Microsoft Word doc

VEGF, IL-8, or c-met-specific short hairpin RNA-expressing oncolytic adenovirus elicits potent inhibition of angiogenesis and tumor growth Ji Young Yoo Department of Medical Science The Graduate School, Yonsei University

VEGF, IL-8, or c-met-specific short hairpin RNA-expressing oncolytic adenovirus elicits potent inhibition of angiogenesis and tumor growth Directed by Professor Chae-Ok Yun The doctoral Dissertation submitted to the Department of Medical science, the Graduate School of Yonsei University in partial fulfillment of the requirements for the degree of Doctor of Medical Science Ji Young Yoo December 2007

This certifies that the doctoral dissertation of Ji Young Yoo is approved Thesis Supervisor : Chae-Ok Yun Joo-Hang Kim Hoguen Kim Hyun Cheol Chung Jaemyeon Lee The Graduate School Yonsei University December 2007

ACKNOWLEDGEMENTS 식물단백질을전공한제가바이러스를이용한암유전자치료에대한막연한동경심으로두려움과설렘으로시작한박사과정이엊그제같은데, 벌써 4년이라는시간이지나부족하지만작은결실을얻게되었습니다. 때로는많이힘들기도지치기도했지만, 논문을완성하고보니고마운분들이너무나많은저는무척이나행운아였던것같습니다. 그분들께이지면을빌어짧게나마감사의뜻을전하고자합니다. 먼저연구에대한기본적인자세와열정, 실험에대한진정한재미와의미를알게해주시고, 다양한연구를할수있는많은기회들을주시고, 무엇보다도열심히하시는모습으로모범이되어주신윤채옥교수님, 항상환한웃음으로격려해주시고, 실험적결과에대해임상적관점에서예리하게짚어주시는따뜻한김주항교수님께깊은감사를드립니다. 제가가진그릇이 10이고, 아무리노력해도 20밖에될수없었을제가, 두분의제자가되어 50이되었다고생각합니다. 더많은노력으로나머지 50을채워나가당당한제자로써교수님의가르침에보답하겠습니다. 그리고바쁘신와중에도부족한이논문의완성을위해꼼꼼하게심사해주신김호근교수님, 정현철교수님, 이재면교수님께깊이감사드립니다. 석사를마치고, 취직을하겠다했을때, 공부하라야단쳐주시고, 서울오시면술사주시며한결같이격려해주시고응원해주시는, 보답해야할게너무많은존경하고사랑하는이상렬교수님께너무나감사드립니다. 다정다감하신손주혁선생님, 열정적이신조병철선생님, 언니처럼다정한최혜진선생님, 바쁘단핑계로연락도자주못드리지만항상먼저챙겨주시고맘써주시는김현희선생님, 친구같은영숙언니, 똑소리나는은희언니, 동물실험의대가황경화선생님, 자기의꿈을위해성실하게노력하는윤아에게도감사의마음을전합니다. 진정한방장의모습을몸소보여주신최고의선배재성선배님, 진정한커피한잔의여유진선언니, 만능맥가이버김인욱선생님, 나의든든한정신적지주민정언니, 열정적이고적극적인경주, 꼼꼼한평환이, 첨엔삐걱거렸으나지금은누구보다편안한친구지훈씨, 애교많은아름이, 곰돌이엄마정선이, 착하고성실한일규, 나의주말파트너송남이, 매력만점첫제자민주, 성실하고반듯한오

준이, 팔방미인성미, 실험배우느라열심인새내기 7 차원지성이와남동생혜원이 에게고맙다는말을전합니다. 가족들보다더많은시간을함께했던실험실선후 배님들께깊은감사를드립니다. 석사시절처음파이펫잡는것부터때로는따끔한충고로, 때로는따뜻한격려로, 진정한실험의재미를알게해준멋진교수님이되신균오선배, 선배만나일복이많아졌다투정부리면서도, 선배가하던실험자세와후배들에게잔소리하는것까지닮아버린저를보며, 너무나많은것을배운것같아항상감사한마음입니다. 세심하게잘챙겨주는중로선배, 큰오빠용훈선배, 작은오빠승식선배, 든든한수권선배, 새신랑연옥선배, 나의든든한석사동기정찬선배, 멋진호희언니, 미스센터진호선배 단백질방선후배님들께깊은감사를드립니다. 바쁘단핑계로연락도않는나를항상걱정하고챙겨주는소중한나의친구들, 기원이, 종숙이, 명희, 의경이, 명숙이, 멀리있어도그존재만으로도큰힘이되는사랑하는친구들현미, 정미, 희영, 미애, 같은꿈을갖고같은길을준비하는든든한친구이박사정순이, 떨어져있어도매일같이전화로나의얘기잘들어주고때로는따뜻하게안아주고, 때로는냉정하게잔소리해주는소중한친구주영이, 만난지 5년이지만가까이있어오래된친구들보다더많은시간을함께하는나의든든하고고마운소중한벗연호에게고마운맘을전합니다. 마지막으로나에게는어리게만보이는철부지막내였지만이제는한가족의가장이된막내지태와올케, 그리고두아이의엄마가되어이제는언니같은동생지현이와제부, 공부하느라자주만나지도놀아주지도못하는고모, 이모를잘따라주는조카들, 그리고여러가지힘든일들이많았지만, 많이부족한큰딸걱정할까말하지않고, 모든일들을이겨내고, 믿고공부에만전념할수있도록사랑과용기를주시는부모님께말로는표현하기힘든감사하는마음과이작은결실을바칩니다. 이러한감사의마음잊지않고언제나열심히할것을다짐하며, 여러분모두들행 복하시길바랍니다.

TABLE OF CONTENTS ABSTRACT 1 I. INTRODUCTION 6 II. MATERIALS AND METHODS 14 1. Cell lines and cell culture 14 2. Construction of expression plasmids expressing VEGF-specific shrna 15 3. Synthesis and transfection of sirnas specific for IL-8 or c-met 16 4. Generation of VEGF -specific shrna-expressing Ads 17 5. Generation of IL-8-specific shrna-expressing Ads 17 6. Generation of c-met-specific shrna-expressing Ads 18 7. Quantitation of VEGF, IL-8, c-met, and MMP-2 by ELISA 20 8. Tube formation assay 22 9. Ex vivo aortic ring sprouting assay 22 10. In vivo Matrigel plug assay 23 11. Migration assay 24 12. Matrigel invasion assay 24 13. Gelatin Zymography 25 14. MTT assay 25 15. Reverse transcription (RT)-Polymerase chain reaction (PCR) i

analysis 26 16. Cell cycle analysis 27 17. HGF specificity assay 28 18. Electron microscope (EM) cytology 28 19. Immunoblotting analysis 29 20. Assessment of anti-tumor effects in human xenograft model 29 21. Evaluation of tumor xenograft by histology and immunohistochemistry 30 22. Intratumoral microvessel density assessment 31 23. Statistical Analysis 31 III. RESULTS 1. Design of shrnas directed to the VEGF mrna 32 2. Construction and effects of shvegf-expressing replicationcompetent Ad on the expression of VEGF 34 3. Construction and effect of shvegf-expressing replicationincompetent Ad on the expression of VEGF 35 4. Ad- E1-shVEGF inhibits angiogenesis in vitro and in vivo 36 5. shvegf expression does not inhibit viral replication 43 6. Enhanced anti-tumor effect of shvegf-expressing oncolytic Ad 45 7. Antiangiogenic effects of VEGF-specific shvegf-expressing ii

oncolytic Ad, Ad- B7-shVEGF 47 8. Improved efficacy of oncolytic Ad-mediated sirna expression over replication-incompetent Ad-mediated irna 51 9. Identification of effective sirna sequences and generation of replication-incompetent Ads expressing shrna specific to IL-8 53 10. Comparison of U6 and CMV promoters in IL-8 knockdown in vitro 56 11. Effect of IL-8-specific shrna expression on the function of endothelial cells in vitro 58 12. Effect of IL-8-specific shrna expression on tumor cell migration, invasion, and MMP-2 expression 60 13. Generation and characterization of oncolytic Ad expressing shil-8 65 14. Oncolytic adenovirus expressing shil-8 inhibits tumor growth in nude mice 68 15. Identification of effective sirna sequences and generation of recombinant Ad expressing shrna specific to c-met 74 16. Comparison of c-met suppression by recombinant adenovirus expressing shmet4, shmet5,or shmet4+5 76 17. Reduced c-met inhibits cell proliferation through mitotic iii

catastrophe by senescence 79 18. Reduced c-met inhibits VEGF expression and accordingly the function of endothelial cells 85 19. Effect of Ad- E1-shMet4+5 on tumor cell migration, invasion, and MMP-2 expression 91 20. Ad- E1-shMet4+5 Suppresses Met signaling 94 21. Inhibition of HGF-dependent or -independent cell proliferation by shmet-expressing recombinant Ad 95 22. Enhanced anti-tumor effect of shmet-expressing recombinant Ad 97 23. In Vivo histologic and immunohistochemical characterization 99 24. Inhibition of tumor metastasis by shmet-expressing Recombinant Ad 103 IV. RESULTS V. CONCLUSION VI. REFERENCES iv

LIST OF FIGURES Figure 1. Design and characterization of vascular endothelial growth factor (VEGF)-specific small interfering RNAs (sirnas) 33 Figure 2. Quantification of VEGF 37 Figure 3. Characterization of replication-incompetent Ad expressing VEGF-specific shrna 40 Figure 4. VEGF-specific shrna-expressing Ads inhibit vascularization in the Matrigel plug assay 42 Figure 5. Cytopathic effects of shvegf-expressing oncolytic Ad 44 Figure 6. shvegf-expressing oncolytic adenovirus Ad- B7-shVEGF and the growth of established tumors and survival of mice 46 Figure 7. Antiangiogenic effects of VEGF-specific shrna-expressing oncolytic Ad, Ad- B7- shvegf, in U343 human glioma xenografts 49 Figure 8. Time-course and magnitude of the VEGF gene silencing effect of Ad- E1-shVEGF or Ad- B7-shVEGF Ad 53 v

Figure 9. Characterization of IL-8-specific small interfering RNAs (sirnas) and the structure of adenoviruses used in this study 55 Figure 10. Quantitation of IL-8 secreted by cells transduced with replication-incompetent Ads 59 Figure 11. Effects of Ad-mediated IL-8-specific shrna expression on migration, tube formation, and vessel sprouting of endothelial cells 61 Figure 12. Inhibition of cancer cell migration, invasion, and MMP-2 expression by Ad- E1-U6shIL8 64 Figure 13. Characterization of IL-8-specific shrnaexpressing oncolytic Ads 67 Figure 14. Effect of oncolytic adenovirus expressing IL- 8-specific shrna in vivo 70 Figure 15. Effect of oncolytic adenovirus expressing IL- 8-specific shrna in vivo 72 Figure 16. Ad- B7-U6shIL8 treatment inhibits tumor metastasis 73 vi

Figure 17. Characterization of c-met-specific small interfering RNAs (sirnas) and the structure of adenoviruses used in this study 75 Figure 18. Quantitation of c-met suppression in various cancer cells transduced with c-met specific shrna expressing Ads, Ad- E1, Ad- E1- shmet4, Ad- E1-shMet5, or Ad- E1- shmet4+5 78 Figure 19. Changes in the cellular morphologies of U343 cells transduced with Ads expressing c-met specific shrna, Ad- E1-shMet4, Ad- E1- shmet5, or Ad- E1-shMet4+5 83 Figure 20. Effects of Ads expressing c-met-specific shrna in endothelial cell functions 89 Figure 21. Inhibition of cancer cell migration, invasion, and MMP-2 expression by Ad- E1- shmet4+5 93 Figure 22. Inhibition of down signal pathway by c-met inhibition 95 Figure 23. HGF-dependent or -independent cell vii

proliferation inhibition by Ad- E1-shMet4+5 96 Figure 24. In vivo antitumor effect of Ads expressing the c-met specific shrna 98 Figure 25. Histological characterization of tumor tissues in U343 human glioma xenografts 102 Figure 26. Therapeutic efficacy of Ad- E1-shMet4+5 on MDA-MB-231 lung metasis tumor model 105 LIST OF TABLES Table 1. Sequences of the four IL-8-specific sirnas examined in this study 54 Table 2. Sequences of the four c-met-specific sirnas examined in this study 74 Table 3. Primer used for the analysis of the gene expression related to cellular senescence 82 viii

ABSTRACT VEGF, IL-8, or c-met-specific short hairpin RNA-expressing oncolytic adenovirus elicits potent inhibition of angiogenesis and tumor growth Ji Young Yoo Department of Medical science The Graduate School, Yonsei University (Directed by Professor Chae-Ok Yun) RNAi, due to its target specificity may be highly effective as a novel therapeutic modality but direct delivery of synthetic sirna still remains a major obstacle for this approach. To induce long term expression and specific gene silencing, novel delivery vector system is also required. To overcome this shortcoming, I constructed an oncolytic adenovirus (Ad)-based shrna expression system (Ad- B7-shVEGF) against vascular endothelial growth factor (VEGF), a key mediator in angiogenesis. To demonstrate VEGFspecific nature of this newly engineered Ad-based shrna, replication- 1

incompetent Ad expressing VEGF-specific shrna (Ad- E1-shVEGF) was also generated. Ad- E1-shVEGF was highly effective in reducing VEGF expression, and elicited anti-angiogenenic effect in vitro as well as in vivo. Similarly, Ad- B7-shVEGF exhibited potent anti-angiogenic effects in the matrigel plug assay in vivo. Moreover, Ad- B7-shVEGF also demonstrated enhanced antitumor effect and survival advantage compared to its cognate control oncolytic Ad, Ad- B7. Tumor histological analysis revealed that Ad- B7-shVEGF induced significant reduction in tumor vasculature, verifying the anti-angiogenic mechanism. Furthermore, the duration and magnitude of the gene silencing effect following infection with Ad- B7- shvegf was longer and more effective than the replication-incompetent Ad, Ad- E1-shVEGF. Taken together, these results suggest that the combined effects of oncolytic viral therapy and cancer cell-specific expression of VEGF-targeted shrna elicits greater anti-tumor effect than an oncolytic Ad alone. To select effective promoters for expression of shrnas, I used IL-8, a potent pro-angiogenic factor. I also manufactured replication-incompetent Ads (Ad- E1-CMVshIL8 and Ad- E1-U6shIL8) under the control of the CMV and U6 promoters, respectively. Ad- E1-U6shIL8 was highly effective in reducing IL-8 expression, and was much more effective in driving IL-8 2

specific shrna than the CMV promoter-driven vector. The reduced IL-8 expression then translated into decreased angiogenesis in vitro as measured by migration, tube formation, and rat aortic ring sprouting assays. In addition to its effect on endothelial cells, Ad- E1-U6shIL8 also effectively suppressed the migration and invasion of cancer cells. I also have generated an efficient oncolytic adenovirus (Ad)-based shrna expression system (Ad- B7- U6shIL8) against IL-8, a potent pro-angiogenic factor. In vivo, intra-tumoral injection of Ad- B7-U6shIL8 significantly inhibited the growth of Hep3B and A549 human tumor xenografts. Histopathological analysis of Ad- B7- U6shIL8-treated tumors revealed an increase in apoptotic cells and a reduction in mcro-vessel density. Finally, Ad- B7-U6shIL8 was also shown to inhibit the growth of disseminated MDA-MB-231 breast cancer metastases. Taken together, these findings demonstrate the utility and anti-tumor effectiveness of oncolytic Ad expressing shrna against IL-8. To develop effective shrna inhibition system, I compared single or dual shrna expression system. c-met, receptor tyrosine kinase for hepatocyte growth factor (HGF) is overexpressed and/or mutated a variety of human tumors. Since HGF-Met signaling contributes to tumor survival, growth, angiogenesis, and metastasis, various approaches have been explored to inhibit the function of HGF or Met. In this study, I generated recombinant 3

adenovirus expressing c-met-specific shrna and compared single and dual shrna expression system, too. All constructed c-met specific shrnaexpressing Ads inhibited c-met expression. Among these, dual shmetexpressing Ad, Ad- E1-shMet4+5, was more effective than single shrnaexpressing-ads in driving c-met specific shrna. Cells infected with shmetexpressing Ads were showed dramatic growth inhibition and characteristic changes in morphology. That is, phenotypes such as enlarged and flattened cell morphology, increased granularity and the appearance of many vacuolated cells were typical senescence-like phenotype and mrna level of the genes commonly associated with cellular senescence like SM22, TGase II, and PAI-I was increased. Also, it was observed that the reduced c-met expression could inhibit cancer cell proliferation by arresting cells at G2/M phase using cell cycle analysis. Reduced c-met expression down-regulated VEGF expression and reduced angiogenesis in vitro as measured by migration, tube formation, and rat aortic ring sprouting assays. In addition to inhibition of endothelial cell functioning, Ad- E1- shmet4+5 effectively suppressed the migration and invasion of cancer cells. Moreover, intra-tumoral injection of Ad- E1-shMet4+5 inhibited tumor growth significantly more than single shmet-specific shrna expresiing Ad, Ad- E1-shMet4 or Ad- E1-shMet5. Histopathological analysis of tumors treated with Ad- E1-shMet4+5 revealed 4

an inhibition of cancer cell proliferation and reduction in vessel density. Furthermore, treatment of Ad- E1-shMet4+5 inhibited the growth of disseminated MDA-MB-231 breast cancer metastases. Taken collectively, these findings demonstrate that the inhibition of c-met function by dual c-met specific shrna expressing Ad suppress cancer cell proliferation via senescence mechanism and tumor growth, invasion, metastasis and angiogenesis. Key words: Cancer gene therapy; vascular endothelial growth factor (VEGF), IL-8, c-met, Hepatocyte growth factor (HGF); short hairpin RNA (shrna), oncolytic adenovirus 5

VEGF, IL-8, or c-met-specific short hairpin RNA-expressing oncolytic adenovirus elicits potent inhibition of angiogenesis and tumor growth Ji Young Yoo Department of Medical science The Graduate School, Yonsei University (Directed by Professor Chae-Ok Yun) I. INTRODUCTION It has recently been shown that the introduction in a mammalian cell of double-stranded oligoribonucleotides (also called sirna) triggers the degradation of the endogenous mrna to which the sirna hybridizes to. Initial investigations of RNAi in cells relied on transfection with synthetic RNA oligonucleotides 1,2, or plasmids designed to drive expression of sirnas through the use of RNA polymerase III promoters 3-5. This knock down technology has been successfully applied to inhibit target gene expression but its utility is limited by its short half-life. To achieve therapeutic in vivo gene silencing in mammalian tissues, it would require intracellular transcription 6

expression rather than transient transfection of dsrna 6. One means to achieve this long lasting expression of sirna is to use a vector-based delivery system such as recombinant viral vectors. E1/E3-deleted replication-defective adenovirus (Ad) has been widely used for cancer gene therapy because it offers, in contrast to other vectors, much higher transduction efficiency and transgene expression in a broad spectrum of cell types 7. However, replication-deficient viral vectors have thus far been used with limited success in cancer gene therapy mainly due to limited transduction efficiency and short duration of therapeutic gene expression. It is thus expected that delivery of shrnas interfering with the expression of genes involved in tumor cell survival using non-replicating vectors will meet similar difficulties. Oncolytic Ads are being developed as selectively replicating antitumoral agents, and currently number of clinical trials with such viruses are ongoing to treat a variety of cancers 8-12. A clear benefit is the potential amplification of its effect in which the replicating vector would be able to infect and deliver the therapeutic gene to adjacent cancer cells, ultimately enhancing the potential of a viral-based therapy to deal with the complexity of a human tumor 13. Angiogenesis, the formation of new capillaries from existing blood vessels, plays an important role in the growth, metastasis, and malignancy of 7

tumors 14. Number of growth factors have been identified as positive regulators of angiogenesis 15. Among them, vascular endothelial growth factor (VEGF) seems to be the predominant growth factor found in a wide variety of conditions associated with angiogenesis 15,16. Therapeutic effect in terms of inhibiting tumor growth and metastasis by inhibiting VEGF activity or disabling the function of its receptors has been shown in the clinic 17-20. In this report, the main objectives were to construct an efficient oncolytic Ad-based shrna expression system and to explore oncolytic Admediated RNAi for efficient and long-term gene silencing. I show here that E1A- and E1B-double mutant oncolytic Ad expressing VEGF-specific shrna, Ad- B7-shVEGF, induces silencing of VEGF gene effectively in vitro as well as in vivo. I show for the first time that VEGF-specific shrna expressed from an oncolytic virus can induce potent anti-angiogenesis, resulting in tumor suppression as well as survival benefits. In addition, I demonstrate for the first time that the oncolytic Ad-mediated shrna expression results in improved efficacy as well as sustained gene silencing effect than the replicationincompetent Ad, Ad- E1-shVEGF. Interleukin-8 (IL-8), a member of the CXC chemokine family, was initially discovered as a leukocyte chemo-attractant, recent studies have revealed that IL-8 plays an important role as a potent pro-angiogenic factor 8

21,22. IL-8 is produced by various tumors including melanoma, lung, prostate, gastric, ovarian, and bladder cancers 23. Experimental and clinical studies have shown positive correlations between IL-8 expression and tumor growth and metastasis 24. Recently, it has been shown that a blockade of IL-8 activity using a neutralizing antibody or an IL-8 antisense oligodeoxynucleotides inhibited both the growth and metastasis of tumors in a variety of animal models 25-27. Using IL-8, I compared two promoters, CMV and U6, in their efficiency of shrna expression. I also investigated the potential of an oncolytic Ad-mediated shrna expression targeted to IL-8 in inhibiting both tumor growth and angiogenesis. Our results show that the U6 promoter is superior to CMV in its ability to express shrna specific to IL-8, and that the expression of IL-8-specific shrna can strongly and specifically silence IL-8 mrna and protein expression in various human cancer cell lines. I show for the first time that IL-8-specific shrna expressed from an oncolytic adenovirus can induce potent anti-angiogenic effects, resulting in inhibition of tumor growth and metastasis. I further show that E1A-mediated suppression of angiogenesis and induction of apoptosis in tumors likely contributes to additional anti-tumor activity, leading to enhanced therapeutic efficacy. Findings in this report strongly suggest that the use of cancer cell-specific 9

replicating oncolytic adenovirus in the delivery of IL-8 specific shrna may hold strong promise for the treatment of cancer. Receptor tyrosine kinases (RTKs) regulate many key processes in mammalian cell growth and survival, organ morphogenesis, neovascularization, and tissue repair and regeneration, especially. Overactivation and/or defective downregulatiuon of RTKs have been implicated as causative factors in the development and progression of numerous human cancers. There are now over 75 known human receptor tyrosine kinase (RTKs), and many of them are known to be proto-oncogenes involved in oncogenesis 28. Based on this, RTKs could be an attractive molecular target for efficient therapeutic tool in anticancer therapy, and it has been developed many of tumor RTK-targeted drugs such as trastuzumab, imatinib, bevacizumab, and gefitinib. c-met, one of RTK proto-oncogenes, is a disulfide-linked α-β heterodimeric receptor tyrosine kinase and is overexpressed and/or dysregulated in a variety of human tumors including melanoma, lung, prostate, gastric, ovarian, and bladder. Various c-met mutations have been well described in multiple solid tumors. It was reported that most of c-met mutations was in juxtamembrane and a cytosolic c- 10

terminal domain with tyrosine kinase activity, so it can induce constitutive activation of the HGF signaling pathway. Since Met signaling contributes to tumor survival, growth, angiogenesis, and metastasis, it could be a potential target for cancer therapy. Various approaches have been explored to inhibit the HGF or Met-mediated function in experimental systems. One of the approaches is the use of neutralizing monoclonal antibody (Cao et al., 2001) or ribozyme 29-31 to block the HGF activity or HGF/Met expression. According to this study, the inhibition of the HGF/Met function suppressed both tumor growth, and metastasis in a variety of animal tumor models. Also, it was also reported other ways to block the HGF/Met interaction like the inhibition of Met tyrosine kinase activity by small molecule inhibitors 17,32,33, impairment of receptor dimerization either by dominant-negative Met 34,35 or by a dualfunction decoy Met receptor that interferes with both HGF binding to Met and Met homodimerization 36, and ligand displacement by a competitive inhibitor of HGF 37. The activation of c-met protein is involved in the induction of vascular endothelial growth factor 38,39 through downregulation of the antiangiogenesis factor thrombospodin-1 40. Recently, it was shown that a 11

selective small molecule inhibitor of c-met, PHA665752, inhibits tumorigenicity and angiogenesis 41 and another small molecule inhibitor of c- Met, PF-2341066, exhibits cytoreductive antitumor efficacy through antiproliferative and antiangiogenic mechanism 42. Using c-met, I compared single or dual shrna expression system, in their efficiency of shrna expression. To effectively inhibit c-met expression, I generated three kinds of recombinant adenovirus expressing single shrna, Ad- E1-shMet4 and Ad- E1-shMet5, and dual shrna, Ad- E1-shMet4+5 and examined their efficiency for c-met knockdown. Our results show that Ad- E1-shMet4+5 can strongly and specifically silence c- Met mrna and protein expression in various human cancer cell lines tested. Also, I observed that reduced c-met expression induce dramatic inhibition of cell proliferation inhibition by senescence mechanism. Functional analyses showed that the inhibition of c-met effectively inhibited not only cell proliferation and tube formation of primary cultured human endothelial cells in vitro but also rat aorta ring sprouting of endothelial cells ex vivo. Furthermore, it significantly suppressed the growth of established U343 human glioma xenograft model in nude mice. In addition, I demonstrate for the first time that the dual c-met-specific shrna expression results in 12

improved efficacy as well as sustained gene silencing effect than the single shrna expression. These observations strongly suggest that the inhibition of c-met expression using dual c-met specific shrna-expressing Ad, Ad- E1- shmet4+5 may hold strong promise for the cancer treatment. 13

II. MATERIAL AND METHODS Cell lines and cell culture All cell lines with the exception of Hep3B, which was maintained in modified Eagle s medium (MEM; Gibco BRL, Grand Island, NY), were cultured in Dulbecco s modified Eagle s medium (DMEM; Gibco BRL) supplemented with 10% fetal bovine serum (Gibco BRL), L-glutamine (2 mm), penicillin (100 IU/ml), and streptomycin (50 µg/ml). A human embryonic kidney cell line expressing the Ad E1 region (HEK293), brain cancer cell lines (U343 and U87MG), liver cancer cell lines (Hep3B, HepG2, Huh7, and Hep1), and a non-small lung cancer cell line (A549) were purchased from the American Type Culture Collection (ATCC, Manassas, VA). Human umbilical vein endothelial cells (HUVECs), isolated from human umbilical cord veins by collagenase treatment as described previously 43, were maintained and propagated in M199 medium (Invitrogen, Carlsbad, CA) containing 20% fetal bovine serum (FBS), penicillin-streptomycine (100 IU/ml), 3 ng/ml basic fibroblast growth factor (Upstate Biotechnology, Lake Placid, NY), and 5 units/ml heparin. HUVECs were used between passages 2 and 7. All cell lines were maintained at 37 o C in a humidified atmosphere at 5% CO 2. 14

Construction of expression plasmids expressing VEGF-specific shrna Two target sirna sequences for VEGF were selected using a dedicated program provided by Ambion Inc. (Ambion, Austin, TX). Two double-stranded RNA oligonucleotides, corresponding to two regions at nucleotides 124-144 (shvegf-1) and 379 399 (shvegf-2) of human VEGF mrna (GenBank accession number gi: 6631028), were then synthesized using Silencer TM sirna construction kit (Ambion, Austin, TX) (Fig. 1A). To generate shrna targeting VEGF, the DNA fragment for the expression of shrna targeting the positions 124-144 or 379-399 of human VEGF was generated by annealing the sense oligonucleotide 5 -gatccc AAGTTCATGGATGTCTATCAGttcaagagaCTGATAGACATCCATGAACTT ttttttggaaa-3 and its cognate antisense oligonucleotide 5 - agcttttccaaaaaaaagttcatggatgtctatcagtctcttgaactgatagacatc CATGAACTTgg-3 or the sense oligonucleotide 5 - gatcccaaatgtgaatgcagaccaaagttcaagagactttggtctgcattca CATTTttttttggaaa-3 and its cognate antisense oligonucloetide 5 - agcttttccaaaaaaaaatgtgaatgcagaccaaagtctcttgaactttggtctgc ATTCACATTTgg-3 for shvegf-1 and shvegf-2, respectively. The 21- nucleotide VEGF target sequences are indicated in uppercase letters, whereas the 9-nucleotide hairpin and the sequences necessary for the directional 15

cloning are depicted in lowercase letters. After digestion with BamHI and HindIII, the fragments were inserted into a psilencer-2.1-hygro-u6 vector (Ambion), resulting in pshvegf-1 and pshvegf-2. The control vector (pscshrna) was constructed by inserting a sequence that expresses a sirna with limited homology to sequences in the human and mouse genomes. Synthesis and transfection of sirnas specific for IL-8 or c-met Four synthetic double-stranded oligonucleotides corresponding to four 21-nt sequences from human IL-8 (GenBank accession number gi: 28610153; Table 1) and Five synthetic double-stranded oligonucleotides specific to human c-met (GenBank accession number gi: 4557746; Table 2), were designed using RNAi software (Ambion: www.ambion.com/techlib/misc/sirna_finder.html) and synthesized using the Ambion Silencer TM sirna construction kit (Ambion, Austin, TX). To determine the most effective sirna, one µg of each of the synthesized sirnas and two control sirnas specific to lamin A/C and luciferase were introduced into Hep3B (IL-8) or U343 (c-met) cells (3 x 10 5 ) in 6-well dishes using lipofectamine plus reagent (Invitrogen, Carlsbad, CA). After 48 hr, cells were harvested and total RNA was extracted using the RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturer s instructions. Semiquantitative RT-PCR was then performed using β-actin as an internal control 16

to normalize gene expression. Generation of VEGF -specific shrna-expressing shuttle vector To generate Ads expressing VEGF-specific shvegf at the E3 region of Ad, shvegf gene excised from pshvegf-2 was first subcloned into psp72-e3 Ad shuttle vector 44 using EcoRI-HindIII, generating a psp72- E3/U6-shVEGF. The newly constructed psp72-e3/u6-shvegf E3 shuttle vector was then co-transformed with a replication-incompetent Ad total vector expressing lacz at E1 region, pdl- E1, or replication-competent Ad total vector, pdl- B7, into Escherichia coli BJ5183 for homologous recombination, generating pad- E1-shVEGF and pad- B7-shVEGF Ad vectors, respectively. Generation of IL-8 -specific shrna-expressing shuttle vector To generate Ads expressing IL-8-specific shrna, DNA fragment targeting position 194-212 of human IL-8 was first generated by annealing the sense oligonucleotide 5 - gatcccgaacttagatgtcagtgcatattcaagagaaatatgcactgacat CTAAGTtttttggaaa-3 and its cognate antisense oligonucleotide 5 - agcttttccaaaaaagaacttagatgtcagtgcatatctcttgaaaatatgcactg ACATCTAAGTgg-3. The 19-nucleotide IL-8 target sequences are indicated in uppercase letters, whereas the 9-nucleotide hairpin and sequences necessary 17

for directional cloning are depicted in lowercase letters. After annealing with sense and antisense oligonucleotides, the fragments were inserted into a psilencer-2.1-hygro-u6 vector (Ambion), resulting in pshu6il-8. The shil-8 gene expression cassette excised from pshu6il-8 was then subcloned into the psp72-e3 Ad shuttle vector 44 using EcoRI-HindIII, generating psp72- E3/U6-shIL8. In addition, the annealed IL-8-specific shrna DNA fragment was subcloned into the psp72-e3/cmv Ad shuttle vector 44, generating psp72-e3/cmv-shil8. The newly constructed psp72-e3/u6-shil8 and psp72-e3/cmv-shil8 E3 shuttle vectors were linearized with XmnI and cotransformed with pdl- E1, a replication-incompetent Ad total vector expressing lacz at E1 region, or pdl- B7, a replication-competent Ad total vector, into Escherichia coli BJ5183 for homologous recombination, generating the pad- E1-U6shIL8, pad- E1-CMVshIL8, pad- B7-U6shIL8, and pad- B7-CMVshIL8 Ad vectors. Generation of c-met -specific shrna-expressing shuttle vector To generate Ads expressing c-met-specific shrna, a DNA fragment targeting position 1987-2007 and 3142-3162 of human c-met was first generated by annealing, the following oligonucleotides were used: c-met #4: sense oligonucleotide: 5 - gatcccaaactagagttctccttggaattcaagagattccaaggagaactct 18

AGTTTttttttggaaa-3 and its cognate antisense oligonucleotide: 5 - agcttttccaaaaaaaaactagagttctccttggaatctcttgaattccaaggaga ACTCTAGTTTgg-3 ; c-met #5: sense oligonucleotide: 5 - gatcccaattagttcgctacgatgcaattcaagagattgcatcgtagcgaac TAATTttttttggaaa-3 and its cognate antisense oligonucleotide: 5 - agcttttccaaaaaaaattagttcgctacgatgcaatctcttgaattgcatcgtagc GAACTAATTgg-3. The 19-nucleotide c-met target sequences are indicated in uppercase letters, whereas the 9-nucleotide hairpin and sequences necessary for directional cloning are depicted in lowercase letters. Each annealed fragments were inserted into a psilencer-2.1-hygro-u6 vector (Ambion) digested with BamHI and HindIII, resulting in pshmet4 and pshmet5. The shmet4 and shmet5 gene expression cassette excised from pshmet4 and pshmet5 was then subcloned into the psp72-e3 Ad shuttle vector using EcoRI-HindIII, generating psp72-e3/shmet4 and psp72-e3/shmet5. In addition, to generate Ad expressing both shmet4 and shmet5, U6-shMet5 region of pshmet5 was amplified by polymerase chain reaction (PCR) with the following primer set: 5 - GTCAAGCTTGAATTCCCCAGTGGAAAGACG-3 as the sense primer and 5 -GTCGAATTCAAGCTTCCAAAAAAAATTAGTTCG-3 as the antisense primer. The primers were designed to create HindIII sites (underlined), and 19

the pshmet5 containing U6-shMet5, was used as a template. The PCR product containing U6-shMet5 was digested with HindIII, and then was subcloned into the psp72-e3/shmet4, generating psp72-e3/shmet4+5. The newly constructed psp72-e3/shmet4, psp72-e3/shmet5, and psp72-e3/shmet4+5 E3 shuttle vectors were linearized with XmnI and co-transformed with pdl- E1, a replication-incompetent Ad total vector expressing lacz, into Escherichia coli BJ5183 for homologous recombination, generating the pad- E1-shMet4, pad- E1-shMet5, and pad- E1-shMet4+5 Ad vectors. To verify homologous recombinants, plasmid DNA purified from overnight E. coli cultures was digested with HindIII, and the digestion pattern was analyzed. Correct homologous recombinant Ad plasmid DNA was digested with PacI and transfected into 293 cells to generate Ad- E1-shMet4, Ad- E1- shmet5, and Ad- E1-shMet4+5 Ads. E1-deleted, replication-incompetent Ad (Ad- E1) was also prepared. Generation of VEGF-, IL-8 -, or c-met-specific shrna-expressing Ads All viruses were propagated in 293 cells, and purification, titration, and quality analysis of all Ads used were performed as previously described. The titer (plaque forming units per ml, PFU/ml) used in this study was determined by limiting dilution assay in 293 cells. 20

Quantitation of VEGF, IL-8, c-met, and MMP-2 by ELISA Concentrations of Human VEGF-A, IL-8, c-met, and MMP-2 in conditioned medium, cell lysate, or tumor tissue lysates were measured using commercially available ELISA kits according to instructions provided by the vendor (VEGF: R & D Systems, Minneapolis, MN; IL-8: Biosource International Inc., Camarillo, CA; c-met: Biosource International Inc.). Cells were plated in six-well plates in medium containing 10% FBS. When the cells reached subconfluence, cells were infected with Ads at different MOIs. Conditioned media and cells were harvested 72 hr after transduction for recombinant Ads. To remove endogenously expressed, medium was replaced with serum free DMEM 30 hr before each respective harvest time point. Tumor tissue was removed from mice and snap-frozen in liquid nitrogen. Tissues were homogenized in ice-cold PBS with protein inhibitor cocktail (Sigma, Cat #P8340). Homogenates were centrifuged in a high-speed microcentrifuge for 10 min and analyzed for total protein content using a BCA protein assay reagent kit (Bio-rad, Hercules, CA). Levels of VEGF, IL-8, c- Met, or MMP-2 in supernatant were determined by ELISA according to the manufacturer s instructions. Serial dilutions of purified recombinant human VEGF-A, IL-8, c-met, and MMP-2 were used to establish standard curves. ELISA results were normalized relative to the total protein concentration in 21

each sample and were calculated as picograms per milligram of total protein. Tube formation assay First, 250 µl of growth factor-reduced Matrigel (Collaborative Biomedical Products, Bedford, MA) was pipetted into a 16-mm diameter tissue culture well and polymerized for 30 min at 37 C. HUVECs incubated in M199 containing 1% FBS for 6 hr were harvested after trypsin treatment and suspended in M199 containing 1% FBS. HUVECs were then plated onto the layer of Matrigel at a density of 2 x 10 5 cell/well, and conditioned media from Ad-infected cells were added. VEGF (Upstate Biotechnology, Lake Placid, NY) at 10 ng/ml and conditioned media from uninfected cells were used as controls. Cells were then allowed to form tubes for 20 hr at 37 C, and photographed ( 40, 100). The area covered by the tube network was determined using an optical imaging technique in which pictures of the tubes were scanned into Adobe Photoshop and quantitated using Image-Pro Plus software (Media Cybermetics Inc., Silver Spring, MD). Ex vivo aortic ring sprouting assay Aortas were harvested from Sprague Dawley rats (6 weeks old), as described previously 45. After removing the surrounding fibro-adipose tissues and rinsing with Hank s balanced salt solution (HBSS) buffer, aortas were cut into 1 mm ring segments. Plates (48-well) were coated with 120 µl of 22

Matrigel; after gelling, rings were placed in the wells and sealed in place with an overlay of 50 µl of Matrigel. Conditioned media from Ad-infected cells (250 µl) were added to thewells. VEGF (20 ng/ml, Upstate Biotechnology) and conditioned media from uninfected cells were used as controls. On day 6, cells were fixed and stained with Diff-Quick. Each ring was scored from 0 (least positive) to 5 (most positive) depending on the degree of vessel sprouting observed. Cultures were scored in a double-blinded manner by three independent observers. In vivo Matrigel plug assay U343 cells (2 X 10 5 cells) were plated in 6-well plates and infected with replication-incompetent Ad (Ad- E1, Ad- E1-shVEGF) or replicationcompetent Ad (Ad- B7, Ad- B7-shVEGF), along with PBS as a negative control. After 2 hr, treated cells were harvested after trypsin treatment, washed three times with 5 ml of HBSS buffer. Cells were then mixed with 600 µl of cold Matrigel and injected with a 1 ml syringe into subcutaneous space above the flank region of the male athymic nu/nu mice. The injected Matrigel rapidly formed a single, solid gel plug. After 14 days, the animals were sacrificed and the skin of each mouse was pulled back to expose the Matrigel plug, which remained intact. To quantify the blood vessel formation, Matrigel plugs were embedded in O.C.T. compound (Sakura Finetec, Torrance, CA), 23

and cut into 10-µm sections. The cyrosections were treated with purified a monoclonal rat anti-mouse CD31 (platelet/endothelial cell adhesion molecule 1; BD Biosciences PharMingen), at a dilution of 1:50 as a primary antibody, and then with goat anti-rat IgG-HRP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) as a secondary antibody. All slides were counterstained with Meyer s hematoxylin. Six mice were used for each group. Migration assay The lower surface of 6.5-mm-diameter polycarbonate filters (8-µm pore size, Corning Costar, Cambridge, MA) was coated by immersion in 0.1% gelatin. Conditioned media from PBS or Ad-infected cells was placed in the lower part of Transwell chambers, and HUVECs, Hep3B, A549, or U343 cells were placed on the filter membrane on the top chamber. Cultures were incubated at 37 C for 4 hr, and cells remaining on the upper surface of the filter were removed with a cotton swab. Filters were stained with H&E, and cells that migrated through to the underside of the filter membrane were counted at 200x magnification. Ten fields were counted for each assay, and experiments were repeated at least three times. Matrigel invasion assay In vitro invasion assays were carried out using Transwell chambers with 6.5-mm diameter polycarbonate filters (Corning Costar, Cambridge, 24

MA) according to the manufacturer s instructions. Conditioned media from Ad-infected cells were placed in the lower chambers and Hep3B, A549, or U343 cells (1 x 10 5 cells/ 100 µl) were seeded onto Matrigel-coated filters in the upper chambers. After 24 hr incubation, cells on the upper surface of the filters were removed with a cotton swab, and filters were fixed with 100% methanol and stained with H&E. Cells that had invaded to the lower side of the filters were viewed under an optical microscope. Invasive activity is expressed as the mean number of cells in ten fields from three independent experiments. Gelatin Zymography Gelatin zymography was performed in 10% SDS polyacrylamide gels containing 0.1% gelatin. Samples with conditioned media used in migration and invasion assay were prepared in nonreducing loading buffer. After electrophoresis, SDS was removed by renaturation buffer (2.5% Triton X-100) to renature gelatinases. Gels were then incubated in developing buffer (50 mm Tris-HCl (ph 7.5), 150 mm NaCl and 10 mm CaCl 2 ) for 16 hr at 37 C, and then were stained with 0.25% Coomassie Blue R 250. All experiments were performed at least in triplicate, and representative experiments are shown. MTT assay 25

Cells were seeded in 24-well plates to 30-70% confluency and were infected with Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 at MOIs of 50 to 500. At indicated times post infection, 250 μlof 3-(4,5- dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT; Sigma Chemical Corp.) in phosphate-buffered saline (PBS) was added to each well. After 4 hr incubation at 37, the supernatant was discarded and the precipitate was dissolved with 1 ml of dimethylsulfoxide (DMSO). Plates were then read on a microplate reader at 540 nm. All assays were performed in triplicate. Number of living cells was calculated from non-infected cells cultured and treated with MTT in the same condition, as were the experimental groups. Reverse transcription (RT)-Polymerase chain reaction (PCR) analysis U343 cells (3 x 10 5 ) in 6-well plates were transduced with Ad- E1- shmet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 at MOIs of 50 MOI. Fortyeight hr later, the cells were harvested and total RNA was extracted by using the RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturer s instructions. For RT-PCR, 1 µg of total RNA was converted to cdna by treatment with 200 units of M-MLV reverse transcriptase and 500 ng of oligo(dt) primer in 50 mm Tris-HCl (ph 8.3), 75 mm KCl, 3 mm MgCl 2, 10 26

mm dithiothreitol, and 1 mm dntps at 37 C for 1 hr. The reaction was stopped by heating at 70 C for 15 min. Two µl of the cdna mixture was then used for enzymatic amplification. PCR was performed in 50 mm KCl, 10 mm Tris-HCl (ph 8.3), 1.5 mm MgCl 2, 0.2 mm dntps, 2.5 units of Taq DNA polymerase, and 0.1 µm each of primers. The sequences of oligonucleotide primers used in RT PCR for verification of senescence and the expected transcript sizes are listed in Table 3. The amplification was performed in a DNA thermal cycler (model PTC-200; MJ Research) under the following condition: denaturation at 94 C for 10 min for the first cycle and for 1 min starting from the second cycle, annealing at 55 C for 1 min, and extension at 72 C for 1 min for 30 repetitive cycles. The reaction was terminated with a final cycle of extension (72 C for 10 min). Semi-quantitative RT-PCR was performed using β-actin as an internal control to normalize gene expression for the PCR templates. Cell cycle analysis Cells were infected with Ad- E1 or Ad- E1-Met4+5 Ads at an MOI of 30. At indicated times post infection, trypsinized and floating cells were pooled, washed with PBS, and fixed in 70% (v/v) ethanol. For assessment of DNA contents, cells were stained with PI and monitored by a fluorescence- 27

activated cell sorter (FACS) (Becton Dickinson, Sunnyvale, CA). Cell cycle distribution was determined with the Modifit LT program (Verity Software House Inc.). HGF specificity assay U343 cells were seeded in 6-well plates to 20% confluency and were infected for 6 hours with Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1- shmet4+5 at MOIs of 50. After 24 hr Ad infection, serum free media with or without HGF/SF (R & D Systems, Minneapolis, MN) of 25ng/ml was added to cells and the effects of c-met inhibition were tested. After 48 hr the cells were washed three times with phosphate-buffered saline (PBS) and cells on the plate were then stained with 0.5% crystal violet in 50% methanol. Electron microscope (EM) cytology U343 cells were infected with Ad- E1, Ad- E1-Met4, Ad- E1- Met5, or Ad- E1-Met4+5 at a MOI of 30. Three days postinfection, cells were gently trypsinized and pelleted for 5 minutes at 3000 rpm in a microcentrifuge tube. After washing with phosphate buffered saline (PBS), pellets were fixed for 4 hours at 4 in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (ph 7.3) containing 2% sucrose and 1 mm calcium chloride. They were then postfixed with 1% OsO 4 in 0.1 M cacodylate-hcl, ph 7.4, for 1 hr. The samples were dehydrated in a gradient series of ethanol, embedded 28

in Epon812, and examinined using an electron microscope (EM 902A, Zeiss, Oberkochen, Germany). Immunoblotting analysis Cells grown in 100 mm dish were infected with Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 at MOI of 200. Three days postinfection, cells were lysed in 50 mm Tris-HCl (ph 7.6), 1% Nonidet P-40 (NP-40), 150 mm NaCl, and 0.1 mm zinc acetate in the presence of a protease inhibitors. Protein concentration was determined using the bicinchoninic acid method (BCA assay; Pierce, Rockford, IL). Cell lysates or immunoprecipitates were fractionated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. Blocked membranes were then incubated with antibodies against c-met (diluted 1:50, Santa Cruz Biotechnology, Inc., San Diego, CA) or p44/42 MAP kinase (Erk), phosphor-p44/42 MAPK (Thr202/Tyr204) (phosphor-erk), MEK1/2, phosphor-mek1/2 (Ser217/221), Akt, phosphor-akt (Ser473) (each diluted 1:500, Cell Signaling Technology, Beverly, MA), and the immunoreactive bands were visualized using a using enhanced chemiluminescence (ECL) (Santa Cruz Biotechnology, Inc.). Assessment of anti-tumor effects in human xenograft model Male athymic nu/nu mice were obtained from Charles River 29

(Yokohama, Japan) at 5-6 weeks of age. All mice were housed and handled in accordance with the Animal Research Committee s Guidelines at Yonsei University College of Medicine, and all facilities are approved by AAALAC (association of assessment and accreditation of laboratory animal care). Mice were implanted with 1 x 10 7 U343 human glioma, Hep3B human hepatoma, and A549 human lung cancer cells subcutaneously in the abdominal region. When tumors reached an average size of 70-100 mm 3, Ads were then administered intratumorally (1 x 10 8 PFU) on days 1, 3, and 5. First day of treatment was designated as day 1. Tumor growth was measured three times weekly by a caliper until the end of the study by measuring the length and width of the tumor. Tumor volume was calculated using the following formula: volume = 0.523LW 2. Evaluation of tumor xenograft by histology and immunohistochemistry I used paraffin bloc slide to observe microvessel density (MVD) in tumor. Tumor tissue was fixed in IHC zinc fixative (formalin-free) (BD Biosciences PharMingen, San Diego, CA), embedded in paraffin, and cut into 3-µm sections (Wax-it; Vancouver, Canada). Representative sections were stained with H&E or Masson s trichrome and then examined by light microscopy. To observe microvessel density (MVD), slides were deparaffinized in xylene and then processed as described previously 46. All 30

slides were counterstained with Meyer s hematoxylin. Blood vessels were counted as described previously. The most vascular area of tumors was identified on low power ( 100) and vessels were counted in ten high-power fields ( 200). The data are presented as mean ± SE of three tumors per group. For detecting cleaved deoxynucleotic acids in situ, the same paraffin slides were also used and performed according to the instructions in the ApoTaq peroxidase in situ apoptosis detection kit (Chemicon International, Temecula, CA) 47. Intratumoral microvessel density assessment Blood vessels were counted as described previously 48,49. The most vascular area of tumors was identified on low power (x100) and vessels were counted in ten high-power fields (x200). The data are presented as mean ± SE for three tumors per group. Statistical Analysis The data are expressed as mean ± standard error (SE), and the significance of differences between group means was determined by the Mann-Whitney test (nonparametric rank sum test) using Stat View software (Abacus Concepts, Inc., Berkeley, CA). Differences were considered significant when P < 0.05. 31

III. RESULTS Design of shrnas directed to the VEGF mrna VEGF-A exists in seven different isoforms, VEGF-121, 145, 148, 165, 183, 189, and 206 amino acids in humans. They are generated as a result of alternative splicing from a single VEGF pre-mrna. To generate interfering RNAs that will degrade all seven VEGF isoforms, I designed two sirna sequences that were within the VEGF 121 mrna (Fig. 1A). Two VEGFspecific shrna-expressing plasmids were then constructed to express shrnas under the control of a murine U6 promoter (RNA polymerase IIIdependent promoter). The 21-nucleotide VEGF-specific targeting sequence separated by a 9-nucleotide short spacer and five thymidines (T5) as termination signal were annealed and cloned into psilencer-2.1-hygro-u6 plasmid, generating VEGF-specific shrna-expressing plasmids, pshvegf-1 and pshvegf-2 (Fig. 1B). K562 cells were then transfected with 1 µg of pshvegf-1 or pshvegf-2, along with psc-shrna plasmid (psilencer-2.1- hygro-u6 encoding scrambled sequence of RNA) as a negative control. At 48 hr post transfection, RT-PCR was performed to determine the endogenous level of VEGF mrna expression. Relative amounts of RT-PCR products for VEGF were obtained by semi-quantitative PCR against β-actin. As shown in Fig. 1C, of the two VEGF-specific shrna-expressing plasmids generated, 32

pshvegf-2 potently suppressed the expression of VEGF mrna (> 80%) compared with pshvegf-1 or control plasmid, psc-shrna. On the basis of these results, I selected shvegf-2 as the functional shrna sequence targeting VEGF and subsequently generated shvegf-2 expressing Ads. Figure 1. Design and characterization of vascular endothelial growth factor (VEGF)-specific small interfering RNAs (sirnas). (A) Location of two VEGF-specific sirnas examined in this study. (B) Sequences for generating two shrna targeting VEGF, shvegf-1 and shvegf-2. (C) shrna-mediated in vitro knockdown of VEGF gene. Cells were transfected for 48 hr with pshvegf-1 or pshvegf-2, and the knockdown of endogenous expression was measured by RT-PCR for VEGF. Densitometric analysis was done and the relative expression for each band was normalized with β-actin. 33

Construction and effects of shvegf-expressing replication-competent Ad on the expression of VEGF Although sirna method is powerful genetic tool for targeting specific knock-down, direct delivery of synthetic sirna is challenging. I therefore chose replication-competent oncolytic Ad and investigated whether shrnas can be delivered and more importantly functionally relevant in cancer cells. Previously, we constructed cancer-specific replicating E1A-mutated (at the Rb binding sites of E1A) and E1B-deleted Ad, Ad- B7 50. Ad- B7 showed significantly improved cytopathic effect and viral replication in a cancer cellspecific manner. To induce efficient and long-term VEGF silencing, I introduced shvegf-2 under the control of an U6 promoter at the E3 region of Ad- B7, generating Ad- B7-shVEGF oncolytic Ad. To investigate whether VEGF-specific shrna could abrogate VEGF expression, U343 glioma cells were infected with Ad- B7 or Ad- B7- shvegf at an MOI of 10 and 20. Non-infected cells were used a negative control. After 24 and 48 hr, the conditioned media was harvested and VEGF ELISA was carried out. Surprisingly, the expression and secretion of VEGF was significantly reduced in cells cultured in the presence of conditioned media infected with both Ad- B7 and Ad- B7-shVEGF (Fig. 2A). The significant suppression of VEGF expression by Ad- B7 and Ad- B7-34

shvegf oncolytic Ads that was observed in the U343 cells was also true in the other cancer cell lines (U87MG, Hep3B, and Hep1). Consistent with our findings, other investigators have also reported that suppression of VEGF expression by E1A-expressing replication competent Ads 51-53. To further demonstrate the effect of E1A-expressing replication-competent Ads in reducing VEGF expression, U343 cells were infected with replicationincompetent E1-deleted Ad (Ad- E1) and replication-competent E1Aexpressing Ads (YKL-1, Ad- B7, Ad- B7-shVEGF) at an MOI of 10 and 20. As shown in Fig. 2B, the VEGF protein level was significantly decreased in U343 cells infected with YKL-1, Ad- B7, and Ad- B7-shVEGF Ads in an MOI-dependent manner as compared to those in cells infected with Ad- E1. Consistent with previous findings, data presented here clearly demonstrate that E1A-expressing replication-competent Ads suppress VEGF expression through a preserved E1A region. Construction and effect of shvegf-expressing replication-incompetent Ad on the expression of VEGF Since E1A protein down-regulates VEGF expression, I next constructed E1-deleted replication-incompetent Ad expressing shvegf-2, Ad- E1-shVEGF, to investigate whether shrna targeting VEGF would knock-down VEGF expression. To determine the effect of shrna expression 35

on VEGF protein levels, a panel of cancer cell lines (U343, Hep3B, Hep1, C33A, U87MG, and A549) was transduced with Ad- E1 or Ad- E1-shVEGF at various MOIs. Four days following transduction, VEGF ELISA was carried out on cell supernatant. Uninfected cells were used as a negative control. As shown in Fig. 2C, the VEGF protein level was significantly decreased in all cancer cell lines transduced with Ad- E1-shVEGF in an MOI-dependent manner. In contrast treatment of cells with Ad- E1, regardless of the applied MOIs, was ineffective. More specifically, Ad- E1- shvegf almost completely knock-downed VEGF expression in the U343 cells at MOIs of 100 (P < 0.001 versus Ad- E1) and this was also true in other cancer cell lines (Hep3B, Hep1, U87MG) at high MOIs. Despite the reduction in endogenous VEGF protein secretion by Ad- E1-shVEGFtransduced cancer cells, the growth rate of these cells was not different from those cells transduced with Ad- E1. Together, these data show that shvegfexpressing replication-incompetent Ad, Ad- E1-shVEGF, reduces VEGF expression with high efficiency. Ad- E1-shVEGF inhibits angiogenesis in vitro and in vivo To determine the functional relevance of VEGF shrna, I first examined its effects on inhibiting capillary tube formation of HUVEC cells in vitro. HUVEC cells seeded onto Matrigel-coated plates were incubated in 36

Figure 2. Quantification of VEGF. Human U343 glioma cells were infected with Ad- B7 or Ad- B7-shVEGF Ad (A) and Ad- E1, YKL-1, Ad- B7, or Ad- B7-shVEGF Ad (B) at an MOI of 10 and 20. VEGF concentration was measured in the culture supernatant at 24 hr and 48 hr after infection by ELISA. (C) Various human cancer cell lines were transduced with Ad- E1 or Ad- E1-shVEGF replication-incompetent Ad at range of 10 ~ 500 MOIs. VEGF concentration was measured in the culture supernatant at 96 hr after transduction by ELISA. Each value represents the mean ± SE of at least three independent experiments. 37

the presence of medium supplemented with human recombinant rvegf 165 protein (10 ng/ml), medium conditioned by U343 control cells, Ad- E1- transduced cells, or Ad- E1-shVEGF-transduced cells. There was no effect within 12 hr of treatment with conditioned media from Ad- E1-shVEGF-transduced cells, but the inhibition was evident at 16 hr and more clearly pronounced at 20 hr with minimal cytotoxicity. Treatment with human recombinant rvegf 165 protein and the conditioned medium collected from Ad- E1-transduced U343 cells led to the formation of organized elongated tube-like structures resembling capillaries with an extensive network (Fig. 3A). However, in the presence of Ad- E1-shVEGFtransduced cells conditioned medium, no such organized structures were observed and the plate was filled with incomplete network of a capillary-like structure. The conditioned medium from Ad- E1-transduced cells only had a mild effect, leading to 8.6% inhibition of tube formation, whereas the conditioned medium from Ad- E1-shVEGF-transduced cells reduced the relative tube length by 49.2% (P < 0.05 versus Ad- E1) (Fig. 3C). These findings demonstrate that shvegf-expressing Ad, Ad- E1-shVEGF, functionally inhibits the formation of tube-like structures in the in vitro matrigel tube-formation assay. 38

Next, I tested the ability of Ad- E1-shVEGF to inhibit vessel sprouting using rat aortic rings (explants) embedded in Matrigel beds. The Matrigel bed mimics physiological extracellular matrix representing its natural composition and architecture. Due to these features, Matrigel enables several cell types, including endothelial cells, to maintain in culture their in vivo phenotype in 3-dimensional organization. The microvessel sprouting was demonstrated by staining the aortic rings with Diff-Quick as described in the materials and methods section. When aortic rings were cultured in U343 cells conditioned medium, rapid microvessel outgrowth was observed within 6 days (Fig. 3B). The extent of angiogenesis was comparable to that observed in the presence of human recombinant rvegf 165 protein (20 ng/ml). In marked contrast, the aortic rings cultured in the presence of Ad- E1- shvegf-transduced cells conditioned medium showed little or no sprouting (P < 0.001 versus Ad- E1). A marked delay in outgrowth of the sprouts with smaller number of microvessels and branches was observed in the Ad- E1- shvegf-transduced cells conditioned medium-treated aortic rings (Fig. 3D). The data further confirms the potent inhibition of angiogenesis induced by Ad- E1-shVEGF. 39

Figure 3. Characterization of replication-incompetent Ad expressing VEGFspecific shrna. (A) Inhibition of tube formation by Ad- E1-shVEGF. HUVEC cells were plated on Matrigel-coated plates and then incubated in the presence of medium supplemented with human recombinant protein rvegf 165, medium conditioned by U343 control cells, Ad- E1-transduced cells, or Ad- E1-shVEGF-transduced cells. After 20 hr, changes in cell morphology were captured using an inverted microscope ( 40, 200). (B) The area covered by the tube network was quantified using Image-Pro Plus software. Experiments 40

were repeated three times and values are means of triplicates of representative of three independent experiment; bars, ± SE. *, P < 0.05 versus Ad- E1 treatment. (C) Inhibition of vessel sprouting by Ad- E1-shVEGF. Rat aortic rings in Matrigel were cultured in the presence of medium supplemented with human recombinant protein rvegf 165, medium conditioned by U343 control cells, Ad- E1-transduced cells, or Ad- E1-shVEGF-transduced cells, and stained with Diff-Quick on day 6. Representative aortic rings were photographed. Original magnification: 40. (D) Vessel sprouting index. *, P < 0.001 versus Ad- E1 treatment. The assay was scored from 0 (least positive) to 5 (most positive) and the data are presented as mean (n=6). Each data point was assayed in sextuplets. I next examined whether Ad- E1-shVEGF and Ad- B7-shVEGF Ads are capable of blocking angiogenesis in vivo. U343 cells were infected with replication-incompetent Ads (Ad- E1 and Ad- E1-shVEGF) at an MOI of 200 or replication-competent Ads (Ad- B7 and Ad- B7-shVEGF) at an MOI of 10. Infected cells were then mixed with cold Matrigel and injected into subcutaneous space above the flank region. Fourteen days later, formed Matrigel were excised and photographed. Matrigel plugs mixed with cells infected with Ad- E1 or Ad- B7 were abundantly filled with angiogenic 41

vessels (Fig. 4A). However, Matrigel plugs mixed with cells infected with VEGF-specific shrna-expressing Ads, Ad- E1-shVEGF and Ad- B7- shvegf, had markedly reduced vascularization after 14 days. The newly formed vessel contents inside the Matrigel plugs were then quantified by counting CD31-positive microvessels (Fig. 4B-C). Treatment with Ad- E1- shvegf or Ad- B7-shVEGF significantly reduced the number of blood vessels by 76.5% (P < 0.001, versus Ad- E1) and 63.2% (P < 0.001, versus Ad- B7), respectively, as compared to their cognate control groups. These results indicate that replication-incompetent and replication-competent Ads are capable of expressing functional shvegf, and efficiently inhibit angiogenesis in vivo. 42

Figure 4. VEGF-specific shrna-expressing Ads inhibit vascularization in the Matrigel plug assay. Male athymic nu/nu mice were injected s.c. at the flank region with Matrigel mixed with U343 cells infected with Ad- E1, Ad- E1- shvegf, Ad- B7, or Ad- B7-shVEGF. Matrigel plugs were removed 14 days after implantation, photographed, and prepared for immunohistological examination. Six mice were used for each group. (A) Representative images of Matrigel plugs removed from mice showing reduction in plug vascularization by Ad- E1-shVEGF- and Ad- B7-shVEGF-treated cells. (B) Immunohistochemical staining of sections of Matrigel plugs with anti-cd31 antibody shows reduced numbers of endothelial cells and vessel structures in Ad- E1-shVEGF- and Ad- B7-shVEGF-treated plugs. Original magnification: 400. (C) The numbers of blood vessels in 100 were counted and averaged. Each column represents the mean ± SE of the blood vessels per group. *, **; P < 0.001 versus Ad- E1 or Ad- B7 treatment, respectively. shvegf expression does not inhibit viral replication To determine whether virally expressed shvegf altered viral replication, VEGF-specific shrna-expressing replication-competent Ad, Ad- B7-shVEGF, was examined for its potential to induce cytopathic effects. The assay is based on the idea that at low concentrations, viruses need to go 43

through multiple rounds of infection, replication, viral release, and reinfection of surrounding cells in order to completely wipe out a cell monolayer. Various cancer cells (A549, Hep1, U343, U87MG, Hep3B, C33A) were infected with Ad- B7 or Ad- B7-shVEGF at MOIs of 0.1-20. For a negative control, cells were infected in parallel with replication-incompetent Ad, Ad- E1. As seen in Fig. 5A, virally induced cytopathic effect appeared to be same or slightly faster in Ad- B7-shVEGF as compared to Ad- B7. This result was consistent over several independent experiments, suggesting that shvegf expression does not inhibit viral replication. Figure 5. Cytopathic effects of shvegf-expressing oncolytic Ad. Cells were infected with Ad- E1, Ad- B7, or Ad- B7-shVEGF at the indicated MOIs. At 4 ~ 10 days post-infection, surviving cells were stained using crystal violet. Replication-incompetent Ad, Ad- E1, served as a negative control. Each cell line was tested at least three times, and data shown are from representative experiments. 44

Enhanced anti-tumor effect of shvegf-expressing oncolytic Ad Results from the matrigel plug assay showed that Ad- B7-shVEGF Ad was notably more potent in inhibiting angiogenesis than Ad- E1-shVEGF Ad. Therefore, the oncolytic Ad expressing VEGF-specific shvegf, Ad- B7-shVEGF, was subsequently assessed for its ability to suppress the growth of U343 human glioma xenograft model established in nude mice and its effects were compared against that of its cognate control Ad, Ad- B7. This U343 tumor model exhibits aggressive tumor growth kinetics. Tumors were generated by subcutaneous injection of cells into the mice abdominal region. When tumors reached an average size of 70-100 mm 3, tumors were injected intratumorally with PBS, Ad- B7 (1 x 10 8 PFU), or Ad- B7-shVEGF Ad (1 x 10 8 PFU) every 2 days for a total of three times (Q2Dx3). As seen in Fig. 6A, control tumors increased to an average size of 2770 ± 27 mm 3 by 48 days after the treatment. By day 48, all mice became moribund in the control group and thus euthanized. In marked contrast, Ad- B7 or Ad- B7-shVEGF Ad-treated tumors reached to an average size of 1660 ± 400 mm 3 and 504 ± 195 mm 3 by 48 days, showing a 37.6% (P < 0.01, versus PBS control) and 81% growth inhibition (P < 0.001 versus PBS and P < 0.01 versus Ad- B7), respectively. Throughout the course of the study, no systemic toxicity, such as diarrhea, loss of weight, or cachexia was observed. Survival rate was also significantly 45

increased in animals treated with Ad- B7-shVEGF Ad. By day 70 following treatment, 100% of the animals in the Ad- B7-shVEGF Ad group were still viable as compared to 38% in the Ad- B7-treated group (P <0.01). In comparison as mentioned previously all animals in the PBS-treated control group were euthanized by day 48 (Fig. 6B). These data indicate that suppression of VEGF expression by oncolytic Ad had strong inhibitory effects on tumor growth, resulting in increased survival. Moreover, no apparent toxicity was noted in animals that received this Ad during the course of these experiments. Figure 6. shvegf-expressing oncolytic adenovirus Ad- B7-shVEGF and the growth of established tumors and survival of mice. Subcutaneous implanted 46

tumors derived from U343 human glioma cells were treated with Ad- B7 ( ) or Ad- B7-shVEGF ( ), along with PBS ( ) as a negative control. (A) Tumor volume was monitored over time (days) after treatment with Ads. The arrow indicates when treatment was given (1 x 10 8 PFU/mouse). Values represent the means ± SE for eight animals per group. (B) Overall survival. The survival analysis of Ads treated mice were performed over a period of 70 days. By day 48, all mice in the PBS-treated groups were moribund and euthanized. In comparison, 100% of mice treated with Ad- B7-shVEGF were viable after 70 days. Antiangiogenic effects of VEGF-specific shvegf-expressing oncolytic Ad, Ad- B7-shVEGF, in U343 human glioma xenografts To verify the proposed mechanism of action of VEGF-specific shvegf-expressing oncolytic Ad, Ad- B7-shVEGF-treated tumors were further investigated by histological examination. Tumors were harvested from each treatment group at 7 days following the three sequential treatments. As seen in Fig. 7A, a marked decrease in number of cells and large areas of necrosis was replaced by fibrous tissue was observed in Ad- B7-shVEGF Adtreated tumors. Majority of remaining tumor mass treated with Ad- B7- shvegf was necrotic, whereas necrotic lesions were barely detectable in the tumors treated with PBS or Ad- B7. To determine whether the reduced size 47

of Ad- B7-shVEGF-treated tumors coincided with reduced neovascularization, microvessel density as assessed using anti-cd31 antibody was determined. A marked decrease in number and size of CD31-positive vessels was observed in Ad- B7-shVEGF Ad-treated tumors. Quantitation of microvessels showed that density was reduced by 79.8% (P < 0.001) and 76.7% (P < 0.001) in response to Ad- B7-shVEGF Ad treatment as compared to PBS and Ad- B7 Ad treatment groups, respectively (Fig. 7B). Apoptotic cells were also more abundantly detected in Ad- B7-shVEGF-treated tumor tissue than in PBS- or Ad- B7-treated tumor tissue. Viral persistence and distribution within the tumor mass was also examined by immunohistochemistry using an antibody specific to Ad hexon protein. Ad particles were detected in wide areas of Ad- B7-treated tumors, most frequently in the peripheral tumor area and in between the area of necrosis and proliferating tumor cells. In comparison few numbers of big clusters of Ad particles in patch style were detected in the peripheral tumor area of Ad- B7-shVEGF-treated tumors (Fig. 7A). This staining pattern observed in the tumors treated with Ad- B7-shVEGF indicates that viral replication occurred actively only in tumor areas devoid of necrosis. Moreover, a marked reduction in VEGF expression was measured by ELISA in homogenates of tumors from the Ad- B7-shVEGF oncolytic Ad-treated 48

group (Fig. 7C). VEGF ELISA demonstrated that treatment of Ad- B7- shvegf Ad resulted in 64.9% (P < 0.001) and 47.4% (P < 0.05) reduction compared to tumors treated with PBS and Ad- B7, respectively. This result closely parallels and supports the decreased neovascularization in the tumor mass and enhanced anti-tumor effect of VEGF-specific shvegf-expressing oncolytic Ad. Figure 7. Antiangiogenic effects of VEGF-specific shrna-expressing oncolytic Ad, Ad- B7-shVEGF, in U343 human glioma xenografts. (A) H & E staining (H &E). Blood vessel density was assessed by 49

immunohistochemical staining for CD31; Hematoxylin counterstained (CD31). Brown staining indicates positive staining for endothelial cells. TUNEL staining of apoptotic cells in tumor tissue; methyl green counter stained (TUNEL). Tumors treated with Ad- B7-shVEGF exhibited significant increase in apoptotic levels. Greater dark brown nuclei with double-strand DNA breaks (indicated by an arrow) can be seen in tumors treated with Ad- B7-shVEGF as compared to PBS- or Ad- B7-treated groups. Immunohistochemical staining of Ad hexon protein to localize Ad in tumor tissue; Hematoxylin counterstained (Ad hexon). Ad-infected cells stain brown as indicated by an arrow. Original magnification: 400. (B) The mean microvessel density for each treatment group was determined by counting the number of CD31-positive vessels in ten high-power fields ( 200) of each section in a blinded fashion. Three tumors per group were analyzed, and all data are shown as mean ± SE. *, P < 0.001 versus PBS or Ad- B7 treatment. (C) VEGF contents in tumors. Treatment of Ad- B7-shVEGF Ad results in significant reduction in VEGF levels as compared to treatment of PBS or Ad- B7. Each data point represents mean VEGF levels for each individual tumor (7 mice per group) and the mean VEGF level for each group is represented by a line. VEGF level is expressed as picograms per milligram of total protein. 50

Improved efficacy of oncolytic Ad-mediated sirna expression over replication-incompetent Ad-mediated sirna To better understand the time-course and magnitude of the gene silencing effect in vitro, I treated U343 cells with either Ad- E1-shVEGF at an MOI of 100 or Ad- B7-shVEGF at an MOI of 0.1, and assessed its gene silencing effect at various time points. The rationale for the difference in MOI for the two Ads is based on the fact that the oncolytic Ad is highly efficient in its cell killing effect that if the MOI is too high the interpretation of the data would be too difficult. As expected, on days 3 and 7 after Ad- E1-shVEGF treatment, VEGF expression levels were 80% and 91% lower than those from untreated control, respectively, demonstrating the efficient suppression of VEGF expression of Ad- E1-shVEGF up to 7 days after transduction (Fig. 8A). However on day 14, the Ad- E1-shVEGF s efficacy was significantly attenuated, showing only 48% suppression compared to untreated control. In marked contrast, VEGF expression level was not visibly reduced in cells infected with Ad- B7-shVEGF even until day 7. However, by day 14 VEGF expression was significantly suppressed by 98%, demonstrating that the propagation, amplification and release of viral progeny to neighboring cells takes ~two weeks to see the functional effects in this model system. I next compared the duration and magnitude of sirna-mediated 51

gene silencing elicited by Ad- E1-shVEGF and Ad- B7-shVEGF in vivo. When U343 tumors reached an average size of 70-100 mm 3, tumors were injected intratumorally with PBS, or with 3 x 10 8 PFU of either Ad- E1- shvegf or Ad- B7-shVEGF Ad every 2 days for a total of three times (Q2Dx3). Tumors were then harvested at day 7 and 21 after viral injection, and VEGF expression was measured by ELISA. As shown in Fig. 8B, following 7 days of treatment, a marked suppression of VEGF expression was observed in tumor tissues treated with Ad- E1-shVEGF or Ad- B7-shVEGF, resulting in 58.5% and 87.3% reduction compared to PBS-treated control tumors, respectively. On day 21 however, there was almost no suppression of VEGF expression for those tumors treated with Ad- E1-shVEGF. In contrast, VEGF expression was significantly suppressed by 62.5% in the tumor treated with Ad- B7-shVEGF compared to PBS-treated control tumors in the same time period. These results demonstrate that in vivo, replication-competent Admediated sirna expression results in improved efficacy as well as sustained gene silencing effect over replication-incompetent Ad. 52

Figure 8. Time-course and magnitude of the VEGF gene silencing effect of Ad- E1-shVEGF or Ad- B7-shVEGF Ad. (A) Human U343 glioma cells were treated with Ad- E1-shVEGF or Ad- B7-shVEGF Ad at an MOI of 100 or 0.1, respectively. VEGF expression level was assessed at 3, 7, and 14 days after viral treatment. Each value represents the mean ± SE of at least three independent experiments. (B) U343 xenograft models were injected intratumorally with PBS, 3 x 10 8 PFU of either Ad- E1-shVEGF or Ad- B7- shvegf Ad every 2 days for a total of three times. Tumors were then harvested on day 7 and 21 after Ad administration and subjected to VEGF ELISA. VEGF level is expressed as picograms per milligram of total protein. Results represent the mean ± SE of three animals per group. Identification of effective sirna sequences and generation of replicationincompetent Ads expressing shrna specific to IL-8 To identify sirna effective sequences against IL-8, I examined four 53

sirnas targeting different regions of human IL-8 mrna (gi:28610153) as described in Table 1. Of the various cancer cell lines assessed, Hep3B expressed the highest IL-8 levels. To evaluate possible off-target effects, Hep3B cells were also transfected with sirna specific to lamin A/C and luciferase. Relative amounts of IL-8 were measured by semi-quantitative RT- PCR, normalized to β-actin. As shown in Fig. 9A, of the four sirnas synthesized, IL-8 sirna No. 2 (194-212) most potently suppressed the expression of IL-8 mrna (> 90%). Transfection with lamin A/C- and luciferase-specific sirna resulted in no significant alteration of IL-8 RNA expression compared to non-transfected HepB cells. On the basis of these results, IL-8 sirna No.2 was chosen to generate shrna specific to IL-8 for use in subsequent experiments. Table 1. Sequences of the four IL-8-specific sirnas examined in this study Number positions sequences #1 178 198 caaggagtgctaaagaactta #2 194 212 gaacttagatgtcagtgcata #3 250 270 aagaactgagagtgattgaga #4 279 297 cacactgcgccaacacagaaa Two IL-8-specific shrna-expressing replication-incompetent adenoviruses, Ad- E1-U6shIL8 and Ad- E1-CMVshIL8 (Fig. 9B), were 54

constructed to express shrnas under the control of the murine U6 promoter (RNA polymerase III-dependent promoter) and the CMV promoter (RNA polymerase II-dependent promoter), respectively. The 19-nucleotide IL-8- specific targeting sequence, separated by a 9-nucleotide short spacer and including five thymidines (T5) as a termination signal was annealed and cloned into the pdl- E1 adenoviral total vector, generating Ad- E1-U6shIL8 and Ad- E1-CMVshIL8. Figure 9. Characterization of IL-8-specific small interfering RNAs (sirnas) and the structure of adenoviruses used in this study. (A) sirna-mediated in vitro knockdown of IL-8 transcripts. Cells were transfected for 48 hr with four IL-8-specific sirnas, along with sirna specific to lamin A/C and luciferase 55

as controls. Knockdown of endogenous expression was measured by RT-PCR for IL-8. Densitometric analysis was performed and the relative expression of each band was normalized to β-actin. (B, C) Schematic representation of the genomic structure of adenovirus vectors used. Comparison of U6 and CMV promoters in IL-8 knockdown in vitro To compare promoter efficiency in the knockdown of IL-8 expression, a panel of cancer cell lines (Hep3B, Huh7, Hep1, HepG2, U87, U251N, U343, and A549) was transduced with Ad- E1, Ad- E1-CMVshIL8, or Ad- E1- U6shIL8 at various MOIs. Non-transduced cell lines were also used as negative controls. After 72 hr, conditioned media was harvested and an IL-8 ELISA was carried out. As shown in Fig. 10A, the synthesis and secretion of IL-8 was significantly reduced by Ad- E1-U6shIL8. In Hep3B cells, IL-8 levels was decreased by 73.7% and 73.4% (P < 0.05 and P < 0.05) compared to non-transduced and control adenovirus (Ad- E1)-transduced cells, whereas the reduction was only 8.2% and 7% with Ad- E1-CMVshIL8. This efficient knockdown of IL-8 by Ad- E1-U6shIL8 was also observed in each of the other cancer cell lines tested. Moreover, Ad- E1-U6shIL8 strongly silenced IL-8 expression even at low MOIs, whereas a notably higher MOI of Ad- E1- CMVshIL8 was needed for effective knockdown (Fig. 10B). Despite the reduction in endogenous IL-8 protein secretion by Ad- E1-U6shIL8-56

transduced cancer cells, the growth rate of these cells was not different from cells transduced with Ad- E1. Together, these data show that IL-8-specific shrna-expressing Ads reduce IL-8 expression, and, more importantly, that the U6 promoter is much more effective in driving IL-8-specific shrna expression than the CMV promoter. Figure 10. Quantitation of IL-8 secreted by cells transduced with replicationincompetent Ads. (A) Various human cancer cell lines were transduced with replication-incompetent Ad- E1, Ad- E1-CMVshIL8, or Ad- E1-U6shIL8 in a range of 20 ~ 1000 MOI. IL-8 concentration was measured in culture 57

supernatants 72 hr after transduction by ELISA. *P<0.05 versus Ad- E1 treatment. (B) Dose-dependent suppression of IL-8 expression by Ad- E1- CMVshIL8 and Ad- E1-U6shIL8. Each value represents the mean ± SE of at least three independent experiments. *P<0.05 versus Ad- E1 treatment. Effect of IL-8-specific shrna expression on the function of endothelial cells in vitro I investigated the effects of Ad-mediated IL-8-specific shrna expression on the migration of HUVECs in vitro. Conditioned medium from A549 cells transduced with Ad- E1-U6shIL8 reduced the migration of endothelial cells by 48.8% (P < 0.001 compared to Ad- E1). In comparison, conditioned medium from Ad- E1-CMVshIL-8-transduced cells inhibited migration only 24% (P < 0.05 compared to Ad- E1) (Fig. 11A). Nearly identical results were observed using conditioned medium from Hep3B cell lines. I then assessed the functional effect of Ad-mediated IL-8-specific shrna expression on capillary tube formation in vitro. HUVEC cells seeded onto Matrigel-coated plates were incubated in the presence of medium supplemented with human recombinant rvegf 165 protein (10 ng/ml) or with conditioned medium from Ad- E1-, Ad- E1-CMVshIL8-, or Ad- E1-58

U6shIL8-transduced cells. As shown in Fig. 11B, conditioned medium from Ad- E1-U6shIL8-transduced cells effectively inhibited HUVEC tube formation, whereas rvegf 165 and conditioned media collected from Ad- E1- and Ad- E1-CMVshIL8-transduced cells resulted in the formation of robust, elongated tube-like structures, which were organized in much larger numbers of cells. In particular, conditioned medium from Ad- E1-U6shIL8-transduced cells diminished relative tube length by 65.5% (P < 0.001 compared to Ad- E1, Fig. 11C). Microvessel sprouting was demonstrated by staining aortic rings with Diff-Quick, as described in materials and methods. As shown in figures 11D and 11E, a marked delay in the outgrowth of sprouts from explants, with a regression in both the number of microvessels and the number of branches, was observed in aortic rings treated with conditioned medium from Ad- E1- U6shIL8-transduced cells, showing a decrease in the number of microvessels of 68.2% (P < 0.001) and 65% (P < 0.001) compared to untransduced and Ad- E1-transduced conditioned media, respectively. In contrast, neither Ad- E1 nor Ad- E1-CMVshIL8 had an inhibitory effect on microvessels sprouting from aortic rings. In fact, the extent of the capillary network was comparable to that observed in the presence of conditioned medium from untransduced 59

cells. These data further confirm the potent inhibition of angiogenesis induced by Ad- E1-U6shIL8. Effect of IL-8-specific shrna expression on tumor cell migration, invasion, and MMP-2 expression I next examined the effect of IL-8-specific shrna expression on tumor cell invasion. Quantitative analysis showed that conditioned media collected from Ad- E1-U6shIL8-transduced Hep3B and A549 cancer cells effectively blocked cellular migration and invasion after 24 hr; by 57.9% (P < 0.05 versus Ad- E1) and 54% (P < 0.05 versus Ad- E1) for Hep3B cells and by 61.1% (P < 0.05 versus Ad- E1) and 53.3% (P < 0.05 versus Ad- E1) for A549 cells (Fig. 12A, 12B). In contrast, conditioned media collected from non-transduced and Ad- E1-transduced cells, failed to affect migration and basal invasion of cancer cells. MMP expression on cancer cells as well as vascular endothelial cells has been shown to regulate neovascularization by enhancing peri-cellular fibrionolysis and cellular locomotion. Of the number of MMP family members, MMP-2 plays a critical role in the angiogenic switch during carcinogenesis. By using conventional gelatin zymography, I compared 60

61

Figure 11. Effects of Ad-mediated IL-8-specific shrna expression on migration, tube formation, and vessel sprouting of endothelial cells. (A) HUVEC cell migration. Migratory cells are represented as the number of migrated cells per high-power field (200x). Ten fields were counted in triplicate from each sample. bars, ± SE. *, P < 0.05 versus Ad- E1 treatment; **, P < 0.001 versus Ad- E1 treatment. (B, C) Tube formation assay. (B) HUVEC cells were plated on Matrigel-coated plates at a density of 2 x 10 5 cells/well and incubated in the presence of medium supplemented with human recombinant rvegf 165 (10 ng/ml), control conditioned medium from Hep3B cells, or conditioned medium from Ad- E1-, Ad- E1-CMVshIL8-, or Ad- E1-U6shIL8-transduced cells. After 20 hr, changes in cell morphology were captured using an inverted microscope. Original magnification: 40x and 100x. (C) The area covered by the tube network was quantified using Image-Pro Plus software. Experiments were repeated three times and values are means of triplicate measurements from one representative experiment; bars, ± SE. **, P < 0.001 versus Ad- E1 treatment. (D, E) Aorta ring sprouting assay. (D) Aortic rings in Matrigel were cultured in the presence of medium supplemented with human recombinant rvegf 165 (20 ng/ml), control conditioned medium from Hep3B cells, or conditioned medium from Ad- E1-, Ad- E1-CMVshIL8-, or Ad- E1-U6shIL8-transduced cells for six 62

days, then stained with Diff-Quick. Representative aortic rings were photographed. Original magnification: 40x. (E) The assay was scored from 0 (least positive) to 5 (most positive) and data are presented as mean ± SE (n=6). **, P < 0.001 versus Ad- E1 treatment. gelatinolytic activity of MMP-2 in response to Ad- E1-, Ad- E1-CMVshIL8, and Ad- E1-U6shIL8-transduced U343 and U251N cancer cells. As shown in Fig. 12C, MMP-2 gelatinolytic activity was reduced in Ad- E1-U6shIL8- transduced cells, whereas Ad- E1- or Ad- E1-CMVshIL8-transduced cancer cells showed MMP-2 activity similar to the levels observed for parental cells. ELISA quantitation of MMP-2 protein in A549 cancer cells transduced with Ad- E1-U6shIL8 indicated that a reduction in the MMP-2 protein level closely paralleled the observed loss of MMP-2 gelatinolytic activity (Fig. 12D). These results strongly suggest that Ad- E1-U6shIL8 inhibits invasion of cancer cells, presumably through the inhibition of MMP-2 activity. 63

Figure 12. Inhibition of cancer cell migration, invasion, and MMP-2 expression by Ad- E1-U6shIL8. (A) Quantitative evaluation of Hep3B and A549 cancer cell migration. Experiments were performed in triplicate and repeated three times, means ± SE are shown. *, P < 0.05 versus Ad- E1 treatment. (B) Tumor cell invasiveness was quantified as the mean number of cells in ten fields of view per filter. Experiments were performed in triplicate and repeated three times, means ± SE are shown. *, P < 0.05 versus Ad- E1 treatment. (C) Zymographic detection of secreted gelatinase activity in conditioned media from U343 and U251N cells treated with Ad E1, Ad E1-64

CMVshIL8, or Ad E1-U6shIL8. This is one representative result from three independent experiments. (D) Modulation of MMP-2 protein expression in A549 cells transduced with Ads. Human MMP-2 concentration in conditioned media was determined by ELISA. Each bar represents the mean ± SE (n = 3). *, P < 0.05 versus Ad- E1 treatment. **, P < 0.001 versus Ad- E1 treatment. Generation and characterization of oncolytic Ad expressing shil-8 To induce long-term IL-8 silencing, I introduced shil-8 under the control of CMV and U6 promoters at the E3 region of Ad- B7, generating Ad- B7-CMVshIL8 and Ad- B7-U6shIL8 oncolytic Ad, respectively (Fig. 9C). To determine whether virally expressed shil-8 altered viral replication, the newly engineered Ads were first examined for their potential to induce cytopathic effects. Multiple cancer cells (A549, U343, Hep1, U87, and Hep3B) were infected with Ad- B7, Ad- B7-CMVshIL8, or Ad- B7- U6shIL8 at MOIs of 0.1-20. As a negative control, cells were infected in parallel with replication-incompetent Ad, Ad- E1. As seen in Fig. 13A, virally induced cytopathic effects appeared to be similar with Ad- B7- CMVshIL8 and Ad- B7-U6shIL8 as compared to Ad- B7. This result was consistent over several independent experiments, suggesting that shil-8 expression does not inhibit viral replication. Next, I examined whether shil-8-expressing oncolytic Ads can 65

functionally knock-down IL-8 expression. Hep3B and A549 cells were infected with Ad- B7, Ad- B7-CMVshIL8, or Ad- B7-U6shIL8 at MOIs of 0.5 and 10, respectively. As shown in Fig. 13B, significant suppression of IL- 8 expression by Ad- B7-U6shIL8 oncolytic Ad was observed in both Hep3B and A549 cell lines with 62.1% (P < 0.001 versus Ad- B7) and 53.4% (P < 0.001 versus Ad- B7) decrease in IL-8 protein level, respectively. To demonstrate IL-8 specificity of Ad- B7-U6shIL8, cells were also infected with Ad- B7-U6shVEGF, an oncolytic Ad expressing VEGF-specific shrna under the control of the U6 promoter. As predicted, no inhibition of IL-8 expression was observed in response to Ad- B7-U6shVEGF treatment. IL-8 protein levels were also considerably decreased in both Hep3B and A549 cells infected with cognate control oncolytic Ad, Ad- B7. This is in agreement with ours and other investigators previous findings that E1A expression suppresses tumor angiogenesis 47,51,53. Together, these data demonstrate that oncolytic Ad- B7-U6shIL8 reduced IL-8 expression with high degree of efficiency as well as high specificity. 66

Figure 13. Characterization of IL-8-specific shrna-expressing oncolytic Ads. (A) Semiquantitative assessment of cytotoxic potency by crystal violet cytopathic effect (CPE) assay. Cells were infected with Ad- E1, Ad- B7, Ad- B7-CMVshIL8, or Ad- B7-U6shIL8 at the indicated MOIs. At 4 ~ 10 days post-infection, surviving cells were stained using crystal violet. Replication-incompetent Ad, Ad- E1, served as a negative control. Each cell line was tested at least three times, and data shown are from representative experiments. (B) Quantitation of IL-8. Hep3B and A549 cells were infected with Ad- E1, Ad- B7, Ad- B7-CMVshIL8, or Ad- B7-U6shIL8 adenovirus (MOI of 0.5 for Hep3B and 10 for A549). IL-8 was measured in culture supernatants by ELISA 24 hr after infection. Each value represents the mean 67

± SE of at least three independent experiments. **, P < 0.001 versus Ad- E1 treatment. Oncolytic adenovirus expressing shil-8 inhibits tumor growth in nude mice Since Ad- B7-U6shIL8 exhibited greater IL-8 suppressive effect than Ad- B7-CMVshIL8, oncolytic Ad- B7-U6shIL8 was chosen to examine its effect in vivo. Hep3B and A549 tumor models established in nude mice showed aggressive growth kinetics. When tumors reached an average size of 100 mm 3, tumors were injected intratumorally with PBS, Ad- B7, or Ad- B7-U6shIL8 every 2 days for a total of three injections (Q2Dx3). As seen in Fig. 14A, tumor volume was greatly reduced in mice treated with both oncolytic Ads (Ad- B7 and Ad- B7-U6shIL8) compared to control mice treated with PBS. At 35 days after viral treatment, tumors of Hep3B tumorbearing mice treated with PBS reached an average tumor volume of 3672 ± 575 mm 3 and the average tumor volumes for the Ad- B7 and Ad- B7- U6shIL8 groups were 942.68 ± 107 mm 3 and 335.61 ± 91 mm 3, respectively. Similarly, mice bearing A549 tumor xenograft showed tumor growth inhibition of 91.7% (P < 0.001) in response to Ad- B7-U6shIL8 treatment. Throughout the course of the study, no systemic toxicity, such as diarrhea, loss of weight, or cachexia was observed. These data indicate that suppression of IL-8 expression by oncolytic Ad had strong inhibitory effects on the growth 68

of these tumor xenografts. The histology of tumors obtained 7 days post final dose was then observed. As shown in Fig. 14B, H & E staining of A549 tumors treated with Ad- B7-U6shIL8 showed dramatic effects as only number of viable tumor cells were observed. In addition apoptotic changes and cell necrosis, surrounded by stromal fibrosis was also seen. Masson s trichrome staining determined collagen type I, stained in blue, as the predominant morphological component of these fibrotic changes. Microvessel density was then examined by immunostaining with anti-pecam (anti-cd31) antibody. A marked decrease in endothelial cells and vessel structures was observed in Ad- B7- U6shIL8-treated tumors compared to PBS- or Ad- B7-treated tumors. Quantitation of microvessels showed that the density was reduced by 44.7% (P <0.05) and 50.6% (P <0.05) in Hep3B and A549 xenograft models, respectively, compared to Ad- B7 (Fig. 15A). To analyze the degree to which Ad- B7-U6shIL8 induced apoptosis in vivo, TUNEL assay was carried out using the same tumor sections. As seen in Fig. 14B, the level of apoptosis was considerably higher in Ad- B7-U6shIL8-treated tumor tissue than in PBS- or Ad- B7-treated tumor tissue. This result closely parallels and supports the decreased neovascularization in the tumor mass and enhanced anti-tumor effect of IL-8-specific shrna-expressing oncolytic Ad. 69

Figure 14. Effect of oncolytic adenovirus expressing IL-8-specific shrna in vivo. (A) Subcutaneous implanted tumors derived from Hep3B and A549 cells were treated with Ad- B7 ( ) or Ad- B7-U6shIL8 ( ), along with PBS ( ) as a negative control. Tumor volume was monitored over time (days) after treatment with Ads. The arrow indicates when Ad was administered (1 x 10 8 PFU/mouse for Hep3B and 2 x 10 8 PFU/mouse for A549). The number of animals per group is indicated in parentheses. Values represent the mean ± SE. *, P < 0.05 versus Ad- B7. **, P < 0.001 versus Ad- B7. (B) Photos of 70

representative tumor sections from animals in each group. Hematoxylin & eosin staining (H & E). Masson's trichrome staining of extracellular matrix (blue) in the tumor tissue sections (M & T) and hematoxylin counter-stained. Microvessels were stained with anti-pecam antibody (CD31) and hematoxylin counter-stained. Brown staining indicates positive staining for endothelial cells. TUNEL staining of apoptotic cells in tumor tissue; methyl green counter stained (TUNEL). Original magnification: 400x. To investigate the biological significance of IL-8 down-regulation in tumor tissue, I assessed IL-8 and VEGF expression in tumor tissues. As shown in Fig. 15B, significant suppression of IL-8 expression by Ad- B7- U6shIL8 oncolytic Ad was observed in both Hep3B and A549 cell lines with 31.5% (P < 0.001 versus Ad- B7) and 71.2% (P < 0.05) reduction, respectively. Moreover, VEGF ELISA demonstrated that treatment with Ad- B7-U6shIL8 resulted in 65% (P < 0.05 versus PBS) and 68.2% (P < 0.05) reduction in Hep3B and A549, respectively (Fig. 15C). These findings suggest that IL-8-specific shrna down-regulated the expression of both IL-8 and VEGF, key mediators of angiogenesis. 71

Figure 15. Effect of oncolytic adenovirus expressing IL-8-specific shrna in vivo. (A) Reduction in microvessel density following treatment with Ad- B7- U6shIL8. Mean microvessel density was determined by counting CD31- positive vessels in ten high-power fields (200x) of each section in a blinded fashion. Error bars represent SE. *, P < 0.05 compared to Ad- B7 treatment. (B) IL-8 expression in tumors. (C) VEGF expression in tumors. Each data point represents mean IL-8 and VEGF levels for each individual tumor. *, P < 0.05 versus Ad- B7. **, P < 0.001 versus Ad- B7. Activation of angiogenesis is also responsible for increased tumor cell entry into circulation and metastasis. Therefore, I investigated whether suppression of IL-8 expression also blocked metastasis of a malignant human 72

breast cancer cell line, MDA231. Administration of Ad- B7-U6shIL8 resulted in a marked reduction of metastatic burden. More specifically, three of 7 mice showed no metastatic spread and the remaining two exhibited minimal tumor spread as compared to control animals treated with PBS or Ad- B7. These data indicate that suppression of IL-8 expression had a strong inhibitory effect on tumor metastasis. Figure 16. Ad- B7-U6shIL8 treatment inhibits tumor metastasis. MD231 human breast cancer cells were injected intravenously to form pulmonary metastatic lesions. After 23 days, mice received intrapleural injections of PBS, Ad- B7, or Ad- B7-U6shIL8. On day 28 after viral injection, mice were sacrificed, lungs were removed and weighed, and tumor nodules were counted to evaluate therapeutic efficacy. Mean of mice treated with PBS, Ad- B7, or Ad- B7-U6shIL8; bars, SE (n = 11 for PBS, n = 12 for Ad- B7, n = 7 for Ad- B7-U6shIL8). * indicates P < 0.05 compared to PBS treatment. 73

Identification of effective sirna sequences and generation of recombinant Ad expressing shrna specific to c-met c-met is the high-affinity receptor tyrosine kinase for hepatocyte growth factor (HGF) or scatter factor (SF) and Met/HGF signaling is involved in many cancers and regulates biological activities 35. It has been reported that c-met was overexpressed and mutated in a variety of malignancies. Because a number of c-met activating mutations, many of which are located in the tyrosine kinase domain, to search effective sirna, I examined five c-met sirnas except for tyrosine kinase domain. Five kinds of c-met sirnas to target different regions of human c-met (gi:4557746) mrna were described in Table 2. Table 2. Sequences of the four c-met-specific sirnas examined in this study Number positions sequences #1 428 446 aaggttgctgagtacaagact #2 754 772 aggaccggttcatcaacttct #3 1258 1256 tggatcgatctgccatgtgtg #4 1987 2007 aaactagagttctccttggaa #5 3142 3162 aattagttcgctacgatgcaa Of the various cancer cell lines assessed, U343 expressed the highest level of c-met, so it was used to compare inhibition efficiency of the five 74

candidate sirnas. To evaluate the possible off-target effects, I also transfected cells with sirna specific to lamin A/C and luciferase as controls. Relative amounts of c-met were measured by semi-quantitative RT-PCR, normalized to β-actin. As shown in Figure 17A, two of the five sirnas synthesized, c-met sirna No. 4 (1987-2007) and No. 5 (3142-3162) potently suppressed the synthesis of c-met mrna (> 90%). Transfection with Lamin A/C- and Luciferase-specific sirna resulted in no significant alteration of c- Met RNA expression compared to non-transfected U343 cells. On the basis of these results, I selected c-met sirna No.4 and No.5 as the most highly functional sirna sequence. Figure 17. Characterization of c-met-specific small interfering RNAs (sirnas) and the structure of adenoviruses used in this study. (A) The effect 75

of sirna-mediated in vitro knockdown of c-met transcripts in U343 human glioma cell. Cells were transfected for 48 hr with four c-met-specific sirnas, along with sirna specific to lamin A/C and luciferase as controls. Knockdown of endogenous expression was measured by RT-PCR for c-met. Densitometric analysis was performed and the relative expression of each band was normalized to β-actin. (B) Schematic diagram of anenoviral construct used. To inhibit c-met expression effectively, I employed the adenovirusdelivered sirna technique and generated recombinant adenovirus expressing c-met shrna No. 4 or No. 5 (shmet4 or shmet5), respectively. Also, for further effective knockdown, I generated adenovirus expressing both No.4 and No.5 in the same adenovirus, which proved to be most effective. These kinds of c-met-specific shrna-expressing recombinant adenoviruses, Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 (Fig. 17B), were constructed to express shrnas under the control of the murine U6 promoter, respectively. Comparison of c-met suppression by recombinant adenovirus expressing shmet4, shmet5, or shmet4+5 Next, to determine the effect of shrna expression on c-met protein levels and compare the knockdown efficiency of three kinds of adenoviruses 76

expressing shmet, Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5, various cancer cell lines (U251N, U343, U87MG, HepG2, SK-Hep1, and A549) were infected with each virus at an MOI of 100 ~ 500. Uninfected cells were used negative control. After 3 days post-infection, transduced cells were harvested and assayed to determine the amounts of c-met protein. As shown in Fig. 18A, the synthesis and secretion of c-met was significantly inhibited by Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5. In U343 cells with Ad- E1-shMet4+5, c-met levels were reduced by 80.4% and 79.8% (P < 0.01 and P < 0.01) compared to untransduced and control adenovirus (Ad- E1)-transduced cells, whereas it was reduced to 64.5% and 63.4% with Ad- E1-shMet4 and was 61.8% and 60.7% with Ad- E1-shMet5. This efficient knockdown of c-met was also observed in other cancer cell lines tested. Also, the c-met protein level was significantly decreased with an MOI-dependent manner and Ad- E1-shMet4+5 among shmet-expressing Ads inhibited c-met expression most effectively (Fig. 18B). These results demonstrate that dual shrna expression system is much more effective than single shrna expression system. 77

Figure 18. Quantitation of c-met suppression in various cancer cells transduced with c-met specific shrna expressing Ads, Ad- E1, Ad- E1- shmet4, Ad- E1-shMet5, or Ad- E1-shMet4+5. (A) Various human cancer cell lines were transduced with Ad- E1, Ad- E1-shMet4, Ad- E1-shMet5, or Ad- E1-shMet4+5 in a range of 100 ~ 500 MOI. c-met concentration was measured in cell lysates 72 hr after transduction by ELISA. *P<0.05 versus Ad- E1 treatment. (B) Dose-dependent suppression of c-met expression by three kinds of Ads expressing c-met specific shrna, Ad- E1-shMet4, Ad- 78

E1-shMet5, or Ad- E1-shMet4+5. Each value represents the mean ± SE of at least three independent experiments. *P<0.05 versus Ad- E1 treatment. Reduced c-met inhibits cell proliferation through mitotic catastrophe by senescence I observed exciting phenotype in cancer cells infected with Ad expressing c-met specific shrna, Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5. After 2 days post-infection, observed with the naked eyes, cells infected with Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 were showed definitely decreased cell confluence and after 3 days postinfection, exhibit enlargement of cell volume (lamellafodia), flattened cell morphology, and the appearance of many vacuolated cells (Figure. 19A), but did not observed characteristic of apoptotic cells. These phenomena were typical senescence like phenotype. After 5 days, most of the cells floated from the culture plate and floated cells demonstrated necrosis-like characteristics such as disruption of intact cellular boundaries, and losses of distinctive nuclear membrane structures. To further confirm, I also performed TUNEL assay. In result, I did not detect TUNEL-positive apoptotic (data not shown). For quantitative analysis of cell proliferation inhibition MTT assay was conducted. As shown in Fig. 19B, U343 cells with Ad- E1-shMet4+5, c- Met levels were reduced by 78.7% and 76.6% (P < 0.01 and P < 0.01) 79

compared to untransduced and control adenovirus (Ad- E1)-transduced cells, whereas the reduction was 73.8% and 71.2% with Ad- E1-shMet4 and was 64.6% and 61.1% with Ad- E1-shMet5. This efficient proliferation inhibition of reduced c-met was also observed in each of the other cancer cell lines tested. The morphological difference observed was further examined under the electron microscope (EM) (Fig. 19C). At three day postinfection at an MOI of 100, in majority of U343 cells infected with Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5, the nuclei became significantly larger and some cells contained several nuclei of unequal sizes and observed an increased number of micronuclei, but did not observe the condensed or fragmented nuclei characteristic of apoptotic cells. I also observed considerable increases in the numbers of vacuoles and electron-dense lysosomes, and additionally observed the convolution or collapse of some nuclear membranes. In 1998, Gonos et al. have found eight genes that are overexpressed in senescent rat embryo fibroblasts. To confirm whether these phenomena were related with senescence; I conducted RT-PCR and determined the level of expression of a set of gene products commonly associated with cellular senescence. Used three genes were SM22, TGase II, and PAI-1 and primer 80

sequences used in RT-PCR were described in Table 3. As shown in Fig. 19D, after 48 hours, total RNA was extracted from cells infected with Ad- E1- shmet4, Ad- E1-shMet5, and Ad- E1-shMet4+5. The levels of mrna of SM22, TGase II, and PAI-1 were significantly increased (compared with Ad- E1). On the basis of these results, I made certain that decrease of c-met expression by shmet, induce senescence of cancer cell. To analyze the mechanisms by which Ad- E1-shMet4+5 inhibits cell proliferation, FACS analysis was applied to analyze the cell cycle of U343 cells after infection with adenovirus for 1 ~ 5 days. As shown in Fig. 19E, at hours 48 of adenovirus infection, the G 2 /M phase cell population was dramatically increased from 16.1 ± 0.9% (Ad- E1) to 42.6 ± 2.9% (Ad- E1- shmet4+5), but the percentage of cells at G 0 /G 1 -phase-cell population was decreased from 43.1 ± 2.3% (Ad- E1) to 22.3 ± 3.5% (Ad- E1-shMet4+5) and S-phase cell population was also slightly reduced by 5.6% compared to control adenovirus (Ad- E1)-transduced cells. No apoptosis peak was detected in sub G 1 population. Taken together, these results showed that apoptosis might not be the major mode of the cell death by knockdown of c- Met, but arrest cells at G 2 -M phase. Taken together, these results suggest that reduced c-met using Ad expressing c-met specific shrna inhibit cancer cell proliferation through mitotic catastrophe by senescence. 81

Table 3. Primer used for the analysis of the gene expression related to cellular senescence Gene primer sequence product (bp) Osteonectin Forward:5'-CTGTGGGAGCTAATCCTG-3' 602 Reverse:5'-GGGTGCTGGTCCAGCTGG-3' SM22 Forward:5'-TGGCGTGATTCTGAGCAA-3' 534 Reverse:5'-CTGCCAAGCTGCCCAAGG-3' TGaseII Forward:5'-CTCGTGGAGCCAGTTATCAACAGCTAC-3' 310 Reverse:5'-TCTCGAAGTTCACCACCAGCTTGTG-3' PAI-1 Forward:5'-GTGTTTCAGCAGGTGGCGC-3' 300 Reverse:5'-CCGGAACAGCCTGAAGAAGTG-3' β-actin Forward: 5'-CGTCTTCACCATGGAGA-3' 310 Reverse: 5'-CGGCCATCACGCCCACAGTTT-3' 82

Figure 19. Changes in the cellular morphologies of U343 cells transduced with Ads expressing c-met specific shrna, Ad- E1-shMet4, Ad- E1- shmet5, or Ad- E1-shMet4+5. (A) After 48 hr of infection with Ad- E1, Ad- E1-shMet4, Ad- E1-shMet5, or Ad- E1-shMet4+5 of 50 MOI, cell morphology was monitored under the microscope. (B) The reduced cell 83

proliferation by the Ads expressing c-met specific shrna was measured by MTT assay. Viability of control cells was set at 100% and viability relative to the control is presented. *P<0.05 versus Ad- E1 treatment. (C) The morphologies of dying cells were monitored using an electron microscope. U343 cells grown in six-well culture plated were infected with MOI of 50. After 72 hr of infection, cells were treated as described in Materials and Methods, and cell morphology was observed under the electron microscope. Representative high power (X 50,000) images are shown. (D) Analysis of gene expression associated with cellular senescence using RT-PCR. Total mrnas were isolated, and semiquantitative RT-PCR was performed using specific primers for genes (human SM22, TGaseII, and PAI-1) over-expressed in senescent cells. β-actin was served as an internal control. (E) DNA content analysis. At 48 hr after transduced with Ad- E1 or Ad- E1-shMet4+5, cells were fixed with ethanol, treated with RNAase and stained with propidium iodide (PI). The stained cells were then analyzed by flow cytometry. Data shown are representative fluorescence histograms of at least three independent experiments. Percentages of G0/G1-, S-, G2/M-, and sub-g1-phase cells were calculated by Modfit program (Verity Software House, Topsham, ME). 84

Reduced c-met inhibits VEGF expression and accordingly the function of endothelial cells c-met also was reported to induce angiogenesis in various cancer by inducing of vascular endothelial growth factor 38,39 and by simulataneously downregulating the antiangiogenesis factor thrombospodin-1 40. Based on these report, I made certain VEGF expression level by c-met inhibition using VEGF ELISA. Various cancer cell lines (U251N, U343, U87MG, HepG2, SK- Hep1, and A549) were infected with Ad- E1, Ad- E1-shMet4, Ad- E1- shmet5, and Ad- E1-shMet4+5 at a MOI of 30 ~ 500. Uninfected cells were used negative control. After 72 hr, conditioned media was harvested and VEGF ELISA was carried out. As shown in Fig. 20A, in accordance with c- Met ELISA result, the synthesis and secretion of VEGF was significantly inhibited by Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5. In U343 cells transduced with Ad- E1-shMet4+5, c-met levels were reduced by 71.4%% and 80% (P < 0.05 and P < 0.05) compared to untransduced and control adenovirus (Ad- E1)-transduced cells, whereas the reduction was 53% and 67% with Ad- E1-shMet4 and was 44.4% and 61% with Ad- E1- shmet5. This efficient inhibition of VEGF by knockdown of c-met was also observed in each of the other cancer cell lines tested. Also, Ad- E1-shMet4+5 of these shmet-expressing Ads inhibited VEGF expression very efficiently. 85

On the basis of these results, I made certain that decrease of c-met expression by shmet, inhibit expression of VEGF. I also investigated the effects of Ad- E1-shMet4+5 on the migration of HUVECs in vitro. When conditioned medium from U343 cells transduced with Ad- E1-shMet4+5 was placed in the lower chamber of a Transwell plate, migration of endothelial cells was reduced by 48.8% (P < 0.001 compared to Ad- E1) whereas conditioned medium from Ad- E1-shMet4+5-transduced cells inhibited migration only 24% (P < 0.05 compared to Ad- E1) (Fig. 20B). I next examined whether decreased c-met by infection of Ad- E1- shmet4+5 would have an inhibitory effect on morphological differentiation of endothelial cells, I conducted in vitro tube formation experiments using twodimensional Matrigel in HUVECs (Figure 20C). To do this, I used U343 conditioned media and HUVECs were infected with conditioned media, and then plated on growth factor-reduced Matrigel. There was no effect within 12 hr of conditioned media treatment, but the inhibition was shown at 19 hr and more clearly at 24 hr with little cytotoxicity. Positive control (10 ng/ml) and conditioned media collected from Ad- E1-infected U343 cells led to the formation of elongated and robust tube-like structures, which were organized by much larger number of cells. In contrast, conditioned media collected from 86

Ad- E1-shMet4+5-infected U343 cells markedly inhibited the width and the length of endothelial and after 24 hours, tubes were broken, shortened, and much thinner at many sites. In particular, conditioned medium from Ad- E1- shmet4+5-transduced cells diminished relative tube length by 44.1% (P < 0.001 compared to Ad- E1, Fig. 20D). These findings demonstrate that a c- Met-specific shrna-expressing Ad, Ad- E1-shMet4+5, functionally inhibits the formation of tube-like structures in the in vitro Matrigel tube-formation assay. As angiogenesis involves multiple steps, I assessed inhibition of vessel sprouting using rat aortic rings (explants) embedded in Matrigel beds. The Matrigel bed mimics a physiological extracellular matrix, representing its natural composition and architecture. Due to these features, Matrigel enables several cell types, including endothelial cells, to maintain their 3-dimensional, in vivo phenotype in culture. Microvessel sprouting was demonstrated by staining aortic rings with Diff-Quick, as described in materials and methods. As shown in figures 20E and 20F, a marked delay in the outgrowth of sprouts from the explants, with a regression in both the number of microvessels and the number of branches, was observed in aortic rings treated with conditioned medium from Ad- E1-shMet4+5-transduced cells, showing a decrease in the 87

number of microvessels of 90.5% (P < 0.001) and 85.7% (P < 0.001) compared to untransduced and Ad- E1-transduced conditioned media, respectively. In contrast, neither Ad- E1 had an inhibitory effect on microvessels sprouting from aortic rings. In fact, the extent of the capillary network was comparable to that observed in the presence of conditioned medium from untransduced cells. These data further confirm the potent inhibition of angiogenesis induced by Ad- E1-shMet4+5. The process of the formation of new blood vessels, angiogenesis, is complex and involves several discrete steps, including extracellular matrix degradation, proliferation, and migration of endothelial cells and morphological differentiation of endothelial cells to form tubes 54. I have first examined the effect of reduced c-met by infection of Ad- E1-shMet4+5 on HUVEC. HUVECs were infected with Ad- E1 and Ad- E1-shMet4+5 at a MOI of 100. After 48 hrs post-infection, cells infected with Ad- E1- shmet4+5 were showed definitely decreased cell confluence and exhibit lengthen cell morphology (Figure. 20G), and eventually the lengthened cells floated from the plate. MTT assay showed that cell viability was inhibited (Fig. 20H). These results indicate that Ad- E1-shMet4+5 inhibits not only proliferation of cancer cell but also of endothelial cell. 88

89

Figure 20. Effects of Ads expressing c-met-specific shrna in endothelial cell functions. (A) Reduced c-met expression down regulates VEGF expression. Various human cancer cell lines were transduced with Ad- E1, Ad- E1- shmet4, Ad- E1-shMet5, or Ad- E1-shMet4+5 with 100~500 MOI. VEGF concentration was measured in the culture supernatant at 72 hr after infection by ELISA. Each value represents the mean ± SE of at least three independent experiments. (B) Inhibition of HUVEC cell migration by Ad- E1-shMet4+5. Migratory cells are represented as the number of migrated cells per highpower field (200x). Ten fields were counted in triplicate from each sample. bars, ± SE. *, P < 0.05 versus Ad- E1 treatment; **, P < 0.001 versus Ad- E1 treatment. (C) Inhibition of tube formation by Ad- E1-shMet4+5. HUVEC cells were plated on Matrigel-coated plates at a density of 2 x 10 5 cells/well and incubated in the presence of medium supplemented with human 90

recombinant rvegf 165 (10 ng/ml), control conditioned medium from U343 cells, or conditioned medium from Ad- E1- or Ad- E1-shMet4+5-transduced cells. After 20 hr, changes in cell morphology were captured using an inverted microscope. Original magnification: 40x and 100x. (D) The area covered by the tube network was quantified using Image-Pro Plus software. Experiments were repeated three times and values are means of triplicate measurements from one representative experiment; bars, ± SE. **, P < 0.001 versus Ad- E1 treatment. (E) Inhibition of vessel sprouting by Ad- E1-shMet4+5. Representative aortic rings were photographed. (F) Vessel sprouting index. The assay was scored from 0 (least positive) to 5 (most positive) and data are presented as mean ± SE (n=6). **, P < 0.001 versus Ad- E1 treatment. Original magnification: 40x (G) Changes in the cellular morphologies of HUVEC cells transduced with Ad- E1-shMet4+5 of 50 MOI. Cell morphology was monitored under the microscope. (H) The reduced cell proliferation by the Ad- E1-shMet4+5 was measured by MTT assay. Viability of control cells was set at 100% and viability relative to the control is presented. *P<0.05 versus Ad- E1 treatment. Effect of Ad- E1-shMet4+5 on tumor cell migration, invasion, and MMP-2 expression I next examined whether the expression of c-met-specific shrna 91

could affect the cellular locomotion/invasion activity of cancer cell lines using a migration/invasion assay. Quantitative analysis showed that conditioned media collected from Ad- E1-shMet4+5-transduced U343 cancer cells effectively blocked cellular migration and invasion after 24 hr: by 65.6% (P < 0.05 versus Ad- E1) and 69% (P < 0.05 versus Ad- E1) for U343 cells (Fig. 21A, 21B). In contrast, conditioned media collected from untransduced and Ad- E1-transduced cells, did not significantly affect migration and basal invasion of cancer cells in this assay. Expression of MMPs in cancer cells (as well as vascular endothelial cells) has been shown to regulate neovascularization by enhancing pericellular fibrionolysis and cellular locomotion. Of the many MMPs, MMP-2 plays a critical role in the angiogenic switch during carcinogenesis. By using conventional gelatin zymography, I compared gelatinolytic activity of MMP-2 in Ad- E1-, and Ad- E1-shMet4+5-transduced U343 cancer cells. As shown in Fig. 21C, MMP-2 gelatinolytic activity was reduced in Ad- E1-shMet4+5- transduced U343 cancer cells, whereas Ad- E1-transduced cancer cells showed MMP-2 activity similar to parental cells. ELISA quantitation of MMP-2 protein in U343 cancer cells transduced with Ad- E1-shMet4+5 indicated that a reduction in the MMP-2 protein level closely paralleled the observed loss of MMP-2 gelatinolytic activity (Fig. 21D). These results 92

indicate that Ad- E1-shMet4+5 inhibits invasion of cancer cells, presumably by the inhibition of MMP activity. Figure 21. Inhibition of cancer cell migration, invasion, and MMP-2 expression by A Ad- E1-shMet4+5. (A) Inhibition of tumor cellular migration by Ad- E1-shMet4+5. U343 (1 X 10 6 cells/ml) cells were added to transwell chamber and treated with conditioned medium from Ad- E1- or Ad- E1-shMet4+5-transduced cells. After 6 hr incubation, migrated cells were quantified by counting the cells that migrated to the lower side of the filter with optical microscopy at 200x magnification. Results are expressed as percentage of control conditioned medium from U343 cells. Experiments were performed in triplicate and repeated three times, means ± SE are shown. *, P < 0.05 versus Ad- E1 treatment. (B) Inhibition of tumor cellular invasion by Ad- E1-shMet4+5. U343 (1 X 10 6 cells/ml) cells were added to transwell 93

chamber coated with Matrigel and treated with conditioned medium from Ad- E1- or Ad- E1-shMet4+5-transduced cells. After 18 hr incubation, the number of invaded cells was quantified by counting as the mean number of cells that invaded to the lower side of the filter with optical microscopy at 200x magnification. Experiments were performed in triplicate and repeated three times, means ± SE are shown. *, P < 0.05 versus Ad- E1 treatment. (C) Zymographic detection of secreted gelatinase activity in conditioned media from U343 cells treated with Ad E1 or Ad- E1-shMet4+5. This is one representative result from three independent experiments. (D) Modulation of MMP-2 protein expression in U343 cells transduced with Ads. Human MMP- 2 concentration in conditioned media was determined by ELISA. Each bar represents the mean ± SE (n = 3). *, P < 0.05 versus Ad- E1 treatment. Ad- E1-shMet4+5 Suppresses Met signaling On signaling by HGF/SF, c-met is activated via phosphorylation of the cytoplasmic domain and further activates downstream pathways such as the mitogen-activated protein kinase (MAPK), PI3K, and STAT or Erk signaling pathways. These pathways are essential for mediating biological activities such as cell migration, proliferation, morphogenesis, and survival. To evaluate the effect of c-met inhibition-induced c-met phosphorylation and downstream pathway, ERK1/2 and Akt phosphorylation were analyzed. As 94

shown in Fig. 22, cells treated with Ads expressing c-met specific shrna, Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5, the basal as well as the inducible activation of AKT and ERK1/2 were significantly inhibited. Figure 22. Inhibition of down signal pathway by c-met inhibition. U343 cells were infected at an MOI of 100 with Ad- E1, Ad- E1-shMet4, Ad- E1- shmet5, or Ad- E1-shMet4+5. After 48 hr, cell lysates were prepared and subjected to SDS-PAGE. Western blot was probed with anti-c-met, anti-erk, anti-phspho-erk, anti-akt, and anti-phspho-akt antibodies and reprobed with an anti-actin antibody to verify equal loading of protein. Inhibition of HGF-dependent or -independent cell proliferation by shmetexpressing recombinant Ad Hepatocyte growth factor/scatter factor (HGF/SF) is a specific ligand for c-met and has a markedly stimulatory activity on mortality of cancer cells but not on their growth 55. To determine whether c-met specific shrna inhibits the proliferation and scattering by HGF, I investigated the role of 95

HGF in U343 cell proliferation. After 24 hours Ad infection, serum free media with or without HGF/SF of 25ng/ml was added to cells and the effects of c- Met inhibition were tested. Increased cell scattering effect was observed in the untransduced and Ad- E1-transduced cells, whereas no scattering effect was observed in the Ad- E1-shMet4+5 transduced cells (Fig. 23A). Next, using MTT assay I investigated the role of HGF in cancer cell proliferation. As shown in Fig. 23B, HGF doesn t inhibit showing that reduction of Met expression can affect Met-dependent cell scattering without affecting cell viability or proliferation. Figure 23. Inhibition of HGF-dependent or HGF-independent cell motility by Ad- E1-shMet4+5. (A) U343 cells were plated in six-well plates to 20% confluency and were infected with Ad- E1-shMet4+5 at an MOI of 50. After 96

24 hr adenoviral infection, serums free media with or without 25 ng/ml of HGF were added to the cells and following 48 hours, viable cells were stained with crystal violet. (B) The reduced cell proliferation by the Ad- E1- shmet4+5 was measured by MTT assay. Viability of control cells was set at 100% and viability relative to the control is presented. Infected cells with Ad- E1-shMet4+5 reduced cell viability or proliferation with or without HGF. *P<0.05 versus Ad- E1 treatment. Cell scattering activity was suppressed in a Met-dependent manner (100x magnification). Enhanced anti-tumor effect of shmet-expressing recombinant Ad Recombinant Ad expressing c-met-specific shrna, Ad- E1-shMet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 were subsequently assessed for its ability to suppress the growth of U343 human glioma xenograft model established in nude mice. Tumors were generated by subcutaneous injection of cells into the mice abdominal region. When tumors reached an average size of 70 mm 3, tumors were injected intratumorally with PBS, Ad- E1, Ad- E1- shmet4, Ad- E1-shMet5, and Ad- E1-shMet4+5 (1 x 10 8 PFU) every 2 days for a total of three times (Q2Dx3). As seen in Fig. 24, the growth of all tumors treated with Ad expressing c-met-specific shrna was substantially delayed compared with that of tumors treated with PBS or with Ad- E1, which does not express c-met-specific shrna. By 45 days after treatment, tumors treated 97

with PBS reached an average tumor volume of 2771.6 ± 467.6 mm 3 and those treated with Ad- E1 reached 1468.1 ± 268.8 mm 3. In marked contrast, the tumor growth was severely suppressed in mice injected with Ad- E1-shMet4 (P < 0.05, versus PBS or Ad- E1 group), Ad- E1-shMet5 (P < 0.05, versus PBS or Ad- E1 group), or Ad- E1-shMet4+5 (P < 0.01, versus PBS or Ad- E1 group). More specifically, in the same time period those treated with Ad- E1-shMet4 reached 841.4 ± 244.2 mm 3, those treated with Ad- E1-shMet5 reached 785.1 ± 201.6 mm 3, and those treated with Ad- E1-shMet4+5 reached 433.3 ± 108.7 mm 3. Throughout the course of the study, no systemic toxicity, such as diarrhea, loss of weight, or cachexia was observed. The survival advantage conferred by Ad- E1-shMet4+5 was statistically significant when compared with either of the Ad- E1 group (P < 0.01, versus Ad- E1 group). 98

Figure 24. In vivo antitumor effect of Ads expressing the c-met specific shrna. Subcutaneous implanted tumors derived from U343 cells were injected with Ad- E1 ( ), Ad- E1-shMet4 ( ), Ad- E1-shMet5 ( ), or Ad- E1-shMet4+5 ( ), along with PBS ( ) as a negative control. Tumor growth was monitored at 2 day intervals after each injection. Arrows indicate when treatment was given 1 X 10 10 viral particles per 50 μl. Data points represent the mean (± SE) tumor size (in cubic millimeters) for each group at the indicated times. Values represent the mean ± SE. *, P < 0.05 versus Ad- E1. **, P < 0.001 versus Ad- E1. In Vivo histologic and immunohistochemical characterization Since Ad- E1-shMet4+5 is far superior to Ad- E1-shMet4 and Ad- E1-shMet5 in suppression of c-met expression, Ad- E1-shMet4+5 was chosen for the subsequent assessment in nude mice. To verify the proposed mechanism of action of c-met-specific shmet-expressing Ad, Ad- E1- shmet4+5-treated tumors were further investigated by histological examination. Tumors were harvested from each treatment group at 7 days following the three sequential treatments. As seen in Figure 25A, most of the tumor mass remaining after treatment with the Ad- E1-shMet4+5 was necrotic as shown by hematoxylin-eosin staining, whereas necrotic lesions were barely detectable in the tumors treated with PBS or Ad- E1. In addition 99

apoptotic changes and cell necrosis, surrounded by stromal fibrosis was also seen. Masson s trichrome staining determined collagen type I, stained in blue, as the predominant morphological component of these fibrotic changes. To determine whether the reduced size of Ad- E1-shMet4+5-treated tumors coincided with reduced neovascularization, microvessel density as assessed using anti-cd31 antibody was determined. A marked decrease in number and size of CD31-positive vessels was observed in Ad- E1-shMet4+5 Ad-treated tumors. Quantitation of microvessels showed that density was reduced by 79.8% (P < 0.001) and 76.7% (P < 0.001) in response to Ad- E1-shMet4+5 Ad treatment as compared to PBS and Ad- E1 Ad treatment groups, respectively (Fig. 25B). To analyze the degree to which Ad- E1-shMet4+5 induced apoptosis in vivo, TUNEL assay was carried out using the same tumor sections. As seen in Fig. 25A, apoptotic cells were also more abundantly detected in Ad- E1-shMet4+5-treated tumor tissue than in PBSor Ad- E1-treated tumor tissue. Next I investigate the biological significance of c-met downregulation in tumor tissue. When U343 tumors reached an average size of 70 mm 3, tumors were injected intratumorally with PBS, or with 1 x 10 8 PFU of either Ad- E1 or Ad- E1-shMet4+5 Ad every 2 days for a total of three times (Q2Dx3). Tumors were then harvested at day 10 and 20 after viral injection, 100

and c-met, VEGF, IL-8, and MMP-2 expression were measured by ELISA. As shown in Fig. 25B, following 10 and 20 days of treatment, a marked suppression of c-met expression was observed in tumor tissues treated with Ad- E1-shMet4+5, resulting in 56.4% and 20.9% reduction compared to Ad- E1-treated control tumors, respectively. Moreover, ELISA results demonstrated that treatment with Ad- E1-shMet4+5 resulted in 32.8% and 49.4% (P < 0.05 versus Ad- E1) reduction in VEGF and 76.7% and 76.7% (P < 0.05 versus Ad- E1) reduction in IL-8, at day 10 and 20 after viral injection. Also, MMP-2 expression reduced in tumors treated with Ad- E1-shMet4+5. These findings suggest that c-met-specific shrna down-regulates the expression of both IL-8 and VEGF, key mediators of angiogenesis and also, MMP-2 expression, appeared to play a crucial role in many stages of tumor progression, including angiogenesis and the invasion and metastasis of tumor cells. Therefore, the antimigratory and antiangiogenic activities of c- Met-specific shrna expression may inhibit tumor growth and metastasis in animals by suppressing tumor angiogenesis, suggesting a crucial role for c- Met in angiogenesis and tumor growth and metastasis in vivo. 101

Figure 25. Histological characterization of tumor tissues in U343 human glioma xenografts. (A) Hematoxylin & Eosin staining (H & E). Masson's trichrome staining of extracellular matrix (blue) in the tumor tissue sections 102

(M & T) and hematoxylin counter-stained. Blood vessel density was assessed by immunohistochemical staining for CD31; Hematoxylin counterstained (CD31). Brown staining indicates positive staining for endothelial cells. TUNEL staining of apoptotic cells in tumor tissue; methyl green counter stained (TUNEL). Tumors treated with Ad- E1-shMet4+5 exhibited significant increase in apoptotic levels. Greater dark brown nuclei with double-strand DNA breaks (indicated by an arrow) can be seen in tumors treated with Ad- E1-shMet4+5 as compared to PBS- or Ad- E1-treated groups. Original magnification: 40 and 400. (B) c-met down-regulation in tumor tissue. Tumor tissues were collected 28 days after final virus injection, and ELISA was done to determine the levels of c-met, VEGF, IL-8, and MMP-2. Treatment of Ad- E1-shMet4+5 Ad results in significant reduction in c-met levels as well as VEGF, IL-8, and MMP-2 as compared to treatment of PBS or Ad- E1. Each data point represents mean c-met, VEGF, IL-8 and MMP-2 levels for each individual tumor (8 mice per group). Each protein levels are expressed as picograms per milligram of total protein. *, P < 0.05 versus Ad- E1. **, P < 0.001 versus Ad- E1. Inhibition of tumor metastasis by shmet-expressing recombinant Ad Activation of angiogenesis is also responsible for increased tumor cell metastasis. Therefore, I investigated whether suppression of c-met 103

expression also blocks metastasis of a malignant human breast cancer cell line, MDA231. As illustrated in Fig. 26A, lung tissues from PBS- and Ad- E1- treated mice possessed approximately 7.6 ± 2.7 and 2.1 ± 0.7 tumor nodules caused by MDA231 metastasis. However, the number of tumor nodules was decreased to 0.4 ± 0.2 after administration of Ad- E1-shMet4+5. In fact, nine of thirteen mice in the Ad- E1-shMet4+5-treated group showed no metastatic spread. Hematoxylin and eosin staining revealed that the PBS- and Ad- E1- treated mice lung had a relatively large number of sizable tumor masses due to the metastasis of the MDA231 (Fig. 26B) and Masson s trichrome staining revealed that most tumor tissues treated with Ad- E1-shMet4+5 were surrounded by stromal fibrosis. Also, microvessel density as assessed using anti-cd31 antibody was determined. A marked decrease in number and size of CD31-positive vessels was observed in Ad- E1-shMet4+5-treated tumors. Quantitation of microvessels showed that density was reduced by 79.8% (P < 0.001) and 76.7% (P < 0.001) in response to Ad- E1-shMet4+5 Ad treatment as compared to PBS and Ad- E1 Ad treatment groups, respectively (Fig. 26C). These results indicate that c-met specific shrna-expressing Ad, Ad- E1-shMet4+5, had a strong inhibitory effect on tumor metastasis. 104

Figure 26. Therapeutic efficacy of Ad- E1-shMet4+5 on MDA-MB-231 lung metasis tumor model. (A) MDA-MB-231 human breast cancer cells were injected intravenously to form pulmonary metastatic lesions. At 28 days following cell injection, MDA-MB-231 tumor-bearing mice received intrapleural injections of PBS, Ad- E1, or Ad- E1-shMet4+5. On day 28 after viral injection, mice were sacrificed, lungs were removed and tumor nodules were counted to evaluate therapeutic efficacy. Mean of mice treated 105