FoxO3a represses lymphangiogenesis in gastric cancer Ara Jo Department of Medical Science The Graduate School, Yonsei University

FoxO3a represses lymphangiogenesis in gastric cancer Ara Jo Department of Medical Science The Graduate School, Yonsei University

FoxO3a represses lymphangiogenesis in gastric cancer Directed by Professor Jae-Ho Cheong The Master's Thesis submitted to the Department of Medical Science, the Graduate School of Yonsei University in partial fulfillment of the requirements for the degree of Master of Medical Science Ara Jo December 2014

ACKNOWLEDGEMENTS 대학을졸업하고석사학위과정을하면서많이부족한저를가르쳐주시고격려해 주신모든분들께감사의말씀을전합니다. 먼저학위과정동안연구의방향을제시해주시고저를잘이끌어주신정재호지도교수님께가장감사드립니다. 아무것도몰랐던제가이논문을쓰고또잘마무리할수있었던것은옆에서항상격려와칭찬을아낌없이주신교수님덕분이라고생각합니다. 또한심사위원으로바쁘신와중에도많은관심을가져주시고부족한실험결과에대한조언과좋은말씀을해주신윤호근교수님과전경희교수님께진심으로감사드립니다. 매일마주하며가장가까운곳에서함께생활한우리팀원들모두감사합니다. 항상걱정해주시고뒤에서챙겨주신박은성교수님, 끊임없이후배사랑해주시는박기청박사님, 실험에대한노하우와많은지식을알려주신기현정박사님, 연구외에도좋은얘기많이해주신이재은박사님, 웃음코드가잘맞아하루하루재미있게생활할수있게해준정민언니, 항상웃는얼굴로인사해주시는심문희선생님, 힘들고힘쓰는일도맡아해주시는이도형선생님, 제논문에관심가져주시고격려해주신박경호선생님, 최윤형선생님께감사드립니다. 그리고가장중요한내단짝혜지, 함께울고웃으며힘들고지칠때정신적으로큰힘이되

어주고항상내편이되어준우리혜지에게도고마움을전합니다. 그외에도실험실생활을하면서도움과격려를준성호오빠, 주만오빠, 재성오빠에게감사를드리고같이학위과정을하면서큰힘이되어준지혜, 지은, 인수오빠, 준혁오빠, 영인오빠그리고급할때마다실험재료를구해주신이진석선생님에게도고마운마음을전합니다. 또한가족과도같은저의영원한친구들인채희, 리나, 수지에게고마움을전합니다. 마지막으로제가학위과정을마치고논문을마무리하는데까지가장큰힘이되고든든한지원군인나의사랑하는우리가족에게진심으로감사를드립니다. 언제나항상저를믿어주시고응원해주셨기에제가지금까지잘성장할수있었던것같습니다. 자나깨나저를위해항상기도하시는외할머니, 저를항상믿어주시고지원해주시는엄마, 뒤에서응원해주시는아빠, 딸처럼하나부터열까지다잘챙겨주시는외삼촌, 큰이모, 작은이모그리고공부하는누나를많이이해해준동생문희에게도고마움과사랑을전합니다. 2014 년 12 월, 논문을마무리하며모든분들께감사의마음을전합니다. 감사합니다. 조아라

TABLE OF CONTENTS ABSTRACT 1 Ⅰ. INTRODUCTION 3 Ⅱ. MATERIALS AND METHODS 11 1. Cell culture 11 2. Establishment of stable Foxo3a knockdown cells 11 3. Establishment of stable FoxO3a-Overexpression cells 12 4. sirna transfection 12 5. ELISA 13 6. Tube formation assay 13 7. Migration assay 14 8. Reverse transcription polymerase chain reaction (RT-PCR) 14 9. Quantitative Real-time PCR 14 10. Western blot analysis 15 11. Electrophoretic mobility shift assay (EMSA) and super-shift assay 16 12. Immunofluorescence assay 17 13. Chemicals 17

14. Reverse Phase Protein Array (RPPA) 17 15. Statistical analysis 18 III. RESULTS 19 1. Phosphorylated FoxO3a is prognostic factor in gastric cancer patients 19 2. FoxO3a and VEGF-C expression showed a negative correlation in SK4 and AGS cell lines 21 3. Secreted VEGF-C levels and tube formation efficiency of human lymphoendothelial cells (HLECs) in conditioned media of AGS were higher than those in SK4 cells 24 4. Silencing FoxO3a in gastric cancer cells enhances the lymphangiogenic activity of HLECs 28 5. Overexpression of FoxO3a in gastric cancer cells inhibits the lymphangiogenic activity of HLECs 33 6. FoxO3a represses VEGF-C expression in gastric cancer cells 37 7. FoxO3a binds to the FHRE of VEGF-C promoter region 40 8. FoxO3a is inactivated by PI3K/AKT pathway 43 9. FoxO3a expression is negatively correlated with lymph node metastasis in gastric cancer patients 47

IV. DISCUSSION 50 V. CONCLUSION 52 REFERENCES 53 ABSTRACT (IN KOREAN) 60

LIST OF FIGURES Figure 1. Exploring prognostic proteins using RPPA in gastric cancer patients 20 Figure 2. FoxO3a and VEGF-C expressions in gastric cancer cell lines 23 Figure 3. Secreted VEGF-C expression and tube formation of HLECs in conditioned media of gastric cancer cell lines 26 Figure 4. Knockdown of FoxO3a in SK4 cells enhances lymphangiogenesis 31 Figure 5. FoxO3a overexpression in AGS cells represses lymphangiogenesis 36 Figure 6. Silencing of FoxO3a in SK4 enhances the VEGF-C expression 39 Figure 7. FoxO3a binds to the FHRE of VEGF-C promoter region 42 Figure 8. FoxO3a localization and PI3K/AKT activity in gastric cancer cells 46 Figure 9. Diagram illustrating the relationship between FoxO3a and lymphangiogenesis in gastric cancer 49

LIST OF TABLES Table 1. FoxO3a expression is anti-correlated with lymph node metastasis in human gastric cancer tissue 48

ABSTRACT FoxO3a represses lymphangiogenesis in gastric cancer Ara Jo Department of Medical Science The Graduate School, Yonsei University (Directed by Professor Jae-Ho Cheong) Lymph node metastasis is the most important prognostic factor and is associated with about 60% of gastric cancer. For cancer cells to metastasize to lymph nodes, lymphangiogenesis is required. FoxO3a, a tumor suppressive transcription factor, is known to be a negative regulator of angiogenesis. Despite that lymphangiogenesis and angiogenesis in cancer are interrelated processes, the role of Foxo3a in lymphangiogenesis remains unknown. Thus, the goal of the study is to investigate the clinical significance and the role of FoxO3a in lymph node metastasis and lymphangiogenesis in gastric cancer. Foxo3a expression was analyzed across a large number of gastric cancer patient tissue samples with reverse phase protein microarray (RPPA). To study the biological function, FoxO3a was knocked down with RNA interference or overexpressed with an expression vector in gastric cancer cells. Tube formation and migration assays were carried out in human lymphatic endothelial cells (HLECs) with conditioned media of FoxO3a-silenced or overexpressed gastric cancer cells, 1

respectively. To assess the DNA binding activity of FoxO3a, electrophoretic mobility shift assay (EMSA) was performed. RPPA analysis revealed that FoxO3a expression was inversely correlated with lymph node metastasis and high expression of phosphorylated FoxO3a was a poor prognostic factor in gastric cancer patients. Silencing of FoxO3a in gastric cancer cells profoundly induced VEGF-C expression and secretion, increased tube formation, and migration of HLECs treated with the conditioned media while overexpression of FoxO3a in gastric cancer cells showed the opposite effects. EMSA with supershift assay demonstrated that FoxO3a binds to the FHRE in the promoter region of VEGF-C. Treatment of LY294002, a PI3K pathway inhibitor, reversed the phosphorylation of FoxO3a and increased the nuclear localization of FoxO3a in gastric cancer cells. Collectively, our data clearly illustrate that FoxO3a binds to VEGF-C promoter and transcriptionally represses the expression of VEGF-C thereby inhibiting tumor lymphangiogenesis. -------------------------------------------------------------------------------------------------------------------------- Key words: foxo3a, vegf-c, lymphangiogenesis, gastric cancer 2

FoxO3a represses lymphangiogenesis in gastric cancer Ara Jo Department of Medical Science The Graduate School, Yonsei University (Directed by Professor Jae-Ho Cheong) Ⅰ. INTRODUCTION Gastric cancer Gastric cancer is the fourth most frequent malignancy and is the second leading cause of cancerrelated death in the world 1-3. The highest incidences of gastric cancer are reported in China, Japan, Korea, and other Eastern Asian countries. It is also one of the most common malignant tumors in Korea 4-6. The overall prognosis of gastric cancer is poor with a 5-year survival rate below 30% for most countries 7. Several potential risk factors include high salt diet, low intake of vegetables and fruits, smoking, chronic gastritis with glandular atrophy and intestinal metaplasia, and Helicobacter pylori (H. pylori) infection. The clinical outcomes of H. pylori infections have been shown to be 3

influenced by various genetic factors and induced the expression of pro-inflammatory cyclooxygenase enzyme (COX-2), which shows up-regulated expression in gastric cancer 8. Some studies have demonstrated the importance of genetic and epigenetic alterations of oncogenes, tumor suppressor genes, and mismatch repair genes in the development of gastric cancer. Compared with other more extensively investigated cancers, such as breast, prostate, and colon carcinomas, the molecular mechanisms involved in the transformation and progression of gastric cancer are poorly characterized. In spite of improvements in early diagnosis and multimodality treatment to this disease, patients at advanced stage frequently have poor prognosis because of the high rate of metastasis 9. About 60% of gastric cancers have lymph node metastasis, one of the most important prognostic factors 10. In addition, preoperative staging has low sensitivity for discrimination in clinical characteristics including lymph node metastasis, and controversies remain about the choice of therapeutic regimens 11-13. Thus, the identification of molecular prognostic markers bearing information on the cancer progression will help to provide a more effective therapeutic approach to the patients with gastric cancer. Lymph node metastasis The extent of lymph node (LN) metastasis is a major determinant for the staging and the prognosis of most human malignancies and often guides therapeutic decisions. The prognosis of gastric cancer patients is also affected by the presence of LN metastasis 14-16. Indeed, lymph node metastasis is the most significant prognostic factor for gastric cancer patients 17. Although the clinical significance of LN involvement is well documented, little has been known about the molecular mechanisms that 4

promote tumor spread via lymphatic vessels to regional and distant LNs. The early step of lymph node metastasis is considered to be intravasation of the cancer cells into lymphatic vessels which are newly induced by lymphangiogenic factors secreted by cancer cells. In that order cancer cells migrate into the lymph nodes, and proliferate there to form a metastasis 18, 19. Clinically, a trustworthy indicator is required to predict the presence of LN metastasis. Furthermore, molecular predictors of LN metastasis can be a prognostic marker and be of a potential therapeutic target. Lymphangiogenesis Lymphaniogenesis is a dynamic process during embryogenesis while is largely absent under normal physiological postnatal conditions 20. Indeed, in adults, lymphangiogenesis only takes place during certain pathological conditions such as inflammation, tissue repair, and tumor growth 21. Under pathological conditions, a major contribution has been established for the proliferation and sprouting of new vessels from preexisting lymphatic vessels. The relative contribution to new vessels from circulating endothelial progenitor cells remains unclear. The identification of several key lymphaticspecific molecular markers and factors that promote lymphatic vessel growth has propelled our understanding of the lymphatic vasculature in both physiological and pathological situations 22. The first and most comprehensively studied, pro-lymphangiogenic factors identified are vascular endothelial growth factor (VEGF)-C and VEGF-D, which bind to a tyrosine kinase receptor, VEGF receptor 3, expressed on the lymphatic endothelium 23. Lymphangiogenesis, the growth of new lymphatic vessels, was found to be critically required for solid tumor growth, invasion and distant lymph node metastases 24, 25. Lymphatic vessels are tubular structures composed of mono-layered 5

lymphatic endothelial cells (LECs) 26, 27. The migration and proliferation of LECs contributes to the sprouting of lymphatic vessels 28. VEGF-C Vascular endothelial growth factor (VEGF)-C is a member of the VEGF family of growth factors, which are conserved secreted glycoproteins that induce vasculogenesis, angiogenesis, lymphangiogenesis and are implicated in many physiological and pathological processes 23. The VEGF family is comprised of VEGF-A,-B,-C,-D and E. Of the three VEGF tyrosine kinase receptors identified thus far (i.e. VEGFR-1,-2 and -3), VEGFR-1 binds VEGF-A and VEGF-B, VEGFR-2 binds VEGF-A,-C,-D and E, and VEGFR-3 binds VEGF-C and VEGF-D. VEGFRs differ with respect to mechanisms of regulation and patterns of expression. For example, VEGFR-1 and VEGFR-2 are expressed almost exclusively by vascular endothelial cells and hematopoietic precursors, whereas VEGFR-3 is widely expressed in the early embryonic vasculature but becomes restricted to lymphatic endothelium at later stages of development and in postnatal life. VEGF-C displays a high degree of similarity to VEGF-A, including conservation of the eight cysteine residues involved in intra- and intermolecular disulfide bonding. Like VEGF-A, both human and murine VEGF-C are alternatively spiced. In addition, VEGF-C mrna is first translated into a precursor from which the mature ligand is derived by cell associated proteolytic processing after secretion. The post-secretion processing also allows VEGF-C to bind to VEGFR-2. Processed VEGF-C induces endothelial cell proliferation and migration, as well as increased vascular permeability. However, the respective roles of VEGFR-2 and -3 in mediating the biological effects of VEGF-C are incompletely understood. Unlike VEGF-A, 6

VEGF-C expression does not appear to be regulated by hypoxia. Based on its expression profile and its binding to VEGFR-3, VEGF-C has been implicated in the development of the lymphatic system 29. Molecular mechanisms of tumor-associated lymphangiogenesis in cancer have been studied and identified VEGF-C and VEGF-D widely accepted as a key molecules of lymphangiogenesis that also play a critical role via VEGF receptor-3 (VEGFR3) on the surface of lymphatic endothelial cells (LECs) 20, 30-32. VEGF-C and VEGF-D promote tumor-associated lymphangiogenesis and lymph node metastasis in vivo models 33-35. Besides, VEGF-C and VEGF-D are often expressed in primary human tumors: they are secreted by tumor cells, tumor associated fibroblasts and immune cells 36-38. It is reported that gastric cancer patients with high expression of VEGF-C protein had a poor prognosis than did those in low VEGF-C expression group 39, 40. Further, VEGF-C binding to VEGFR-3 induces lymphatic vessel development and lymph node metastasis 41. A significant correlation between lymph node metastasis and VEGF-C expression has been reported in gastric cancer 39, 42. Other growth factors reported to be lymphangiogenic are fibroblast growth factor-2 (FGF-2) 43 and platelet-derived growth factor B (PDGFB) ( 44. The lymphangiogenic effect of FGF-2 occurs to be indirect via VEGF-C and D. Gastric cancer cells produce PDGFB that is a regulator of lymphangiogenesis in gastric cancer 44, 45. In addition, Angiopoietin-2 is essential for establishing the lymphatic vasculature 46, 47. VEGF-C, D/VEGFR-3 is a key for primary proliferation of lymphatic vessels, whereas angiopoietin-2 is crucial in later remodeling stages 43, 48. 7

FoxO3a Forkhead box (FOX) proteins are superfamily of evolutionarily conserved transcriptional factors which play important roles in both normal biological processes and cancer development 49-51. FOXO transcription factor members include FOXO1, FOXO3, FOXO4 and FOXO6 52-54. These FOXO factors contribute to the regulation of various processes such as cell cycle regulation, cell size determination, apoptosis, differentiation, resistance to stress, DNA damage repair and energetic metabolism 55-59. Recently, increasing FOXO subfamilies have been suggested to be involved in carcinogenesis and cancer metastasis 60-62. Of note, oncogenic growth factor signaling pathways regulate FOXO function by phosphorylation 63. FOXO transcription factors are directly phosphorylated by PI3K/AKT signaling pathway, resulting in their sequestration in the cytoplasm 64-67. The FOXO1, FOXO3 and FOXO4 knockdown mouse has been shown to develop lymphomas in vivo 68. Among FOXO transcription factors, FoxO3a was shown to be associated with tumor suppression activity and angiogenesis 69, 70. Also FoxO3a inhibition promotes cell transformation, angiogenesis and tumor progression 71, 72. In addition, ERK down regulates FoxO3a expression by directly phosphorylating FoxO3a at Ser294, Ser344 and Ser425, which leads to cell proliferation and tumorigenesis 73-75. Aforementioned, FOXO functions as a tumor suppressor in cancer 76. Loss of function of FOXO is a pivotal event in tumorigenesis 57, 77. FOXO proteins contain a conserved DNA-binding domain (Forkhead domain) which recognizes the consensus sequence, TTGTTTAC 78-80. An active form of FOXO translocates to the nucleus, binding to cognate DNA sequence and partner proteins, transcriptional co-activators or co-repressors, regulate the transcription of multiple target genes involved in tumor suppression 81-84. 8

FoxO3a binds to a consensus Forkhead response element (FHRE) in the VEGF promoter and then recruits histone deacetylase2 (HDAC2) to the promoter leading to decreased histones acetylation, and concomitant transcriptional inhibition of VEGF. FoxO3a dependent repression of VEGF in breast cancer cells was recently reported 85. FoxO3a is a crucial downstream effector of the PI3K/AKT signaling pathway 86. Phosphorylated by AKT, FoxO3a is mislocalized in cytoplasm and cannot function as a tumor suppressive transcription factor in cancer cells 62, 87. PI3K/AKT pathway The phosphatidylinositor-3-kinase (PI3K)/AKT signaling pathway is one of the critical signaling cascades that is estimated to be present in >30% of various types of human cancers, and plays a key role in the extracellular growth factor stimulation to various cellular processes, including cell proliferation, survival, migration, genomic instability, angiogenesis and metastasis 88. Activation of PI3K is triggered by growth factor binding to receptor tyrosine kinases (RTKs). PI3Ks are lipid kinases that consist of three different classes (classⅠ, Ⅱ and Ⅲ) according to their structures and mechanism of activation. ClassⅠ PI3Ks are composed of a catalytic subunit and a regulatory subunit. PI3K enzyme acts on membrane PI to generate the second messenger lipid PI-3,4,5-triphosphate. PI- 3,4,5-triphosphate recruits phosphatidylinositol-dependent kinase 1 and AKT kinase to the membrane. AKT is a serine-threonine kinase that is regulated mainly following activation of the second messenger phospholipid kinase PI3K. The AKT kinase remains a subject of diverse investigations due mainly to its critical roles in a variety of pathways and cellular processes. AKT is activated through phosphorylation of Thr308 and Ser473 67. 9

In clinical trials, the number of drugs that target proteins involved in this pathway. For example, flavonoid derivative LY294002 is a PI3K inhibitor that acts in the ATP-binding site of PI3K enzyme and targets the PI3K/AKT axis. Together with the previous knowledge and the notion that lymphangiogenesis and angiogenesis in cancer are interrelated processes, we hypothesize that tumor suppressive transcription factor FoxO3a would suppress lymphangiogenesis in gastric cancer cell line models. 10

Ⅱ. MATERIALS AND METHODS 1. Cell culture The human gastric cancer cells SK4 and AGS were cultured in RPMI 1640 medium (Hyclone, South Logan, Utah)) containing 10 % fetal bovine serum, 100 U/ml of penicillin sodium and 100 μg /ml of streptomycin sulfate at 37 in a humidified incubator containing 5 % CO 2. Cultured human lymphatic endothelial cells (HLECs) were purchased from Promo Cell (Promo Cell, Heidelberg, Germany). This cell line was maintained in complete medium (Endothelial cell Growth medium 1; Promo Cell, Heidelberg, Germany) on gelatin-coated dishes. HLECs were used between passages 5 and 8. 2. Establishment of stable FoxO3a Knockdown cells For depletion of FoxO3a, short-hairpin RNAs (shrnas) targeted to FoxO3a were purchased from Origene Technologies (Origene, Suite 200, Rockville, USA). Knockdown vector of FoxO3a was transfected into SK4 cells. SK4 cells were grown at 80% confluence and then transfected with 4 μg knockdown vector of FoxO3a or Nontarget vector (Origene, NT-TR30013) for 24h using the TransIT- 2020 transfection reagent (Mirus Bio, Madison, WI53711, USA)following the manufacturer s instruction. Stably transfected cell lines were selected using 4 μg /ml Puromycin (Merck, Darmstadt, Germany). Stably transfected cells were harvested when plates were 80-90% confluent. Subsequently, stable knockdown cells, designated as SK4-shFoxO3a, were confirmed by western blot used in the 11

following experiments. 3. Establishment of stable FoxO3a-Overexpression cells For overexpression of FoxO3a, human green fluorescent protein-foxo3a construct was obtained from Origene Technologies (Origene, Suite 200, Rockville, USA). Overexpression vector of FoxO3a were transfected into AGS cells, AGS cells at 70-80% confluence and then transfected with 4 μg of FoxO3a expression vector or empty vector (pcmv-ac-gfp) for 24h using TransIT-2020 transfection reagent (Mirus Bio, Madison, WI53711, USA) following the manufacturer s instruction. Stably transfected cell lines were selected using 400 μg /ml G418 (Geneticin, an analog of neomycin). Stably transfected cells were harvested when plates were 80-90% confluent. Subsequently, stable overexpression cells, designated as AGS-FoxO3a, with successful optimization of gain of expression were determined with Western blotting. 4. sirna transfection To silence FoxO3a gene expression, SK4 cells were plated in 6-well plates at 1X10 5 cells for 16 hour before transfection. sirna complexes, prepared by incubating 25nM of indicated non-targeted RNA (NT) and FoxO3a sirna (Bioneer, Daejeon, Korea) with 8 μl X-tremeGENE sirna transfection reagents (Roche, Werk Penzberg, Germany) for 20min, were added slowly to the cell plates in a final volume of 2ml. After treatment for 7 hour with the sirna, the medium contacting sirna and transfection reagents was removed and cells cultured for 48h with 10%FBS RPMI1640 fresh culture medium. Gene knockdown was assessed by western blotting at 48 hours after sirna transfection. 12

FoxO3a small interfering RNA (sirna) and non-targeted RNA (NT) were purchased from Bioneer (Bioneer, Daejeon, Korea). sirnas with the following sense and antisense sequences were used: FoxO3a construct 1, 5 -GACGAUGAUGCGCCUCUCU-3 (sense) 5 -AGAGAGGCGCAUCAUCGUC-3 (antisense); FoxO3a construct 2, 5 -CAGUUCUAACUUCACUGUU-3 (sense) 5 -AACAGUGAAGUUAGAACUG 3 (antisense) Non-target construct, 5 -CCUACGCCACCAAUUUCGU-3 (sense) 5 -ACGAAAUUGGUGGCGUAGG-3 (antisense) 5. ELISA SK4 and AGS cell lines were seeded at 2X10 5 cells/well of a 10cm 2 dish and cultured for 24, 48, 72 hour. The amounts of VEGF-C in the cell medium were estimated using VEGF-C human ELISA kit (Abcam, Cambridge, UK) according to the manufacturer s instructions. For each experiment, triplicate samples were measured for statistical significance. 6. Tube formation assay HLECs (1X10 5 ) were cultured in a 24well plate coated with 150 μl Growth factor reduced Matrigel in MV1 when cell attachment for 1 hour, the MV1 medium was replaced with conditioned medium and continues cell culture for 24 hour. Tube length was quantified after 8 hours by measuring the total cumulative tube length in 3 random microscopic fields with a computer-assisted microscope using the program Image J. The original magnification used was X100. 13

7. Migration assay HLECs (10 6 /well) were seeding in MV1 media on 24 well plates. When the cells confluence 100%, we scratch on cells using tips and the media were removed from the cells and replaced with conditioned media from cultured media of gastric cancer cells. Migration of HLECs was determined after 12, 24 hour incubation. 8. Reverse transcription polymerase chain reaction (RT-PCR) Total cellular RNA was extracted from gastric cancer cells using the Trizol reagent according to the manufacturer s protocol, and PCR was performed using the Onestep RT-PCR kit (intron, Gyeonggi-do, Korea). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. PCR amplification was performed using the oligonucleotide primers of FoxO3a sense 5 - AGAGCTGAGACCAGGGTAAA-3, antisense 5 -GACAGGCTTCACTACCAGATTC-3 ; VEGF- C sense 5 -AGGCCACGGCTTATGCAA-3, antisense 5 -TAGACATGCATCGGCAGAA-3 ; GAPDH sense 5 -GTCAGTGGTGGACCTGACCT-3, antisense 5 - TGTTGAAGTCAGAGGACACC-3, with 35 cycles of 30s at 94, 30s at 55, and 2min at 72. PCR products were resolved on 1.5% agarose gels, stained with ethidium bromide, and photographed. 9. Quantitative Real-time PCR Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer s protocol, and real-time PCR reactions were conducted in an illumina Eco Real-time PCR system (Illumina, San Diego, CA). Complementary DNA (cdna) was synthesized using the M- 14

MLV Reverse transcriptase (m.biotech, Hanam, Korea). Real-time PCR was performed on 2X QuantiSpeed SYBR No-Rox kit (Philekorea,Deajeon, Korea). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. Primers used in real-time PCR were as follows: FoxO3a sense 5 -AGAGCTGAGACCAGGGTAAA-3, antisense 5 - GACAGGCTTCACTACCAGATTC-3 ; VEGF-C sense 5 -TTCCACCACCAAACATGCAG-3, antisense 5 -GGGACACAACGACACACTTC-3 ; GAPDH sense 5 - GTCAGTGGTGGACCTGACCT-3, antisense 5 -TGTTGAAGTCAGAGGACACC-3. Relative expression of FoxO3a and VEGF-C mrnas was given as FoxO3a/GAPDH and VEGF- C/GAPDH. The thermal profile for PCR was 95 for 3min, followed by 40 cycles of 95 for 10s and 60 for 30s. Thermo cycling was carried out in a final volume of 10 μl containing 1 μl of a cdna sample. Each sample was run in triplicate. The melting curve of each tube was examined to confirm a single peak appearance. 10. Western blot analysis Gastric cancer cells were collected and lysed on ice for 30 min in lysis buffer [1% Triton X-100, 50 mmol/l HEPES (ph 7.5), 150 mmol/l NaCl, 25 mmol/l β-glycerophosphate, 25 mmol/l NaF, 5 mmol/l EGTA, 1mmol/L EDTA as a protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN)]. The lysates were clarified by centrifugation for 13,000 rpm for 30 min. Equal amounts of protein were loaded onto a sodium dodecyl sulfate polyacrylamide gel (12 % polyacrylamide) followed by electrophoresis at 100V for 3 hour and transferred to polyvinylidene floride membrane at 100V for 1 hour. Subsequently, PVDF membrane was incubated in TBS-T with 5% skim milk (blocking solution) 15

for 1 h at room temperature. The PVDF membrane was, respectively, incubated 4 overnight with the target primary antibody. Anti-FoxO3a (dilution 1:1,000; cell signaling Technology, Massachusetts, USA), Anti-phospho FoxO3a (dilution 1:1,000; cell signaling Technology, Massachusetts, USA), VEGF-C (dilution 1:1,000; cell signaling Technology, Massachusetts, USA) and anti-β-actin (dilution 1:1,000; Sigma-Aldrich, USA) antibodies were diluted in TBS-T (TBS/Tween 20: 2% skim milk). The appropriate secondary antibodies were applied (1:5000, horseradish peroxidase-conjugated antirabbit and anti-mouse) at room temperature for 1 hours. Labeled bands were detected by enhanced chemiluminescence (ECL; ThermoScientific, USA). 11. Electrophoretic mobility shift assay (EMSA) and super-shift assay The DNA binding activity of FoxO3a was confirmed with a 32 P-labeled oligonucleotide containing FoxO3a transcription factor binding sites found in the VEGF-C promoter region. The DNA-protein binding detection kit (Promega, Madison, WI, USA) was used with modifications. In brief, DNAbinding reactions were carried out in a final volume of 25μl of buffer containing 10mM Tris (ph 7.5), 100mM NaCl, 1mM DTT, 1mM EDTA, 4% (w/v) glycerol, 0.1 mg/ml sonicated salmon sperm DNA, 15μg of nuclear extract, and oligonucleotides. Oligonucleotides containing consensus VEGF-C (IDT, SanDiego, CA, USA) was end-labelled to a specific activity of 5ⅹ10 5 CPM with γ-[ 32 P]-ATP and T4- polynucleotide kinase, followed by purification on a Nick column (GE Healthcare, Piscataway, NJ). Reaction mixtures with radio-labelled oligonucleotides were incubated at room temperature for 20 minutes, and resolved on 6.5% non-denaturing polyacrylamide gels after addition of 3 μl bromophenol blue (0.1%). Gels were dried and subjected to autoradiography. For super-shift assays, 3 16

μg of antibody (Ab) was added for 20 minutes at room temperature after the initial incubation. Abs specific for were purchased from Cell Signaling Technology. 12. Immunofluorescence assay Cells were washed with phosphate-buffered saline and fixed with 4% paraformaldehyde for 10 min. Nuclei were premeabilized by treatment with 0.5% Triton X-100/phosphate-buffered saline for 5 min. Cells were blocked in 5% BSA/phosphate-buffered saline for 20 min and incubated with rabbit Foxo3a (Cell Signaling Technology, Danvers, MA, USA) for 1 h at room temperature. Subsequently, cells were incubated with Alexa Fluor 488 anti-rabbit secondary antibody (Invitrogen, Carlsbad, CA) for 1 h at 4 (1:2000). Images were collected using a confocal microscope (LSM Meta 700, Carl Zeiss, Oberkochen, Germany) and were analyzed using the Zeiss LSM Image Browser software program, version 4.2.0121.The intensity of nuclear staining was determined by using the area measurement 13. Chemicals PI3K inhibitor, LY294002, was obtained from Invitrogen (Carlsbad, CA, USA) and dissolved in DMSO at a concentration of 50mM and stored at -20 until used. 14. Reverse Phase Protein Array (RPPA) For RPPA assays, cells were lysed in lysis buffer (1% Triton X-100, 50mM HEPES, ph7.4, 150mM NaCl, 1.5mM MgCl2, 1mM EGTA, 100mM NaF, 10Mm Na Pyrophosphate, 1mM Na3VO4, 10% 17

glycerol, 1mM PMSF, and 10ug/mL aprotinin). Samples were incubated on ice for 30min and centrifuged at 13,000rpm. Protein concentrations in the resulting supernatants were determined by the BCA method (Pierce, Rockford, IL). After dilution to 1mg/mL concentration, lysates were boiled in 1%SDS. Five serial dilutions were prepared for each lysate on 96well plates using additional lysis buffer with 1% SDS, and samples then transferred to 384well plates. Protein was then spotted onto nitrocellulose-coated glass slides (FAST Slides, Schleicher & Solutions Inc.,). Slides were stored at - 20 after printing. To detect protein levels, the RPPA slides were first blocked for endogenous peroxidase, avidin, and biotin protein activity and then incubated with primary and secondary antibodies. Antibodies were purchased from Cell Signaling Inc., (Danvers, MA, USA). The antibody signals were amplified using a Dako cytomation catalyzed detection system. 15. Statistical Analysis Student s t-test was used to evaluate the data. P values of <0.05 were considered statistically significant. 18

Ⅲ. RESULTS 1. Phosphorylated FoxO3a is prognostic factor in gastric cancer patients The Forkhead box class O (FoxO) transcription factors are implicated in a broad array of cellular functions including cell differentiation, apoptosis and DNA damage 89. Also, down regulation of FoxO proteins are associated with cancer progression and tumorigenesis 62. To find out for prognostic factors in gastric cancer patients, we performed RPPA in tissue of 673 gastric cancer patients. Cox univariate analysis for proteins related to survival revealed an inverse correlation between the phosphorylated FoxO3a, inactive form of FoxO3a, and prognosis of gastric cancer patients. The cluster analysis of RPPA was presented as a heat map (Figure 1). 19

Figure 1. Exploring prognostic proteins using RPPA in gastric cancer patients tumors. The result of reverse phase protein array (RPPA) was presented as a heat map. The color represents the expression level of the protein. Red represents high expression, while green represents low expression. Cox univariate analysis of protein expression profiling showed that proteins related to critical cancer pathways are prognostic in 673 gastric cancer patients. Among these, p-foxo3 was a significant poor prognostic factor for gastric cancer. 20

2. FoxO3a and VEGF-C expression showed a negative correlation in SK4 and AGS cell lines To select model cell lines, we first examined the expressions of FoxO3a and VEGF-C in 7 gastric cancer cell lines (MKN28, SK4, AGS, MKN45, NCIN87, KATOⅢ, HS746T) by RT-PCR and immunoblotting (Figure 1A,C). FoxO3a mrna was variously expressed in 7 gastric cancer cell lines examined whereas VEGF-C mrna was expressed significantly only in AGS cells. For quantitative measurement, Real time RT-PCR was performed in SK4 and AGS cells. The expression levels were normalized against the house keeping gene GAPDH. There was an inverse relationship between FoxO3a and VEGF-C in gastric cancer cell lines of SK4 and AGS cells (Figure 1B). Based on these results, we selected SK4 and AGS as model cell lines for downstream analysis in this study. 21

(A) (B) (C) 22

Figure 2. FoxO3a and VEGF-C expressions in gastric cancer cell lines (A) FoxO3a and VEGF-C mrna levels in 7 gastric cancer cell lines were analyzed by RT PCR. Total RNA (1ug) isolated using the Trisol was reverse transcribed using the SuperscriptⅢ reverse transcriptase and random primers, and the resulting first strand cdna was used as template in the RT- PCR. GAPDH was used as loading controls. All experiments were performed in triplicate. (B)Total RNA was extracted from these cells and analyzed for FoxO3a and VEGF-C mrna expression using real-time PCR (qrt-pcr). The expression of the FoxO3a gene in SK4 cell was high than AGS cell. In contrast, the expression of the VEGF-C gene in AGS cell was high than SK4 cell. (C) FoxO3a protein levels were analyzed in 7 gastric cancer cell lines by western blot assay using anti-foxo3a antibody. β-actin level was used as loading controls. The experiment was repeated three times with reproducible results. 23

3. Secreted VEGF-C levels and tube formation efficiency of human lymphendothelial cells (HLECs) in conditioned media of AGS were higher than those in SK4 cells Lymphangiogenesis, the sprouting of new lymphatic vessels from pre-existing ones, and the permeability of lymphatic vessels are regulated by vascular endothelial growth factor (VEGF)-C via its known receptors VEGFR-3 (Flt-4) 90. Because VEGF-C is a soluble factor that can be released from gastric cancer cells, we next investigated whether the model cell lines, SK4 and AGS cells, secret VEGF-C and thereby tube formation of HLECs being stimulated. To this end, we examined the secreted VEGF-C from the conditioned media of SK4 and AGS cells, using ELISA and immunoblotting. We measured secreted VEGF-C expression in conditioned media of SK4 and AGS cells on day 1, 2 and 3 (Figure 3A). We then performed western blot analysis of secreted VEGF-C protein with conditioned media at 72hour time point in SK4 and AGS cells (Figure 3B). Higher proliferation and tubule formation efficiency were observed when HLEC was cultured in the conditioned media of AGS than of SK4 cells (Figure 3C, D and E). The results suggest that AGS cells with low-level FoxO3a expression secreted high level of VEGF-C thereby promoting HLEC proliferation and tubule formation more efficiently than SK4 cells. 24

(A) (B) (C) 25

(D) (E) Figure 3. Secreted VEGF-C expression and tube formation of HLECs in conditioned media of gastric cancer cell lines (A)The time course of the secreted VEGF-C expression in cultured media SK4 and AGS gastric 26

cancer cells. Conditioned media extracts were analyzed by VEGF-C ELISA kit. (B) Secreted VEGF- C expression in media of cultured gastric cancer cells were taken at 72h after seeding by western blot with media concentration method using a methanol. (C) The proliferation of HLECs with media of 72h cultured gastric cancer cells was observed in a 10 cm 2 dish. Cultured media of AGS cells induces HLECs proliferation more than SK4. (D) The tube formation of HLECs with media of 72h cultured gastric cancer cells was observed in a 24-well plate coated with growth factor reduced matrigel. (E)Tube length of tube formation quantified after 7hours by measuring the total cumulative tube length in 3 random microscopic fields with a computer-assisted microscope using the program Image J. The original magnification used was X100. Data were expressed as mean +SD, n=3. Significant differences (* P<0.05) were determined by t-test using Graphpad prism 5.0 software. 27

4. Silencing FoxO3a in gastric cancer cells enhances the lymphangiogenic activity of HLECs Based on the above results, we next investigated the effect of FoxO3a in gastric cancer cells on tube formation and migration of HLECs in vitro, which are crucial for tumor lymphangiogenesis. First, we examined the effect of FoxO3a suppression using sirna. SK4 cells, highly expressing FoxO3a, were transfected with 2 different constructs of FoxO3a sirnas. After transfection, media changed with MV1 media and after 72hours the cells and conditioned media were collected. FoxO3a protein levels were determined by western bot. (Figure 4A). Human lymphatic endothelial cells (HLEC) were seeded on a growth factor reduced matrigel and examined the tube formation capacity Compared with control, vector only or non-target groups, tube formation of HLECs were significantly promoted by the conditioned media of FoxO3a knock-down SK4 cells (Figure 4B, C). Capillary-like structures were measured under light microscopy after 8 hours. In addition, the conditioned media of FoxO3a knockdown SK4 cells induced migration of HLECs (Figure 4D, E). Taken together, these results suggest that knockdown of FoxO3a in SK4 cells promote lymphangiogenesis. 28

(A) (B) 29

(C) (D) 30

(E) Figure 4. Knockdown of FoxO3a in SK4 cells enhances lymphangiogenesis (A) FoxO3a was specifically down-regulated by FoxO3a sirna. SK4 cells were trypsinized and seeded on 6well plates at 2X10 5 cells per well. Cells were transfected with 25nM non-targeted sirna and 2 different constructs of FoxO3a sirnas for 7hour with 2% FBS RPMI1640 media. After 72h, the cells were collected and FoxO3a protein levels were determined by western blot. β-actin expression was monitored for normalization. (B) After transfection, media changed with MV1 media and after 72h the cells and cultured media were collected. The silencing of FoxO3a in SK4 cells profoundly induced tube formation of HLECs treated with corresponding conditioned media than in 31

non-targeted control sirna transfected cells. Images show tube formation 7 h after HLECs seeding on matrigel. The data represent the average from there independent experiments (error bars = SD). (C) The tube formation of HLECs quantified after 7hours by measuring the cumulative tube length in 3 random microscopic fields with a computer-assisted microscope using the program Image J. The original magnification used was X100. Data were expressed as mean +SD, n=3. Significant differences (* P<0.05) were determined by t-test using Graphpad prism 5.0 software. (D) Migration of HLECs with cultured media of knocked down FoxO3a SK4 cells or with the vector only control transfected cells. (E)Images of approximate same field were taken at 12 and 24 hour after seeding, respectively. Representative images of three fields were shown (n=3, 100x). 32

5. Overexpression of FoxO3a in gastric cancer cells inhibits the lymphangiogenic activity of HLECs Next, we investigated the effect of FoxO3a overexpression in gastric cancer cells on lymphangiogenesis activity of HLECs. AGS cells, which expresses very low level FoxO3a, were transfected with the FoxO3a expression vector. After transfection, media changed with MV1 media after 72hour the cells and conditioned media were collected. FoxO3a protein levels were determined by western blot. (Figure 5A). To investigate the effect of FoxO3a expression in cancer cells on lymphangiogenic activity, HLECs were seeded on a growth factor reduced matrigel. Compared with control, vector only or non-target groups, tube formation of HLECs were significantly repressed by the conditioned media of FoxO3a overexpressed AGS cells. Capillary-like structures were measured under light microscopy after 8 hours. To quantify lymphangiogenesis, the parameters analyzed were cumulative tube length using an image J (Figure 5B, C). In addition, the conditioned media culture of FoxO3a overexpressed AGS cells suppress migration of HLECs (Figure 5D, E). These results suggest that overexpression of FoxO3a in AGS cells repress lymphangiogenesis. Collectively, we found that FoxO3a expression in cancer cells has a negative effect on lymphangiogenesis in our model systems. 33

(A) (B) (C) 34

(D) (E) 35

Figure 5. FoxO3a overexpression in AGS cells represses lymphangiogenesis (A)Foxo3a was overexpressed in AGS cells by FoxO3a overexpression plasmid vector. AGS cells were trypsinized and seeded on 6well plates at 2X10 5 cells per well. Cells were transfected with 1 μg plasmid vector for 7hour with 2% FBS RPMI1640 media. After 72h, the cells were lysed and subjected to western blot analysis with FoxO3a antibody. β-actin expression was monitored for normalization. (B) After transfection, media changed with MV1 media and after 72h the cells and cultured media were collected. The FoxO3a overexpression in AGS cells profoundly reduced tube formation of HLECs treated with corresponding conditioned media than in vector only control transfected cells. Images show tube formation 7 h after HLECs were seeded on matrigel. The data represent the average from there independent experiments (error bars = SD). (C) The tube formation of HLECs quantified after 7hours by measuring the cumulative tube length in 3 random microscopic fields with a computer-assisted microscope using the program Image J. The original magnification used was X100. Date were expressed as mean +SD, n=3. Significant differences (*P<0.05) were determined by t-test using Graphpad prism 5.0 software. (D) The migration of HLECs with cultured media of FoxO3a overexpressed AGS cells or with the vector only control transfected cells. (E) Images of approximate same field were taken at 12 and 24hour after seeding, respectively. Representative images of three fields were shown (n=3, 100X). 36

6. FoxO3a represses VEGF-C expression in gastric cancer cells Together with the observation that knockdown of transcription factor FoxO3a in gastric cancer cells promote lymphangiogenesis in HLECs, we next investigated whether the FoxO3a in gastric cancer cells would transcriptionally regulate the expression of lymphangiogenic factors. A recent study showed that FoxO3a could bind to the VEGF-A gene promoter and inactivate transcription of VEGF- A 85. Interestingly, we found that the FoxO3a binding site, FHRE, is also present in VEGF-C gene promoter region. VEGF-C is the most representative molecule inducing lymphangiogenesis in cancer. Therefore, we set out to investigate whether FoxO3a regulate VEGF-C expression and thereby influencing tube formation efficiency of HLECs. To do this, secreted VEGF-C in the conditioned media of gastric cancer cells were detected using western blot with media concentration method and ELISA. We used FoxO3a shrnas to inhibit FoxO3a signaling in FoxO3a high expressing cells, SK4 cell. As a result, when FoxO3a was knocked down in SK4 cells, secreted VEGF-C was significantly increased (Figure. 6A). Also, when FoxO3a was knocked down or overexpressed using either sirna or plasmid vector respectively, FoxO3a and secreted VEGF-C expression showed a negative correlation in SK4 and AGS cells. Secreted VEGF-C was measured by specific VEGF-C ELISA (Figure 6B, C). These results indicate that FoxO3a repressed lymphangiogenesis of HLECs by transcriptionally downregulating VEGF-C. 37

(A) (B) (C) 38

Figure 6. Silencing of FoxO3a in SK4 enhances the VEGF-C expression (A)Western blot analysis showing inducible shrna knockdown of endogenous FoxO3a in SK4 cells. shfoxo3a line was seed at 10 5 cells per well in 6-well plates and cultured for 72 hour. At this time point, using a conditioned media, secreted VEGF-C expression was estimated by western blot with media concentration method using a methanol. (B) FoxO3a was specifically down-regulated by FoxO3a sirna in SK4 cells. Foxo3a was overexpressed in AGS cells by FoxO3a overexpression plasmid vector. SK4 and AGS cells were trypsinized and seeded on 6well plates at 2X10 5 cells per well. Cells were transfected with 2% FBS RPMI1640 media for 7hours. After 72h, the cells were collected and FoxO3a protein levels were determined by western blot. β-actin expression was monitored for normalization. (C) Secreted VEGF-C was measured in conditioned media by specific VEGF-C ELISA. 39

7. FoxO3a binds to the FHRE of VEGF-C promoter region To investigate whether FoxO3a directly binds to VEGF-C promoter region thereby repressing transcriptional activity, we next performed electro mobility shift assay (EMSA) to directly show that FoxO3a binds to the FHRE in promoter region of VEGF-C (Figure 7A). The probe corresponding to the VEGF-C promoter region was designed as described under Materials and Method. Our results showed that there was a shifted band when the probe was employed. When the five-fold nuclear protein was used in the test, the shifted band was induced significantly. The results suggest that FoxO3a binds to the FHRE of VEGF-C gene promoter (Figure 7B). Taken together, these results demonstrated that FoxO3a can repress the VEGF-C through binding to VEGF-C promoter region. 40

(A) (B) 41

Figure 7. FoxO3a binds to the FHRE of VEGF-C promoter region (A) Design of the probe including forkhead response element (FHRE) of VEGF-C promoter region. (B) Incubation of nuclear extract in SK4 and AGS with 32 P-labeled FoxO3a-VEGF-C sequence produced a DNA-protein band shift. To confirm the binding of FoxO3a to the FoxO3a-VEGF-C sequence, these EMSA reactions were further incubated with anti-foxo3a antibody. The addition of this antibody resulted in a super shifted complex in addition to the DNA-protein band. These data confirmed the presence of FoxO3a in the nuclear protein complex that binds the FoxO3a binding site of the VEGF-C promoter. 42

8. FoxO3a is inactivated by PI3K/AKT pathway The PI3K/AKT signaling pathway is activated in many human cancers, and plays a key role in the extracellular growth factor stimulation to various cellular processes, including cell proliferation, survival, and angiogenesis 88. PI3K downstream is the serine/threonine kinase AKT. AKT is activated through phosphorylation of Thr308 and Ser473 67. FoxO3a is a crucial downstream target of the PI3K/AKT signaling pathway 86. FoxO3a function is suppressed by PI3K/AKT pathway. When phosphorylated by AKT, FoxO3a is inactivated and cannot function as a transcription factor in cancer cells 62, 87. Specific small molecule inhibitors of PI3K (LY294002) were used to suppress endogenous PI3K pathway. To identify the FoxO3a regulation by PI3K/AKT pathway in gastric cancer cells, we examined the time course of phosphorylated AKT (p-s473) and p-foxo3a (p-s253) expressions in SK4 and AGS cells with LY294002 (50uM) by western blot. As a result, we observed that LY294002 treatment decreased phosphorylation levels of AKT and FoxO3a in SK4 and AGS cells. (Figure 8A). After 2 hours of treatment, we observed FoxO3a localization using immunocytochemistry (ICC) method. Of note, SK4 cells which express high level of FoxO3a at baseline showed minimal change in nuclear localization of FoxO3a after LY294002 treatment compared to AGS cells which showed dramatic increase in nuclear localization of FoxO3a upon LY294002 treatment (Figure 8B). These results suggest that FoxO3a nuclear localization that is dependent on phosphorylation is regulated by the PI3K/AKT pathway. 43

(A) 44

(B) 45

Figure 8. FoxO3a localization and PI3K/AKT activity in gastric cancer cells (A) SK4 and AGS cells were treated the PI3K-specific inhibitor LY294002 (50uM) for the indicated times. Expression of AKT and FoxO3a phosphorylated proteins were examined by western blotting of the total cell lysates and β-actin was used as the loading control. (B) SK4 and AGS cells were seeded in glass coated plates at 80% confluence and treated with LY294002 (50uM). After 2 hours of treatment, ICC shows nuclear localization of FoxO3a in SK4 and AGS cells. Hoechst was used to stain the nuclei. 46

9. FoxO3a expression is negatively correlated with lymph node metastasis in gastric cancer patients Lastly, we examined whether the expression of FoxO3a and phospho-foxo3a in gastric cancer patient tissues is correlated with lymph node metastasis. In RPPA data, we evaluated the quantified expression levels of both FoxO3a and phospho-foxo3a in relation to lymph node status. In accordance with the projected role of FoxO3a in lymphangiogenesis we explored in this study, there was an inverse correlation between FoxO3a expression and lymph node metastasis in gastric cancer patients (Table 1). 47

Table 1. FOXO3a expression in anti-correlated with lymph node metastasis in human gastric cancer tissue 48

Figure 9. Diagram illustrating the relationship between FoxO3a and lymphangiogenesis in gastric cancer Stimulation with growth factors leads to activation of the phosphatidylinositol-3 kinase (PI3K)/AKT signaling cascade. Phosphorylation of Foxo3a by AKT leads to their exclusion from the nucleus, and a reduction in the expression of key genes involved in protein degradation. Our data of gastric cancer cells revealed that Foxo3a nuclear localization is significantly but inversely associated with VEGF-C expression, suggesting FoxO3a negatively regulates VEGF-C expression. Signals mediated through VEGF-C and their receptors have been shown to be essential for gastric cancer carcinogenesis, cell migration, lymphangiogenesis and lymph node metastasis. 49

Ⅳ. DISCUSSION Lymph node metastasis is the major cause of relapse and death in gastric cancer patients. Thus, it is critical to identify reliable predictive markers and molecular targets for lymph node metastasis to improve clinical outcome of gastric cancer. The primary aim of this study was to determine the clinical significance and biological role of FoxO3a in gastric cancer progression, in particular lymphangiogenesis. We found that phosphorylated FoxO3a, indicative of inactivation of FoxO3a, increased in gastric cancer tissues and was correlated with poor prognosis. Moreover, high level of FoxO3a in patient tumor was significantly negatively correlated with lymph node metastasis. Since angiogenesis and lymphangiogenesis are interrelated processes and FoxO3a is already known to suppress VEGF-A, we speculated that FoxO3a might be involved in regulation of lymphangiogenesis as well. Together with the clinical tumor protein assay data indicating FoxO3a is inversely correlated with lymph node metastasis, we prompted to test the hypothesis that FoxO3a would suppress lymphangiogenesis. Functional studies revealed that overexpression of FoxO3a inhibited tube formation and migration of HLECs, while silencing of FoxO3a caused opposite effects. These data indicate that decreased FoxO3a in gastric cancer promotes lymphangiogenesis. Further, we found that VEGF-C, a major regulator of lymphangiogenesis, has a FHRE in the promoter region that consolidated the hypothesis. We revealed that FoxO3a binds to the FHRE in the promoter region of VEGF-C through EMSA. Further, gene silencing of FoxO3 using RNA interference increased the expression of VEGF-C in gastric cancer cells. Together with EMSA results, these data indicate that FoxO3a could transcriptionally repress VEGF-C expression in gastric cancer cells to decrease the tube formation and migration of HLECs. According to the recent study, FoxO3a suppresses VEGF-A 50