes were viewed and photographed under ImageXpressMICRO imaging system. Right panel: quantiﬁcation of total tube formation by measuring tube length using MetaXpress software.

We developed a cell-based assay to directly test the confirmed this result by using FACS to determine the ability of Axl mAbs to block Axl/Gas6-mediated cell ability of Axl mAbs to block Gas6 binding to cell growth. We generated a stable cell line in Baf3 pro-B surface Axl on A549 cells (Figure 8Bb). Axl mAbs cells that overexpresses Axl (Supplementary Figure 3A). 12A11, 8B5 and 3G9 downregulated receptor expression Expression of Axl in Baf3 cells resulted in Gas6- in A549 cells, whereas 4F8 had no effect (Figure 8C). dependent growth (Supplementary Figure 3B), which 4F8 was the only antibody that detected Axl in western could be inhibited by AxlFc in a dose-dependent manner blotting, suggesting that it recognizes a linear epitope (Supplementary Figure 3C). We immunized mice with (data not shown). In addition, 12A11 and 3G9 inhibited human Axl extracellular domain (ECD), and screened a Gas6-induced Axl phosphorylation in H1299 cells panel of hybrid supernatants using this assay. We (Figure 8C). identified four antibodies that blocked Gas6-induced These monoclonal antibodies did not cross-react with Baf3Axl cell growth (Supplementary Figure 3D). murine Axl (Supplementary Figure 4A), nor did they Purified antibodies showed dose-dependent inhibition cross-react with related receptors Tyro3 and Mer of Baf3Axl cell growth, with the strongest blocker, (Supplementary Figure 4B). The Axl mAbs did not 12A11, exhibiting an IC50 of B100 ng/ml (Figure 8A). compete with each other in cross-blocking assays (data To understand the mechanisms of Axl inhibition by not shown), suggesting that they bind different epitopes. these mAbs, we further characterized these antibodies. To localize the binding epitopes of mAbs, different Axl mAbs 3G9 and 8B5 blocked ligand binding to the portions of Axl ECD were in vitro transcribed and receptor in solid-phase ELISA, whereas 12A11 and 4F8 translated (Figure 8D) and then used as antigens in had no effect on ligand binding (Figure 8Ba). We ELISA. The epitopes for 3G9 and 8B5 appeared to be

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3449 10 Up-regulated Down-regulated

MYCN (9.1) IFI44L (-98) Axl DKK3 HLX (4.5) GJA4 (-56) 1 GAS7 (4.5) Axl (-20) Ang-2 HDAC9 (4.1) IFIT1 (-14) E2F1 (3.9) SCG5 (-8.2) -10 CXCR4 (3.6) CYTL1 (-8.1) PMCH (3.4) DPP4 (-7.3) ANG-2 (3.3) DKK3 (-5.3) Fold of change over control -100

GL2 Axl-D4

60000 50000 40000 30000 20000 DKK3-1 Ang-2

total tube length 10000 0 GL2 Axl-D4 DKK3-1 Ang-2 siRNA

Figure 6 Gene expresson profile of human umbilical vein endothelial cells (HUVECs) with Axl knockdown. (a) List of most up- and downregulated genes following Axl knockdown. HUVECs were transfected with control or Axl siRNA and RNA was prepared 72 h posttransfection. Microarray and data analysis were described in Materials and methods. Fold of change over control is shown. Definition for the genes in the list: MYCN: N-Myc; HLX: H2.0-like homeobox; GAS7: growth arrest-specific 7; HDAC9: histone deacetylase 9; E2F1: E2F transcription factor 1; CXCR4: chemokine (C-X-C motif) receptor 4; PMCH: promelanin-concentrating hormone; ANG-2: angiopoietin 2; IFI44L: interferon-induced protein 44-like; GJA4: gap junction protein, alpha 4; IFIT1: interferon- induced protein with tetratricopeptide repeats 1; SCG5: secretogranin V, CYTL1: cytokine-like 1; DPP4: dipeptidylpeptidase 4; DKK3: Dickkopf 3. (b) Quantitative RT–PCR for DKK3 and Ang-2 mRNA levels following Axl knockdown. (c) Effect of DKK3 or Ang-2 siRNA knockdown on endothelial tube formation. Left panel: tubes were viewed and photographed under ImageXpressMICRO imaging system. Right panel: quantification of total tube formation by measuring tube length using MetaXpress software. located within the first Ig-domain, consistent with their tumor samples showed that treatment of tumors with ability to block ligand binding to the receptor, whereas Axl mAb 12A11 resulted in 20% decrease in Ki67 the epitopes for 12A11 and 4F8 were mapped to the first staining (Figure 9c) and approximately 50% increase in fibronectin domain (Figure 8D). caspase 3 staining (not shown) 72 h postdosing. These results suggest that Axl mAbs attenuate A549 xenograft growth by downregulation of receptor expression, which Axl mAbs inhibit A549 NSCLC xenograft tumor growth in turn leads to increased apoptosis and deceased cell To evaluate whether mAb inhibition of Axl affects proliferation of tumor cells. These results are consistent tumor cell growth in vivo, A549 cells were implanted with Axl knockdown in A549 cells, confirming that Axl subcutaneously in nude mice. When the tumor size plays an important role in the regulation of tumor reached 100 mm3 (Day 0 in Figure 9a), animals were growth of this NSCLC cell line. randomized and treated with either Axl mAbs or a control antibody at 30 mg/kg, twice weekly. Axl mAbs significantly attenuated A549 tumor growth compared with control, around 40% inhibition 40 days postdosing Discussion (Figure 9a). To investigate the mechanism of tumor growth inhibition by Axl mAbs, we performed a In this study, we have investigated various roles of Axl pharmacodynamic study. Tumors were excised at 0, in tumorigenesis including tumor growth, metastasis as 24, 48 and 72 h after administration of antibodies and well as angiogenesis. We used inducible shRNA to tumor lysate was generated. Western blotting analysis inhibit Axl expression in human cancer cell lines and showed that the Axl mAbs downregulated Axl expres- demonstrated that Axl expression promotes tumor sion in tumors at 24 h postdosing, which was sustained growth in NSCLC and breast cancer models. Previous through 72 h (Figure 9b). Ki67 and caspase 3 staining of studies showed that Axl expression promotes tumor

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3450

Figure 7 Axl knockdown has additive effect with anti-vascular endothelial growth factor (VEGF). Human umbilical vein endothelial cell (HUVEC) cell were transfected with control, Axl or Dickkopf-homologue 3 (DKK3) siRNAs with or without anti-VEGF and cocultured with pulmonary artery smooth muscle cell (PASMCs), Tube formation were analysed 3 days after coculture by staining with ﬂuorescein isothiocyanate (FITC)-conjugated anti-CD31 antibodies. Upper panels: tubes were viewed and photographed under ImageXpressMICRO imaging system. Lower panel: quantiﬁcation of total tube formation by measuring tube length using MetaXpress software.

growth in xenograft models of glioblastoma (Vajkoczy genesis, local invasion and formation of metastasis are et al., 2006) and breast cancer (Holland et al., 2005), clinically the most relevant. Recent experimental pro- using an Axl dominant-negative mutant or shRNA, gress has identiﬁed several molecular pathways and respectively. However, in both of these studies, tumor cellular mechanisms that underlie the multistage process growth rate was compared between a control cell line of metastasis. These include tumor invasion, tumor cell and a cell line, in which either Axl or its signaling was dissemination through the bloodstream or lymphatic downregulated before implantation into mice. Although system, colonization of distant organs and outgrowth of this approach is useful, it does not assess the direct metastases (Christofori, 2006). Axl expression is asso- impact of Axl inhibition on tumor growth. Our ciated with invasiveness and metastasis in various approach took the advantage of inducible expression cancers including breast (Meric et al., 2002; Zhang of AxlshRNA and measured the growth of the same cell et al., 2008), lung (Shieh et al., 2005) and gastric line in the presence or absence of Axl expression. Our (Sainaghi et al., 2005) cancers, as well as in glioblastoma results show that the impact of Axl expression on tumor (Hutterer et al., 2008). The involvement of Axl in growth depends on the cell model used. Although Axl promoting cancer cell migration and invasion has been knockdown in A549 NSCLC almost completely inhib- demonstrated in vitro (Vajkoczy et al., 2006; Tai et al., ited tumor growth, the effect of Axl knockdown on 2008; Zhang et al., 2008). In breast cancer models, H1299 NSCLC and MDA-MB-231 breast cancer was ectopic expression of Axl was sufﬁcient to confer a moderate (about B30% growth inhibition). The higher highly invasive phenotype to weakly invasive MCF7 sensitivity of A549 cells to Axl knockdown may be due cells. Conversely, inhibition of Axl signaling by shRNA to the presence of an increased gene copy number of Axl knockdown or an anti-Axl antibody decreased mobility in these cells (unpublished result). and invasiveness of highly invasive breast cancer cells Metastasis accounts for approximately 90% of all (Zhang et al., 2008). Although these studies established cancer deaths; and of all processes involved in carcino- a role for Axl in promoting cell migration and invasion

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3451 B a 1.80 120 1.60 100 12A11 3G9 1.40 80 4F8 1.20 4F8 60 8B5 1.00 8B5 12A11

% viability 0.80 40 control

OD450nm 3G9 20 0.60 AxlFc control μ 0.40 0 (0.2 g/ml) 010500.4 2 μg/ml 0.20 0.00 1 10 100 1000 10000 Ab concentration (ng/ml) B b 100 No Gas6 100 100 100 No Gas6 80 80 No Gas6 80 80 No Gas6 8B5 60 60 60 3G9 12A11 60 4F8 control 40 40 40 40 % of Max control control control 20 20 20 20 0 0 0 0 0 1 2 3 4 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 10 10 10 10 10

0.6

12A11 8B5 3G9 4F8 0.5 C 1 2 24 1 2 24 1 2 24 1 2 24 hrs 0.4

150 Axl 0.3

OD450nm 0.2 37 β-actin 0.1

0 Gas6 -+++ mAbs - control 12A11 3G9 Axl2 Axl3 Axl4 Axl5 Ctrl 2.5

2 control 4F8 1.5 8B5 4F8 8B5 12A11 3G9 12A11 Axl-2(aa1-134) - + - + 1

OD450nm 3G9 Axl-3(aa1-221) - + - + no Abs Axl-4(aa1-324) + + + + 0.5 Axl-5(aa1-435) + + + + 0 Axl-2Axl-3 Axl-4 Axl-5 Ctrl Figure 8 Characterization of Axl mAbs. (A) Axl mAbs inhibit Baf3Axl cell growth. Baf3Axl cells were grown in medium containing 200 ng/ml growth arrest-speciﬁc 6 (Gas6), and treated with Axl mAbs at indicated concentrations for 48 h. Cell viability was measure by CellTirer Glo assay. (B) mAbs 3G9 and 8B5 block Gas6 binding to Axl (a: enzyme-linked immunosorbent assay (ELISA) and b: ﬂuorescence-activated cell sorting (FACS)). (C) Downregulation of Axl expression in A549 (western blotting analysis) and Gas6- induced phosphorylation of Axl in H1299 cells (ELISA) by mAbs. (D) Epitope mapping. Various portions of Axl extracellular domain (ECD) were in vitro transcribed and translated and were used as antigen in ELISA.

in vitro, whether expression of Axl results in metastasis functional signiﬁcance of Axl in metastasis. Our results has not been elucidated. In this study we have used an show that Axl expression is required for MDA-MB-231 orthotopic breast cancer model to investigate the cells to establish metastatic foci in the lung, because Axl

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3452 control

1200 control 3G9 1000 12A11 8B5 800

600

400

200 Tumor volume ( mm^3)

0 05010 20 30 40 Days First dose last dose

0 24 48 72 hr Tumor ID 4 7 13 21 29 36 2 17 35 11 28 47 12A11 Axl

12A11 β-actin

Tumor ID 4 7 13 22 41 45 23 27 4 21 30 48

Axl 3G9 β-actin

Tumor ID 437 13 9 40 34 39 49 18 50 54

Axl

8B5 β-actin

Figure 9 Axl mAbs inhibit A549 xenograft tumor growth. (a) Tumor growth curve. mAbs were administrated i.p. at 30 mg/kg, twice a week, starting when the mean tumor size reached 100 mm3. The ﬁrst (day 0) and last doses (day 40) of mAbs administration were indicated by arrows. The graph is representative of three independent experiments. Error bars represent standard error of the mean (n ¼ 10 for each group in each experiment). (b) Axl mAbs downregulate Axl expression. Mice were treated with mAbs at 30 mg/kg and tumors excised at the indicated time points. Cell lysates from tumors were used in western blot analysis for Axl expression. (c) Ki67 staining of tumor samples 72 h after administrating of Axl mAb12A11.

knockdown by shRNA completely abolished the ability with avb3 integrin complex (D’Arcangelo et al., 2006). of cells emerging from the primary tumors to colonize Axl and Gas6 are coexpressed in tumor associated the lungs. However, once metastatic foci had established vascular cells in gliomas (Hutterer et al., 2008). In in the lung, Axl did not seem to have a significant impact primary human breast cancer samples, we found strong on the out growth of the metastases. These results Axl staining in tumor stromal cells (unpublished implicate Axl in early stages of MDA-MB-231 cancer observation). Therefore Axl may impact tumorigenesis cell metastasis, and provided the first in vivo evidence not only in tumor cells but also by modulating tumor that directly links Axl to metastasis. stromal function such as angiogenesis. In this study, we Gas6/Axl signaling plays an important role in have performed experiments to understand the mechan- vascular biology by modulating survival and migration isms involved in the regulation of endothelial cell of vascular smooth muscle cells and endothelial cells functions by Axl. Our data show that Axl expression (Melaragno et al., 1999). The expression of Gas6/Axl is is not required for endothelial cell proliferation, as Axl upregulated in injured arteries and Gas6 induces class A knockdown by RNAi had little impact on the growth of scavenger receptor through activation of PI3K/Akt, these cells. However, inhibition of Axl expression leading to the formation of foam cells, an important step impaired endothelial tube formation. in atherosclerosis (Cao et al., 2001). Activation of Axl To understand the mechanisms involved, we per- signaling mediates antiapoptotic effects of laminar shear formed mRNA expression profiling to identify genes that stress in endothelial cells possibly through its association may be modulated by Axl expression. Axl knockdown

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3453 resulted in the downregulation of DKK3 and upregula- In this study, we have characterized a panel of tion of Ang-2, two factors known to be involved in hybridoma antibodies against Axl. We developed a angiogenesis. DKK3 is among the Dickkopf family of cell-based screen for monoclonal antibodies that block secreted modulators of Wnt signaling (Krupnik et al., Axl biological function. The mAbs we characterized 1999), and is strongly expressed in the developing heart here include two (3G9 and 8B5) that block ligand and blood vessels of mice and chickens (Monaghan et al., binding to the receptor, and one (12A11) that does not 1999). A recent study demonstrated DKK3 protein interfere with ligand binding, but inhibits receptor expression in blood vessels of highly vascularized phosphorylation. All three mAbs induce downregula- neoplasms, including GBM, high-grade non-Hodgkin’s tion of receptor expression. Most importantly, these lymphoma, melanoma and colorectal carcinomas (Mon- mAbs attenuated A549 NSCLC tumor growth. These aghan et al., 1999). Stable overexpression of murine results not only confirm the role of Axl in promoting DKK3 in B16F10 cells significantly increased micro- tumor growth as indicated by the shRNA studies, but vessel density in the C57/BL6 melanoma model. In also suggest that monoclonal antibodies may provide an addition, an in vitro study in primary endothelial colony- effective strategy for treatment of cancers that over- forming cells showed that siRNA knockdown of DKK3 express Axl. As these mAbs do not cross-react with did not significantly affect cell proliferation, but murine Axl, it was not possible to assess the importance decreased tube formation in matrigel (Untergasser of Axl for tumor angiogenesis in vivo and its impact on et al., 2008). These results are similar to our findings in the tumor stroma. We are currently developing cross- HUVEC after Axl or DKK3 knockdown. reacting mAbs to investigate these potential functions of Axl knockdown in HUVECs led to upregulation of Axl in vivo. Ang-2. The angiopoietin signaling system plays a key In summary, we have validated a role for Axl in role in the regulation of angiogenesis, vascular home- promoting tumor growth and provided the first evidence ostasis and vascular regression (Yancopoulos et al., that Axl expression is required for metastasis of breast 2000). This signaling system consists of angiopoietin 1 cancer cells to the lung. We demonstrated that Axl and 2 and their receptor Tie2 (Dumont et al., 1994). enhances endothelial tube formation by acting through Ang-1 is an agonist that supports EC survival and the angiopoietin and DKK3 signaling system. Our data endothelium integrity (Suri et al., 1998) through the suggest that therapeutic antibodies against Axl could PI3K/Akt signaling pathway (Peters et al., 2004). On the block Axl functions not only in the tumor itself but also other hand, Ang-2 (Maisonpierre et al., 1997) is a possibly in the surrounding stroma. The additive signaling antagonist and inhibits the action of Ang-1 to effect of Axl inhibition with anti-VEGF suggests that promote vessel destabilization. In the presence of blocking Axl function might be a useful approach for VEGF, Ang-2 provides an important angiogenic stimu- enhancing antiangiogenic therapy and warrant further lus (Yancopoulos et al., 2000). However, when VEGF investigation. signaling is inhibited or absent, the destabilizing action of Ang-2 on vessels leads to cell death and capillary regression (Yancopoulos et al., 2000). Our results Materials and methods demonstrate that Axl knockdown is additive with anti- VGEF to inhibit endothelial tube formation. This Antibodies and cell lines information as well as protocols additive effect may be mediated through upregulation including siRNA transfection, generating inducible AxlshR- of Ang-2 when Axl expression is inhibited. NA cell lines, proliferation and migration assays, coculture, Taken together, our results suggest the following microarray, generation and characterization of Axl mAbs are model for modulation of endothelial cell function by provided in Supplementary materials. Axl. Expression of Axl in EC results in upregulation of the proangiogenic factor DKK3 and downregulation of Ang-2, which in turn facilitates signaling of Ang-1/Tie2 Xenograft experiments and results in the increase in angiogenesis. All studies were conducted in accordance with the Guidance for the Care and Use of Laboratory Animals (NIH), and The multiple roles that Axl plays in tumorigenesis approved by Institutional Animal Care and Used Committee make it an attractive therapeutic target for cancer. (IACUC). The detailed protocols are described in Supplemen- Anti-Axl antibodies are specific reagents that can be tary materials. used to investigate the role of Axl in tumorigenesis, and provide therapeutic potential to inhibit Axl-dependent signaling in patients. An antagonistic mAb can achieve its function by various mechanisms: (1) blocks Conflict of interest ligand binding to the receptor, (2) promotes receptor degradation, (3) blocks receptor dimerization or (4) The authors declare no conflict of interest. mediates antibody-dependent cell-mediated cytotoxicity. As many tumors may exhibit ligand-independent Acknowledgements activation of Axl, it is important to develop and compare mAbs that act through different modes of We thank Dr Avi Ashkenazi and Dr Jing Qing for reading the action, and select the lead mAb based on efficacy. article and providing helpful suggestions and comments.

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3454 References

Angelillo-Scherrer A, Burnier L, Lambrechts D, Fish RJ, Tjwa M, Janssen JW, Schulz AS, Steenvoorden AC, Schmidberger M, Strehl S, Plaisance S et al. (2008). Role of Gas6 in erythropoiesis and anemia Ambros PF et al. (1991). A novel putative tyrosine kinase receptor in mice. J Clin Invest 118: 583–596. with oncogenic potential. Oncogene 6: 2113–2120. Angelillo-Scherrer A, de Frutos P, Aparicio C, Melis E, Savi P, Katagiri M, Hakeda Y, Chikazu D, Ogasawara T, Takato T, Kumegawa Lupu F et al. (2001). Deficiency or inhibition of Gas6 causes M et al. (2001). Mechanism of stimulation of osteoclastic bone platelet dysfunction and protects mice against thrombosis. Nat Med resorption through Gas6/Tyro 3, a receptor tyrosine kinase signaling, 7: 215–221. in mouse osteoclasts. JBiolChem276: 7376–7382. Berclaz G, Altermatt HJ, Rohrbach V, Kieffer I, Dreher E, Andres Keating AK, Salzberg DB, Sather S, Liang X, Nickoloff S, Anwar A AC. (2001). Estrogen dependent expression of the receptor tyrosine et al. (2006). Lymphoblastic leukemia/lymphoma in mice over- kinase axl in normal and malignant human breast. Ann Oncol 12: expressing the Mer (MerTK) receptor tyrosine kinase. Oncogene 25: 819–824. 6092–6100. Budagian V, Bulanova E, Orinska Z, Duitman E, Brandt K, Ludwig A Korshunov VA, Daul M, Massett MP, Berk BC. (2007). Axl mediates et al. (2005a). Soluble Axl is generated by ADAM10-dependent vascular remodeling induced by deoxycorticosterone acetate-salt cleavage and associates with Gas6 in mouse serum. Mol Cell Biol 25: hypertension. Hypertension 50: 1057–1062. 9324–9339. Korshunov VA, Mohan AM, Georger MA, Berk BC. (2006). Axl, a Budagian V, Bulanova E, Orinska Z, Thon L, Mamat U, Bellosta P receptor tyrosine kinase, mediates flow-induced vascular remodel- et al. (2005b). A promiscuous liaison between IL-15 receptor and ing. Circ Res 98: 1446–1452. Axl receptor tyrosine kinase in cell death control. EMBO J 24: Krupnik VE, Sharp JD, Jiang C, Robison K, Chickering TW, 4260–4270. Amaravadi L et al. (1999). Functional and structural diversity of Cao WM, Murao K, Imachi H, Sato M, Nakano T, Kodama T et al. the human Dickkopf gene family. Gene 238: 301–313. (2001). Phosphatidylinositol 3-OH kinase-Akt/protein kinase B Lai C, Lemke G. (1991). An extended family of protein-tyrosine kinase pathway mediates Gas6 induction of scavenger receptor a in genes differentially expressed in the vertebrate nervous system. immortalized human vascular smooth muscle cell line. Arterioscler Neuron 6: 691–704. Thromb Vasc Biol 21: 1592–1597. Lemke G, Rothlin CV. (2008). Immunobiology of the TAM receptors. Caraux A, Lu Q, Fernandez N, Riou S, Di Santo JP, Raulet DH et al. Nat Rev Immunol 8: 327–336. (2006). Natural killer cell differentiation driven by Tyro3 receptor Lu Q, Gore M, Zhang Q, Camenisch T, Boast S, Casagranda F et al. tyrosine kinases. Nat Immunol 7: 747–754. (1999). Tyro-3 family receptors are essential regulators of mamma- Christofori G. (2006). New signals from the invasive front. Nature 441: lian spermatogenesis. Nature 398: 723–728. 444–450. Mahadevan D, Cooke L, Riley C, Swart R, Simons B, Della Croce K Chung BI, Malkowicz SB, Nguyen TB, Libertino JA, McGarvey TW. et al. (2007). A novel tyrosine kinase switch is a mechanism of (2003). Expression of the proto-oncogene Axl in renal cell imatinib resistance in gastrointestinal stromal tumors. Oncogene 26: carcinoma. DNA Cell Biol 22: 533–540. 3909–3919. Craven RJ, Xu LH, Weiner TM, Fridell YW, Dent GA, Srivastava S Maisonpierre PC, Suri C, Jones PF, Bartunkova S, Wiegand SJ, et al. (1995). Receptor tyrosine kinases expressed in metastatic colon Radziejewski C et al. (1997). Angiopoietin-2, a natural antagonist cancer. Int J Cancer 60: 791–797. for Tie2 that disrupts in vivo angiogenesis. Science 277: 55–60. D’Arcangelo D, Ambrosino V, Giannuzzo M, Gaetano C, Capogrossi Melaragno MG, Fridell YW, Berk BC. (1999). The Gas6/Axl system: a MC. (2006). Axl receptor activation mediates laminar shear stress novel regulator of vascular cell function. Trends Cardiovasc Med 9: anti-apoptotic effects in human endothelial cells. Cardiovasc Res 71: 250–253. 754–763. Meric F, Lee WP, Sahin A, Zhang H, Kung HJ, Hung MC. (2002). Dumont DJ, Gradwohl G, Fong GH, Puri MC, Gertsenstein M, Expression profile of tyrosine kinases in breast cancer. Clin Cancer Auerbach A et al. (1994). Dominant-negative and targeted Res 8: 361–367. null mutations in the endothelial receptor tyrosine kinase, tek, Monaghan AP, Kioschis P, Wu W, Zuniga A, Bock D, Poustka A reveal a critical role in vasculogenesis of the embryo. Genes Dev 8: et al. (1999). Dickkopf genes are co-ordinately expressed in 1897–1909. mesodermal lineages. Mech Dev 87: 45–56. Gould WR, Baxi SM, Schroeder R, Peng YW, Leadley RJ, Peterson O’Bryan JP, Frye RA, Cogswell PC, Neubauer A, Kitch B, Prokop C JT et al. (2005). Gas6 receptors Axl, Sky and Mer enhance platelet et al. (1991). axl, a transforming gene isolated from primary human activation and regulate thrombotic responses. J Thromb Haemost 3: myeloid leukemia cells, encodes a novel receptor tyrosine kinase. 733–741. Mol Cell Biol 11: 5016–5031. Hafizi S, Dahlba¨ck B. (2006). Signaling and functional diversity within Peters KG, Kontos CD, Lin PC, Wong AL, Rao P, Huang L et al. the Axl subfamily of receptor tyrosine kinases. Cytokine Growth (2004). Functional significance of Tie2 signaling in the adult Factor Rev 17: 295–304. vasculature. Recent Prog Horm Res 59: 51–71. Hasanbasic I, Rajotte I, Blostein M. (2005). The role of gamma- Prasad D, Rothlin CV, Burrola P, Burstyn-Cohen T, Lu Q, Garcia de carboxylation in the anti-apoptotic function of gas6. J Thromb Frutos P et al. (2006). TAM receptor function in the retinal pigment Haemost 3: 2790–2797. epithelium. Mol Cell Neurosci 33: 96–108. Holland SJ, Powell MJ, Franci C, Chan EW, Friera AM, Atchison RE Rothlin CV, Ghosh S, Zuniga EI, Oldstone MB, Lemke G. (2007). et al. (2005). Multiple roles for the receptor tyrosine kinase axl in TAM receptors are pleiotropic inhibitors of the innate immune tumor formation. Cancer Res 65: 9294–9303. response. Cell 131: 1124–1136. Hong CC, Lay JD, Huang JS, Cheng AL, Tang JL, Lin MT et al. Sainaghi PP, Castello L, Bergamasco L, Galletti M, Bellosta P, Avanzi (2008). Receptor tyrosine kinase AXL is induced by chemotherapy GC. (2005). Gas6 induces proliferation in prostate carcinoma cell drugs and overexpression of AXL confers drug resistance in acute lines expressing the Axl receptor. J Cell Physiol 204: 36–44. myeloid leukemia. Cancer Lett 268: 314–324. Sasaki T, Knyazev PG, Clout NJ, Cheburkin Y, Go¨hring W, Ullrich A Hutterer M, Knyazev P, Abate A, Reschke M, Maier H, Stefanova N et al. (2006). Structural basis for Gas6-Axl signalling. EMBO J 25: et al. (2008). Axl and growth arrest-specific gene 6 are frequently 80–87. overexpressed in human gliomas and predict poor prognosis in Shankar SL, O’Guin K, Kim M, Varnum B, Lemke G, Brosnan CF patients with glioblastoma multiforme. Clin Cancer Res 14: 130–138. et al. (2006). Gas6/Axl signaling activates the phosphatidylinositol Ito T, Ito M, Naito S, Ohtsuru A, Nagayama Y, Kanematsu T et al. 3-kinase/Akt1 survival pathway to protect oligodendrocytes (1999). Expression of the Axl receptor tyrosine kinase in human from tumor necrosis factor alpha-induced apoptosis. J Neurosci thyroid carcinoma. Thyroid 9: 563–567. 26: 5638–5648.

Oncogene Multiple functions of Axl in tumorigenesis YLiet al 3455 Sharif MN, Sosic D, Rothlin CV, Kelly E, Lemke G, Olson EN et al. tumor endothelial cells and supports capillary formation. Int J (2006). Twist mediates suppression of inflammation by type I IFNs Cancer 122: 1539–1547. and Axl. J Exp Med 203: 1891–1901. Vajkoczy P, Knyazev P, Kunkel A, Capelle HH, Behrndt S, von Shieh YS, Lai CY, Kao YR, Shiah SG, Chu YW, Lee HS et al. (2005). Tengg-Kobligk H et al. (2006). Dominant-negative inhibition of the Expression of axl in lung adenocarcinoma and correlation with Axl receptor tyrosine kinase suppresses brain tumor cell growth tumor progression. Neoplasia 7: 1058–1064. and invasion and prolongs survival. Proc Natl Acad Sci USA 103: Stitt TN, Conn G, Gore M, Lai C, Bruno J, Radziejewski C et al. 5799–5804. (1995). The anticoagulation factor protein S and its relative, Gas6, Varnum BC, Young C, Elliott G, Garcia A, Bartley TD, Fridell YW are ligands for the Tyro 3/Axl family of receptor tyrosine kinases. et al. (1995). Axl receptor tyrosine kinase stimulated by the vitamin Cell 80: 661–670. K-dependent protein encoded by growth-arrest-specific gene 6. Sun W, Fujimoto J, Tamaya T. (2004). Coexpression of Gas6/Axl in Nature 373: 623–626. human ovarian cancers. Oncology 66: 450–457. Wu CW, Li AF, Chi CW, Lai CH, Huang CL, Lo SS et al. (2002). Suri C, McClain J, Thurston G, McDonald DM, Zhou H, Oldmixon Clinical significance of AXL kinase family in gastric cancer. EH et al. (1998). Increased vascularization in mice overexpressing Anticancer Res 22: 1071–1078. angiopoietin-1. Science 282: 468–471. Yancopoulos GD, Davis S, Gale NW, Rudge JS, Wiegand SJ, Tai KY, Shieh YS, Lee CS, Shiah SG, Wu CW. (2008). Axl promotes Holash J. (2000). Vascular-specific growth factors and blood vessel cell invasion by inducing MMP-9 activity through activation of formation. Nature 407: 242–248. NF-kappaB and Brg-1. Oncogene 27: 4044–4055. Zhang YX, Knyazev PG, Cheburkin YV, Sharma K, Knyazev YP, Untergasser G, Steurer M, Zimmermann M, Hermann M, Kern J, Orfi L et al. (2008). AXL is a potential target for therapeutic Amberger A et al. (2008). The Dickkopf-homolog 3 is expressed in intervention in breast cancer progression. Cancer Res 68: 1905–1915.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene