Discovery and characterization of tubulin-interactive peptides

Tubulin is a heterodimer composed of an ? and a ? subunit. Polymerization of these dimers results in the formation of microtubules that are important in many cellular functions, such as cell division. Rapidly dividing cells, such as cancer cell, are very sensitive to compounds that disrupt microtubule biochemistry. Natural products have yielded several tubulin active poisons used in the clinic today. Recent exploration into marine natural products has produced several new molecules with tubulin activity. One such compound is vitilevuamide, a bicyclic 13 amino acid peptide, isolated from marine ascidians. Phage display is a combinatorial technique used to find short peptides that bind a desired target. We have isolated several peptides from a phage display library that bind tubulin. The focus of this study is to compare the natural product vitilevuamide and the peptides found by screening phage libraries. Although phage display may select some cytotoxic peptides, we hypothesized that many others would bind motifs that modulate tubulin function in a noncytotoxic fashion. The effect of the peptides on tubulin polymerization, aggregation, quenching of tubulin fluorescence and prevention of time- and temperature-dependent denaturation of tubulin was determined. The ability of these peptides to prevent drug-induced aggregation was also examined. Sequence similarity analysis was used to generate possible explanation for characteristics displayed by the phage-derived peptides. One phage display peptide isolated was found to have sequence similarity to a conserved region of tau protein. Another peptide had similarities to regions on tubulin that play roles in both lateral and longitudinal contacts in polymerization of microtubules. These studies have shown that phage display can produce peptides that have affinity for tubulin, but do not function like traditional cytotoxic natural product peptides.

DISCOVERY AND CHARACTERIZATION OF TUBULIN-INTERACTIVE PEPTIDES by Michael Charles Edler, Jr. A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Phannacology and Toxicology The University of Utah May 2003 Copyright © Michael Charles Edler, Jr. 2003 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Michael Charles Edler, Jr. This dissertation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. Chair: Louis R. Barrows f 7 William K. Nichols Gleii! R. Hanson THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: I have read the dissertation of Michael Charles Edler, Jr. in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable: (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. Louis R. Barrows Date Chair, Supervisory Committee Approved for the Major Department Approved for the Graduate Council ABSTRACT Tubulin is a heterodimer composed of an a and a p subunit. Polymerization of these dinlers results in the formation of microtubules that are important in many cellular functions, such as cell division. Rapidly dividing cells, such as cancer cell, are very sensitive to compounds that disrupt microtubule biochemistry. Natural products have yielded several tubulin active poisons used in the clinic today. Recent exploration into marine natural products has produced several new molecules with tubulin activity. One such compound is vitilevuamide, a bicyclic 13 amino acid peptide, isolated froIn Inarine ascidians. Phage display is a combinatorial technique used to find short peptides that bind a desired target. We have isolated several peptides from a phage display library that bind tubulin. The focus of this study is to compare the natural product vitilevuamide and the peptides found by screening phage libraries. Although phage display may select SOlne cytotoxic peptides, we hypothesized that many others would bind motifs that modulate tubulin function in a noncytotoxic fashion. The effect of the peptides on tubulin poI)'1nerization, aggregation, quenching oftubulin fluorescence and prevention of time- and temperature-dependent denaturation of tubulin was determined. The ability of these peptides to prevent drug-induced aggregation was also examined. Sequence similarity analysis was used to generate possible explanation for characteristics displayed by the phage-derived peptides. One phage display peptide isolated was found to have sequence sinlilarity to a conserved region of tau protein. Another peptide had similarities to regions on tubulin that play roles in both lateral and longitudinal contacts in polymerization of micro tubules. These studies have shown that phage display can produce peptides that have affinity for tubulin, but do not function like traditional cytotoxic natural product peptides. v TABLE OF CONTENTS ABSTRACT.................................................................................... IV LIST OF FIGURES........................................................................... VIII LIST OF TABLES............................................................................. Xl ACKNOWLEDGMENTS........ .. ...... . ...... .. ... .... ... ... .. .... ... .. . .. . .. . .. . .. . .. ..... XII Chapter 1. INTRODUCTION .................................................................... . Cancer............................ .................................................... 1 Tubulin Biochemistry ............. , ..... . .. ....... ... ... ........... . .. . ..... . .. .. .... 1 Tubulin and Microtubule Structure.. ...... .. ...... .... .. ... ... ... ..... . ... ........ 9 Microtubule Associated Proteins. .... ... ..... ... . ..... . .. . .. . .. ... . ..... . .. . .. .... 10 Tubulin Drug Binding Sites ..................................... '" ..... . ...... .. . . 15 Assays for Characterizing Tubulin-Ligand Interactions..... ............... ..... 20 Phage-Display........... ............................................................. 27 Research Objectives................................................................. 31 2. CHARACTERIZATION OF VITILEVUAMIDE.. ...... ................. ........ 33 Introductioll........................................................................... 33 Materials and Methods..... .. ...... . .. . ... ..... ......... ...... .. . . .. . .. .... .. . ... .. 34 Results............................................................................. .... 47 Discussiol1 ................................ '" .. , ................ , . '" .. ... . .. ...... ... 91 3. PHAGE DISPLAY SCREENING OF TUBULIN................................. 101 Introduction........................ .................. ...... ........................... 101 Materials and Methods........................................................... ... 101 Results and Discussion.............................................................. 114 Conclusion......... ... . ...... ... .... . . .. ....... ........ .. . ... . .. . .. . ..... ... ...... . .... 153 4. COMPARISON AND DISCUSSION OF TUBULIN INTERACTIVE PEPTIDES. ....... ......... ........ ............ .......... ..... ... ......... ...... ........ 156 Similarity of SSF Peptide. . ......... ...... ..... .......... .. . ..... . ..... ... ...... .... 161 Similarity R 1 and R2 Peptides... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 Conclusions....................................................................... .... 167 Future Studies ...................................................... ,. .. ... ...... ..... 168 REFERENCES........ ... ..... .... ... ............ ...... ........ ....... ........ ... .......... ..... 170 Vll LIST OF FIGURES Figure 1.1 Structure oftubulin heterodiIner at 3.7-A resolution, consisting of an alpha (right) and beta (left) subunit.. .. ...... .... ... .. . ... .. ... . .. . .. ... . ... .. . .. 2 1.2 Microtubule structures............................................................. 6 1.3 Organization of the human tau protein.. ...... ... ..... . ...... ... .. ... ... .... .. .. 13 1.4 Proposed position oftubulin drug-binding sites................................ 16 1.5 Life cycle of filalnentous phage.. . .. ...... ........ .... .. ... .... .. ... . .. ... ... .... 29 2.l Structure ofvitilevuamide........................................................ 35 2.2 SDS-polyacrylamide gel electrophoresis of phosphocellulose-purified tubulin............................................................................. ... 38 2.3 Effect of vitilevuamide on cell cycle progression ........... '" ... .. . ..... . .. .. 49 2.4 Dose-dependent in vitro inhibition of glutamate-induced phosphocellulose purified tubulin polymerization.... .. ...... . .. .... .. . .. ...... 52 2.5 ICso calculation of the inhibition of PC tubulin polymerization by vitilevualnide ............... , ......... ...... . .. .... ...... .... . . ........ . .. ...... . ... . 54 2.6 Effect ofvitilevuan1ide on the critical concentration oftubulin.......... .... 57 2.7 Induction oftubulin aggregation by vitilevuamide, dolastatin 10, and vinblastine. . ........ . .. . ..... . ... ..... . ........ . .. . .. . ... ...... .. . ... . .... . ... ... .. . . 61 2.8 Inhibition of viti levu amide-induced tubulin aggregation by low doses of dolastatin 10......................................................................... 64 2.9 Inhibition of viti levu amide-induced tubulin aggregation by low doses of vinblastine.. ... ......... ...... ... .. .... ..... .... ... ...... .. .... .. . ...... .. . .. . ...... . 66 2.10 ICso calculation of dolastatin 10 and vinblastine inhibition of 68 vitilevuamide induced tubulin aggregation ............................. '" ..... . 2.11 Inhibition of dolastatin 10 induced tubulin aggregation by low doses of vitilevllamide. ......... .... ..... .... .. ... ... . ..... ............ ......... . .. . ..... . ... . 72 2.12 ICso calculation of vitilevuamide inhibition of dolastatin 10 induced tubulin aggregation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 75 2.13 Prevention of time and tenlperature dependent denaturation of tubulin by vitilevuamide........................................................................ 78 2.14 Effects of vitilevuamide on the fluorescence oftubulin....................... 81 2.15 Effect of vitilevuamide on vinblastine or dolastatin 10 binding to tubulin. 84 2.16 Inhibitory effects of vitilevuamide, hemiasterlin, and vinblastine on the binding of radio labeled GTP to tubulin ................................... '" .... 86 3.1 Set-up for screening phage libraries against tubulin........................... 105 3.2 Screening oftubulin column with 15mer phage library.................... ... 108 3.3 Enrichment of tubulin screened library.......................................... 115 3.4 Titering of individual clones from tubulin screen.............................. 117 3.5 Enhancement of MAP tubulin polymerization by SFF peptide.......... ..... 125 3.6 Possible induction oftubulin aggregation by SFF peptide.................... 129 3.7 Possible enhancement of vinblastine and dolastatin 10 induced aggregation by SSF peptide. . ... ... ...... .. . ...... ..... . ..... . .. . .. ... . .. . ..... . ... 132 3.8 Possible prevention oftinle and temperature dependent denaturation of tubulin by SSF peptide.. ..... . .. . ..... . ... ..... . .. . .. . .. . ........ ... . ... .. .... .. ... 136 3.9 Tubulin polymerization inhibition by SSF peptide. ...................... ..... 140 3.10 Effect ofRl and R2 peptide on tubulin aggregation........................... 144 3.11 Inhibition of dolastatin 10 induced tubulin aggregation by Rl peptide...... 147 3.12 Inhibition of dolastatin 10 induced tubulin aggregation by R2 peptide.. .... 149 IX 4.1 Sequence alignn1ent for A. SSFGVSFVKQA VRPR and B. GSAL VGFA.. 163 4.2 Location ofR1 and R2 peptide similarity located on the alpha subunit of tubulin ................................................................ , .. ... . .. .... . ... 165 x 1.1 2.1 LIST OF TABLES Major Characteristics of Tubulin Binding Agents ................... , ......... . Best-fit Values for the Extent ofPolYll1erization and Rate of Polymerization of Phosphocellulose Purified Tubulin Treated with Vitilevual11ide ....................................................................... . 2.2 Reduction in the Extent and Rate of Viti levu amide-Induced Tubulin 22 56 Aggregation by Dolastatin 10 and Vinblastine.................................. 70 2.3 Reduction in the Extent and Rate of Dolastatin 10-Induced Tubulin Aggregation by Vitilevuamide..................................................... 74 2.4 Stabilization of Colchicine Binding Activity of Tubulin by Vitilevuamide, Vinblastine, and Hemiasterilin. . ... ...... .. . .. ... . ..... ... ... ..... . ... .. . ... ..... .. 89 2.5 Antitumor Activity of Viti levu amide Against Murine P388 Lymphocytic Leukenlia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.6 Major Characteristics ofTubulin Binding Agents Compared to Vitilevuamide.... .... .. ... . ... ..... . ... .. ... . ..... . .. . ... .. .... ..... ... . ..... . ..... ... 99 3.1 Quantitation of Phage Binding to Tubulin and Milk Protein................... 120 3.2 Sequences of Phage Isolated from Tubulin Screen......................... ..... 121 3.3 Effect of SSF Peptide on Tubulin Polytl1erization Extent and Rates......... 127 3.4 Effect ofRl Peptide on Tubulin Polymerization Extent and Rates........... 142 3.5 Reduction in the Extent and Rate of Dolastatin 10-Induced Tubulin Aggregation R1 and R2 Peptides............................................... .... 151 3.6 Major Characteristics of Tubulin Binding Agents Compared to VitilevuaIl1ide, SSF Peptide, Rl and R2 Peptides............................. 154 ACKNOWLEDGEMENTS I would like to extend my sincere appreciation to my advisor, Dr. Louis R. Barrows, for his guidance, encouragement, support, understanding and friendship during this research endeavor. I would also like to thank n1y thesis committee lnelnbers for their time and constructive criticisn1 of this manuscript and suggestions during my research. I am grateful to the members of my laboratory who made graduate work both interesting and fun. A special thanks to Chris Pond for his helpful suggestions and ideas during my research. I would also like to thank Kate Marshall for her friendship and help in editing this manuscript. I am also very grateful for the support and friendship of A.1. Baucum, who helped me get through graduate school unscathed. Special thanks to my good friends leffHardin and Tiffany Potter for allowing me to step away from my research and enjoy life. Finally, I an1 deeply grateful to my mother, Patricia Edler, and my father, Michael Edler Sr. for the continual support and love throughout this process. I would also like to thank my brother and sisters for their support and encouragement. Without them, this endeavor would have had a lost meaning. CHAPTER 1 INTRODUCTION Cancer According to the National Vital Statistics Report, cancer was the second leading cause of death in the year 2000 (l). Approximately 550,000 Americans died in the year 2000 fronl malignant neoplasias. Surgery conlbined with radiation therapy has been successful in treating about 30 % of cancer patients. Many patients still require chelnotherapy to treat their cancer and the development of new agents has begun to provide useful tools in this treatnlent. Discovery of new agents that possess unique mechanisms of action or new potencies against refractory cancers will be necessary for success in this endeavor. Many chemotherapy treatments utilize agents that target the intracellular protein tubulin in treating many cancers. This dissertation focuses on peptide or depsipeptide agents that target tubulin and are prototype drugs for the treahnent of cancer. Tubulin Biochenlistry Tubulin is a 100 kDa heterodimeric protein consisting of an alpha subunit (50 kDa) and a beta subunit (50 kDa) (2-4) (Figure 1.1). Polymerization of these 2 Figure 1.1. Structure oftubulin heterodimer at 3.7-A resolution, consisting of an alpha (right) and beta (left) subunit. The alpha subunit contains a site for both zinc and l11agnesium ion binding, as well as a site for GTP binding at the nonexchangeable site (N­site). The beta subunit contains the binding site for the drug taxol, as well as a site for GTP binding at the exchangeable site (E-site). Structure was obtained from Nogales et al. (9). Magnesium Ion GTP (N-site) Taxol Zinc Ion GDP (E-site) B-Tubulin a-Tubulin V.J heterodiIllers by the hydrolysis of GTP to GDP results in the assembly of nlicrotubules. Microtubules are dynanlic structures involved in nlany cellular processes. These include maintaining cell structure, facilitating cell division, transport of vesicles along axons and dendrites of nerve cells, and transport of other cellular organelles throughout the cell. Microtubules accomplish l11any of these tasks by switching between states of growing, rapid shortening (catastrophe) and regrowth (rescue), what is also known as dynamic instability. Microtubules can remain in a steady state of unidirectional movenlent whereby subunits are moved through the microtubule by polymerizing predonlinately at one end of the l11icrotubule and depolymerizing at the other end of the microtubule, a process known as treadmilling. While dynanlic instability is a nl0re general property of microtubules in the cell, treadmilling of the nlicrotubule allows for the separation of chromosomes during cell division. GTP Binding The dynamic properties of microtubules described above are derived frolll the binding and hydrolysis of GTP. Each monomer of the tubulin heterodinler binds a GTP nlolecule in a separate and unique fashion (5-7). On alpha tubulin, GTP binding occurs at the monomer-monomer interface within the dimer. This site is commonly referred to as the N-site because the GTP is nonexchangeable and always remains in the GTP form. Removal of the GTP molecule has been accolllplished only by denaturing the tubulin heterodimer, rendering it inactive (8-10). A magnesium ion is also bound to the N-site and is important in preserving heterodimer integrity (11). p-tubulin binds GTP at a site known as the E-site or exchangeable site. GTP binds to the E-site and facilitates 4 5 polymerization by hydrolysis of the GTP to GDP after the addition of another dinler. The GDP then becomes nonexchangeable and the two dinlers are polymerized (12). The tubul in dimers continue stacking on top of each other to form long single chain protofilaInents (Figure 1.2A). These protofilaments associate with each other to form nlicrotubules (discussed below). Microtubule stability is a tightly regulated process and the maintenance of a GTP cap on the end of the microtubule is extren1ely important in maintaining stability (13). Experinlents have shown that microtubules only require one GTP-tubulin cap per protofilament to maintain stability. When a microtubule loses this GTP cap, it rapidly depolymerizes. The ability of the microtubule to lose its GTP cap is central to the dynamic instability properties of micro tubules (Figure I.2B). This model oftubulin dynamic instability is supported by results utilizing nonhydrolyzable analogs of GTP. Tubulin can polymerize into microtubules when incubated with nonhydrolyzable analogs ofGTP, but because the formed microtubules do not lose their GTP cap they no longer possess their dynamic properties (14). Sulfhydryl Content Tubulin dimers contain lnany sulfhydryl groups (8 cysteine residues in p-tubulin, ] 2 in a-tubulin) that can be alkylated (3,4). Even snlall amounts of alkylation cause the inactivation of tubulin heterodimer polymerization (15-17). These cysteines are heterogeneously distributed throughout the heterodinler and have been exploited in order to examine confomlational changes in the heterodimer. Cysteine pairs on the p subunit can be cross-linked with 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB). Other cysteines are divided into slow and fast reacting groups. DTNB is now used to measure the effects of 6 Figure 1.2. Microtubule structures. A: Microtubule protofilament comprised oftubulin heteroditners stacked end to end with a polymerizing plus end and depolymerizing minus end. a-tubulin is shown with a nonexchangeable GTP and ~-tubulin containing GDP where dimer stacking has occurred. Plus-end ~-tubulin contains a GTP cap. B: Rapidly depolymerizing microtubule after Joss of GTP cap fron1 the polymerizing end. ProtofilaInents in nlicrotubule are arranged in a ~-lattice with an a-tubulin of one protofilament in contact with a ~-tubulin of the adjacent protofilaInent and vice versa. C: Ring structure of microtubule conlprised of 13 proto fi laments. 7 A. - + B. c. 8 antilnitotic drugs on the overall conformation of the tubulin dimer. Colchicine and many of its analogs have been shown to decrease DTNB reactivity with tubulin and completely protect approximately 1.4 cysteine residues (18). Ce11ular Role of Microtubules During interphase, nlicrotubules play an essential role in the formation and maintenance of the cell cytoskeleton. Microtubules interact with both intermediate fiianlents and microfiiaments, conlposed of actin, to accomplish this. However, these structures are not static and are in constant flux between states of growing and shortening that can be seen by microscopy. Upon entering into mitosis, the cell disassembles n10st of the nlicrotubules involved in maintaining the cytoskeleton and transports them to the Initotic spindle. The mitotic spindle is highly dynamic (20-fold more than cytoskeletal microtubules) and undergoes rapid changes in length that are required for chromosonle segregation (19, 20). Nucleation of micro tubules requires a microtubule-organizing center (MTOC) that is composed ofy-tubulin. y-tubulin shares high sequence homology with both Ct.- and ~-tubulin and forms a ring structure at theMTOC that serves as a tenlplate for microtubule growth (21). Microtubules are polar in nature, with a non­growing minus end (depolymerizing) attached to the centrosonle/MTOC, both during interphase and mitosis. The growing plus end (polytnerizing) is attached to the kinetochore of the chronlosome during nlitosis and attached to the cell periphery during interphase. 9 In Vitro Tubulin Polymerization Microtubules formed in vitro require the presence of both GTP and microtubule associated proteins (MAPs), the latter of which are discussed later in this introduction (22,23). MagnesiUln is not required, but its addition enhances polymerization (24). Other compounds can induce polymerization and include dimethyl sulfoxide (25), glycerol (26), and glutamate (27). Induction oftubulin polymerization with such C0111pounds allows for the examination oftubulin polymerization in vitro in the absence of MAPs. In vitro tubulin polymerization is also very tenlperature sensitive and polynlerization does not occur at 4°C. As the temperature is raised, the rate of polymerization increases with an optimal polymerization temperature at 37°C. Addition or subtraction of dilnethyl sulfoxide, glycerol and glutamate will affect the rate of polynlerization. Adding microtubule-stabilizing drugs such as taxol will increase the rate of polynlerization even at lower temperatures (28). Tubulin and Microtubule Structure Microtubules are generally 25 nm in diameter and vary in length. Microtubules fomled in vitro can reach lengths of up to 20 ~un, and can exceed 1 00 ~nl in vivo, particularly in axons (29). Microtubules are composed predominately of 13 parallel protofilaments (30) (although nlicrotubules composed of as little as 10 and as nlany as 15 have been seen) arranged in a ~-lattice structure (31) (Figure 1.2B/C). This ~-lattice structure is created by offsetting each protofilament such that an a-subunit of one protofilanlent is bound to a ~-subunit of the adjacent protofilament and vice versa. Approximately 1670 a-~ heterodinlers per ~m of microtubule are in this configuration. 10 The structure of Inicrotubules was originally examined by electron n1icroscopy of negatively stained san1ples. The general arrangement of tubulin subunits could be seen, but the technique resulted in artifacts (structures that do not resemble Inicrotubules) from the staining and had lin1ited resolution. The advent of cryoelectron microscopy has provided a lnore understandable picture of microtubule arrangement and the rapid freezing involved in this technique has provided microtubule samples in varying structural states (32). This analytical technique has given insight into the necessary structural states required during both asserrlbly and disassen1bly of n1icrotubules in the cell. Elucidation of the crystal structure of tubulin at 3.7 A has provided even more insight into the interaction oftubulin dimers in microtubules. Nogales et al. were the first to accomplish this in 1998, and over the past few years molecular modeling has attempted to detemline precise amino acid contacts between tubulin heterodimers during polymerization (9). Modeling of the microtubule structure was accomplished by docking the crystal structure oftubulin at 3.7 A with a 20 A map of the microtubule obtained by cryoelectron microscopy. This model has provided new insights into the precise mnino acid contacts required for microtubule polymerization. Microtubule Associated Proteins Microtubule-associated proteins (MAPs) represent a diverse set of multifunctional proteins that exert unique effects on the structural and functional propeliies oftubulin. MAPs can be generally classified into three groups according to their function, but SOlne proteins n1ay share multiple properties. 11 Microtubule Motor Proteins The first group of MAPs is the nlicrotubule nl0tor proteins. These proteins nlove along the nlicrotubule either towards the plus end (kinesins) or the minus end (dyneins). These proteins are involved in the transport of organelles and other macromolecules around the cell. Some proteins in this family are involved in aiding the separation of chronl0somes during mitosis, such as the centromere-associated protein E (CENP-E) (33- 36). Most nlotor proteins do not affect microtubule dynamics, but others facilitate the organization of MTOCs away from the centrosonle (37). Microtubule Destabilizers The second group of MAPs is the microtubule destabilizers. Katanin is an ATPase that possesses Inicrotubule-severing activity and it is localized to the centrosome (38). Katanin is composed of two subunits that serve two unique but inlportant functions. The first subunit, p60, is the ATPase involved in microtubule severing, but is only active in the absence of the second subunit, p80. In addition, p80 localizes katanin to the centrosonle due to the presence ofWD40 repeats in its structure (39). Another nlicrotubule destabilizer, Op 18-stathmin, binds to tubulin dimers and increases microtubule catastrophe rate (40). Op 18 directly hydrolyzes GTP at the plus ends of microtubules and relTIOVeS the GTP cap and causes the microtubule to depolynlerize (41). Microtubule Stabilizers The last group of MAPs is classified by their ability to copolymerize or co localize with tubulin during polYITIerization. Their specific function appears to be related to 12 stabilization of the microtubule architecture and promoting its polymerization. Microtubule stabilizers are generally positively charged proteins that are thought to act by shielding the negatively charged tail of the tubulin dilner. There is some dispute about the exact location and nlanner of nlicrotubule binding and whether the binding site overlaps with the binding site of microtubule nlotor proteins. Microtubule stabilizer binding is regulated by phosphorylation. Structurally, the tertiary structures of microtubule stabilizing proteins are poorly understood. Removal of these proteins by phosphocellulose chrOlnatography abolishes the ability of tubulin to polymerize in vitro, even in the presence of GTP. MAPs 1, 2 and 4 all help prOlTIote the polymerization of tubulin, along with tau protein. Tau is one of the best characterized of the MAPs and consists of a collection of altenlatively spliced proteins ranging between 37 to 45 kDa in size (Figure 1.3). RNA splicing and differential phosphorylation results in the expression of In ore than 60 different tau isofonns (42). Tau is mainly found in neuronal cells, but it can also be found in astrocytes, oligodendrocytes, and nluscle cells. Tau appears to be extremely ilnportant in neural development and isofoffil expression increases as brain development progresses. There is significant interest in tau due to its abnoffilal association with Alzheinler's disease. Degradation of tau occurs via calciunl-dependent protease activity (43). Figure 1.3 represents the mnino acid organization of hUl11an tau protein with phosphorylation sites, alternative RNA splicing regions, and nlicrotubule-binding domains. Functional studies have shown that tau promotes tubulin polymerization by lowering the critical concentration needed for polymerization and by preventing the Figure 1.3. Organization of the human tau protein. The C-ternlinal region (right side) where tubulin-binding repeats are located. Also shown are the regions generated from alternative splicing and areas of phosphorylation sites. The arrow indicates the tubulin­binding hotspot. 13 • • • /0 441 lr?I}:::.. : F..7. ..:. ::.;.: ..:..:. ::. .:...; .;: ..;.. ::.;r ...:..,. ,:..: ..;. .:;.;:. .;.. ::.;; ..." r··;.';·~: ··.·· ·'·'··1- \1 ... ··r::::::f::,:,::::::::.:·::·:,li:!:.:.:::1iiiii!i .- ·~c • • •• _ •••• 0 o •• • Tubulin-Binding Repeats TBH= Tubulin Binding Hotsoot Alternative RNA Splicing Regions Resulting in Different Tau Isoforms Phosphorylation site (Serine/Threonine Kinase/ Other Kinases) Determined by NetPhos 2.0. .--. .J;;:.. 15 depolymerization of fonned microtubules. Tau is thought to stabilize Inicrotubules in the cell, which is demonstrated by transfection studies where the addition of tau to fibroblast cel1s promoted both assembly and maintenance of micro tubules (44, 45). In vitro, tau binds to nlicrotubules, facilitates its nucleation and elongation, shields against disassenlbly, and induces thetn to foml bundles. Binding of tau to tubulin is facilitated through several conserved repeat regions in the C-tenninal portion of tau. This region of tau varies in the number of repeats, ranging from one to four. Binding of tau to nlicrotubules requires both the repeat donlain plus the regions just outside the repeat donlain (46, 47). These exterior regions of tau act like ')aws" to hold the tubulin dimers together. The regions outside the repeat "jaws" domains appear to pronl0te bundling of tnicrotubules (48). Although the repeat dmnains possess moderately weak microtubule binding, there are several 'hotspots' (e.g., the region between repeats 1 and 2) (Figure 1.3), which could regulate affinity of tau for microtubules, particularly during the developmentally regulated alte111ative splicing of tau (49). Tubulin Drug Binding Sites Studies of tubulin-binding drugs date back to 1968 when tubulin was found to colocalize with radioactive colchicine (5). Today, hundreds of compounds and their analogs are known to bind tubulin and alter its biochemistry. Even with all the newly found compounds that bind to tubulin, only five or six distinct sites for binding have been identified (Figure 1.4). The colchicine, taxol and vinca sites are where all the clinically relevant drugs bind to date. 16 Figure 1.4. Proposed position of tubulin drug-binding sites. Map of tubulin dimer with a (bottOln) and p (top) subunits. The vinca domain (V) is proposed to be located on the top of the p-subunit near the polymerizing end and overlapping the exchangeable GTP site (E). This domain contains three distinct sites for tubulin drugs. The taxol (T) site was detennined by x-ray crystallography studies and may also contain a site nearby for the drug laulimalide (L). The colchicine site (C) is located at the intradinler interface near the nonexchangeable GTP binding site (N). V-Site ~E-Site L-Site T-Site C-Site N-Site (3- Subunit u­Subunit 17 18 Colchicine Site As stated earlier, the colchicine site was the first tubulin drug-binding site discovered and photo-cross-linking studies later found that its binding pocket is n10st likely at the monomer-monomer interface (50). Many agents have now been found to bind at the colchicine site. Most of these cOlupounds are simple in structure, but as a group are very structurally diverse. They include the combrestatins (51), podophyllotoxin (52), benzoylphenylureas (53), stegnacin (54), the curacins (55), and 2- nlethoxyestradiol (56). None of these agents have been shown to interfere with the exchange of GTP at the E-site of tubulin (57). Interestingly, many of these agents induce a GTPase reaction apart from microtubule assembly (58). Taxol Site Agents binding the taxol site on tubulin actually promote the polymerization of tubulin and comprise those conlpounds that belong to the family of compounds known as taxoids. In vitro these agents promote the hyperassembly of microtubules with improved stability. In cells, these agents generally produce shorter, more bundled microtubules with increased cellular density. These agents do not alter nucleotide exchange, and the presence of an exchangeable GTP is not required to induce assembly (57). However, adding GTP does increase the overall polymerization of micro tubules in the presence of taxol drugs (59,60). Crystal structure analysis oftubulin co-crystallized with taxol has given direct evidence of the taxoid-binding site. Taxol binds near the M-loop in ~-tubulin in a hydrophobic pocket at a point of lateral interactions between inter-dimers (9, 61). In a-tubulin, an additional eight anlino acids occupy this hydrophobic pocket. Taxol is 19 thought to possibly mimic the extra eight anlino acids of a-tubulin when bound to p­tubulin and may provide additional stability to the nlicrotubule. Many new conlpounds with clinical potential as anticancer agents bind in this site and include discodermolide (62), eleutherobin (63), the epothilones (64), and the sarcodictyins (60). Another compound, laulilnalide, also prolnotes the fonnation of microtubules like taxol, but does not bind in the taxol site (65). When incubated together, both paclitaxel and laulimalide bind to microtubules in near stoichiometric anlounts relative to tubulin content (65). The binding of one drug (taxol) to the nlicrotubule does not prevent the binding of the other drug (laulimalide), indicating separate binding sites. The site where laul inlalide binds may help identify agents that are effective against cancers that are resistant to paclitaxel treatment. Vinca D0111ain Another site oftubulin drug binding contains three distinct subsites in close proximity on the tubulin dimer. This site is known as the "vinca domain" because all agents that bind in this "donlain" inhibit the binding of radiolabeled vincristine and vinblastine to tubulin. Vinca alkaloid compounds like vinblastine, vincristine, and vinorelbine all bind to residues 175-213 in p-tubulin resulting in a disruption of longitudinal contacts between tubulin dimers and distorted protofilament structures incapable ofpolynlerization (66). Vinca site agents bind within the "vinca domain" and c0111petitively inhibit the binding of other vinca site agents (67). The compounds rhizoxin and nlaytansine also belong in the class of vinca site agents. 20 Peptide Sites Two additional sites are also located in this "vinca domain" that noncompetitively inhibit the binding of vinca site agents. Noncompetitive inhibition indicates that agents that bind the two additional sites displace vinca site agents but do not bind to the same site. The first site is cOlnmonly referred as the "peptide-site" and was originally characterized by Ernest Hamel using Phomopsin A and dolastatin 10 (67). Recently, it has been shown that a second site exists that nonconlpetitively inhibits both vinca site agents and peptide site agents. Spongistatin was found to exhibit this trait and comprises the only confirmed member of a class of drugs that noncompetitively inhibits both vinca site agents and peptide site agents (68). Data from our laboratory suggest that vitilevumnide 111ay also bind this site, but this remains unconfirmed. Assays for Characterizing Tubulin-Ligand Interactions Compounds that bind to tubulin exert unique effects on the structure and biochenlical properties of the tubulin dilner. These effects can be as profound as the inhibition or promotion of polymerization, or less consequential effects such as quenching oftubulin fluorescence. The following section will attempt to explain several of the il11portant assays used to elucidate the propel1ies of cOlnpounds that bind to tubulin. Sonle of these assays, such as in vitro polynlerization and tubulin aggregation, provide data about the effects of drug binding on tubulin function. Other assays, such as protection of tubulin from denaturation and protection of tryptophan fluorescence, provide specific data about drug effects that might not correlate with pharnlacologic activities, but nevertheless permit comparison mnong tubulin drugs. The activities of a 21 selection oftubulin interactive drugs relevant to this work are presented in Table 1.1. In vitro Polymerization Assay The most widely used and significant assay of tubulin function is the in vitro polymerization assay (69). As stated earlier, purified tubulin can undergo polymerization in vitro in the presence of GTP at 37°C. Polymerization can be readily monitored by changes in turbidity. For this assay, turbidity is measured on a spectrophotometer at 350 nm. The 350 mn wavelength was chosen to increase the sensitivity of the assay. A shorter wavelength would result in the interference from absorption oftubulin and GTP. Turbidity is an approximation to the total light-scattering cross section (70). When the particles being measured are smaller than the wavelength of light used, the turbidity possesses an inverse fourth-power dependence on the wavelength and is a function of the size and shape of the scattering particles. In the case of nlicrotubule formation, the rods being formed can be considered very long in relation to the wavelength of light being used. In this situation, the turbidity is a function of the total mass concentration of the scattering particles (70). When tubulin polytnerization occurs the long rod-like microtubules cause the light to scatter, resulting in an increase in absorbance readings. This increase in absorbance can be attributed to an increase in microtubule or aggregate content in solution. Agents that bind tubulin and inhibit its polymerization prevent this light scattering. The in vitro tubulin polymerization assay provides a rapid detennination of a compound's ability to inhibit tubulin polymerization and provides evidence that tubulin is the target protein. In vitro polytllerization assays can also identify agents that prOlnote tubulin assembly such as taxol. Compounds that bind tubulin, but do not affect Table 1.1 Major Characteristics of Tubulin Binding Agents Drug Inhibition of Induces Inhibits Inhibits Tubulin Aggregation Vinblastine Dolastatin 10 Polymerization Induced Tubulin Induced p.tM (ICso) Aggregation Tubulin Aggregation - Do lastatin 10 0.59 Yes Yes -- Phomopsin A 2.8 Yes Yes Yes Cryptophycins 1.0-5.0 Yes Yes Yes - Hemiasterlins 0.98 Yes ND Enhances r S pongi statin 3.6 No ND Yes Taxol Promotes Yes ND ND Stabilizes Tubulin Denaturation Yes Yes Yes - N/D ND ND Protects Tryptophan Fluorescence Yes Yes Yes N/D ND ND 1'0 1'0 23 tubulin polymerization will not provide a positive result in this assay. These compounds may still have an effect on tubulin properties; other experiments are needed to nleasure these effects. Deternlination of the Critical Concentration of Tubulin Polymerization The in vitro tubulin pol)'lnerization assay described above requires a mininlum concentration of protein to produce an observable effect. This concentration of tubulin is comnl0nly referred as the critical concentration of tubulin polyn1erization. Using the in vitro tubulin polytnerization assay the critical concentration is determined by measuring the amount of polytnerization of several different concentrations of tubulin and plotting the concentration oftubulin versus anlount of polymerization. Drawing a regression line through the points on the graph and calculating the x-intercept provides the critical concentration. When tubulin is polymerized without any inhibitors, the critical concentration of tubulin normally ranges from 2-7 JlM. Most compounds that inhibit tubulin polymerization will cause an increase in the critical concentration of tubulin. The in vitro polynlerization assay can thus be used to determine whether a compound increases the critical concentration of tubulin polymerization or causes substoichiOlnetric reductions in the maxinlum extent of asserrlbly. Tubulin Aggregation Assay Another property oftubulin-binding drugs is the ability to induce tubulin aggregation. Aggregation oftubulin occurs Inainly by the association of the hydrophobic 24 regions on one tubulin heterodinler with another tubulin heterodinler. Tubulin active agents bind to the heterodinler and induce structural changes that result in the exposure of some of these hydrophobic groups and facilitate aggregate formation. Many compounds, such as vinblastine, vincristine, cryptophycin, and dolastatin 10 induce the formation of tubulin aggregates at high concentrations. Aggregate formation, like microtubule fonnation, can also be lTIonitored by a light scattering assay. Drug-induced tubulin aggregation can occur in the absence of GTP and therefore is different from microtubule fomlation, which requires GTP. Not all conlpounds that interact with tubulin and prevent polymerization pronlote the fonnation of aggregates. This assay allows for the discrinlination of COlTIpounds that do or do not aggregate tubulin. However, this assay does not allow for the quantitation of the size or structure of the aggregates. Some compounds will produce large aggregates while other compounds produce much smaller ones. If the aggregate size is too snlall, the light scattering assay cannot detect thenl. Electron microscopy nlust be used to determine the nature of the aggregate fonllatiol1. It is of some interest that all agents found to date that cause tubulin aggregation also prevent the time-dependent decay of colchicine binding (71). Inhibition of Drug-Induced Tubulin Aggregation The aggregation process discussed above can also be inhibited when a second tubulin drug is added to the incubation, but at lower concentrations. Inhibition most likely occurs due to conlpetition for the binding site by both drugs, reducing the aggregation effect of the high concentration agent. These types of experiments can lend a better insight into similarities of the binding characteristics of the two agents. 25 Bis-ANS Binding Assay Tubulin in solution undergoes structural changes (denaturation) that are both time-dependent and tenlperature-dependent (72). 4,4'-dianilino-l, 1 '-binaphthyl-5, 5'­disulfonic acid (bis-ANS) is a fluorescent hydrophobic probe that binds tightly to tubulin and inhibits polymerization (73). As tubulin begins to denature, hydrophobic regions begin to appear, allowing for n10re than one site for binding ofbis-ANS. It should be pointed out that denaturation oftubulin in solution over time is different from the exposure of hydrophobic regions due to the binding of high concentration of a tubulin drug that results in aggregation. Upon binding to these hydrophobic regions, bis-ANS fluorescence is greatly enhanced. The more hydrophobic regions available, the greater the fluorescence increases due to increased bis-ANS binding. Drugs that bind tubulin and stabilize these structural changes show a marked decrease in fluorescent intensity from bis-ANS binding over tilne (74). The bis-ANS binding assay can determine if the drug stabilizes tubulin denaturation. Tubulin Fluorescence Quenching Assay When intrinsic tubulin tryptophans and tyrosines are excited at 284 nm it results in a characteristic emission spectrum. Ligands that bind to tubulin and quench any of the surface-exposed tryptophans or tyrosines show a red shift in the emission spectra of tubulin. Vinca site agents have been shown to quench approximately three surface­exposed tryptophans on tubulin regardless of their ability to form tubulin aggregates. Colchicine site agents do not have this property (75). By binding a ligand to tubulin and exciting the tryptophans and tyrosines, one can begin to classify the sites where the ligand mayor may not be binding. Agents that do not quench any tryptophans or tyrosines can be ruled out as possible vinca-site agents. This assay cannot determine if the drug being tested binds near the colchicine site. Ligand Binding Assays Drug Binding Sites 26 Ligand binding assays are used to determine the relative affinity of two tubulin­binding drugs. Binding assays can also determine whether one drug inhibits the binding of a second drug to the tubulin dimer. If inhibition does occur, the relationship of the inhibition is determined by plotting the data on a Lineweaver-Burke (double-reciprocal) plot. If the lines on the plot all intercept at the y-axis the relationship is conlpetitive and if the lines all intercept at the x-axis the relationship is noncompetitive. Sometimes a drug nlay inhibit the binding of another drug but the binding relationship cannot be determined. These binding assays can provide information about an unknown drug's binding site by comparing its binding relationship to other known drugs for tubulin, such as the colchicine site, the vinca site, and the peptide site. GTP Binding As indicated above, tubulin polymerization requires the binding of GTP to the beta subunit of tubulin. Some agents that bind to tubulin will prevent GTP binding and result in an inhibition oftubulin polymerization. Studies can be performed in order to determine if the drug being tested inhibits the binding of GTP. This is accomplished by measuring the ability of a drug to inhibit [8-3H] GTP binding to tubulin. Perfonning 27 these binding studies can determine if inhibiting GTP binding is one mechanisnl of action for inhibition of pOl)'ll1erization for these drugs. Colchicine Stabilization Assays are perfomled to determine whether a drug stabilizes colchicine binding to tubulin when coincubated. Many drugs that do not inhibit colchicine binding to tubulin will stabilize colchicine binding. Not every tubulin interactive drug stabilizes colchicine binding so this assay will determine if the drug possesses this property. Phage-display Screening with phage display libraries or biopanning, is a powerful cOITlbinatorial technique used to discover a vast array of binding peptides or antibodies for a particular target. It was first discovered in 1985 by George P. Smith at the University of Missouri (76). An entire issue of Gene (volmne 128, number 1) was dedicated to the examination of this powerful technique. Phage-display technology is the result of two simple but powerful concepts. First, an insertion mutation into the 5' end of one of the viral coat protein genes will lead to the expression of the corresponding peptide sequence on the surface of the phage virus, as long as the mutation does not cause a disruption in any essential gene products. Second, insertion of a randOln oligonucleotide library into this locus results in the creation of a library of phage-displayed peptides that corresponds to the sequences of the oligonucleotide inserts. Utilizing proper screening techniques, one can isolate a single phage clone and purify its DNA for sequence analysis to reveal the sequence of the 28 inserted peptide. Certain lilnitations do exist in such a system and are focused on the ability of the phage coat protein to tolerate the peptide insert. Phage particles tolerate some inserts better than others and those inserts that result in fatal mutations or mutations that slow the viral replication too much will be lost from the library. Other mutations that prOlnote replication or provide an advantageous property to the phage particle can result in a biased library that will be skewed towards peptides with these sequences. Both situations result in libraries that are not random and can result in biased results when performing a screen. Several different phage vectors can be used when creating a phage library. These include baculovirus, T4 phage, Iv phage, bacterial flagella, pili, and cell-surface proteins. The filamentous bacteriophage M13 has been the mostly widely used and most successful vector for creating libraries. Our laboratory utilized such a vector, which will be the focus of this introduction. Depending on the strain being used, the M 13 particle is approxinlately one n1icron in length. Circular, single stranded (ss) DNA is sheathed in a tube con1posed of thousands of major coat proteins from the p VIII gene. Some libraries are engineered to express the insertion mutation on this protein. There are four minor coat proteins also expressed on the tips of the phage particle. The pIlI gene product is represented five tilnes on the coat surface and can tolerate large oligonucleotide inserts. Viral infection occurs when the pIlI coat protein attaches to the F-pilus of the K91kan E. coli host and transfers its ssDNA into the bacterium. The ssDNA is converted to double stranded (ds) DNA that provides the template for future progeny of ssDNA (Figure 1.5). Phage display has been used to find various pharmacologically relevant peptides and antibodies. These include peptides that aid in drug targeting (77), peptides that 29 Figure 1.5. Life cycle of filamentous phage. Beginning in the upper right comer. A. Phage particle displaying randon1 peptide sequence is used to screen (biopan) against a target protein. Isolated phage are then incubated with E. coli. B. Phage particle binds to the F pilus of the E coli and transfers its single-stranded DNA into the bacterium. This is classified as the infecting "+"-strand. C. This is then converted into the double-stranded replicative form (RF). Through a rolling circle mechanism the bacterium produces phage proteins from the RF "-"-strand. D. Phage particles are packaged at the cell surface into new phage particles and extruded through the cell membrane. The phage is nonlytic. B Binding to F+ Pilus ~ !Infection c BIOPANNING l.J.,lliJ~.ILt.I.l.ILlLILILILlLILI.C..... ~~ t! G ~ \9 ~'\"I'\1'\7'\1'\1'\l'\l"\1'\7'\'/'\1'\I'\l'\1\r\~ DS~~ ~ Peptide Display ...... - Insertion Points Outer Membrane Inner Membrane CYTOPLASM Gene Products _ pVI c::> pVII pili ~ pVII + piX Assembly pV A 30 D ~ pV pV/ssDNA Complex Late Early pV amount Low i pVamount high 31 inhibit matrix metalloproteinases (78) and the discovery of recombinant, yet fully humanized monoclonal antibodies with anti-tumor activity (79). Biopanning of cell surface makers of cultured tumor cells has yielded probes for these markers. Tumor­bearing nude mice have even been injected with phage libraries to discover peptides that are specific for neoplasia associated vascular tissue (77). Published dissociation constants of phage derived peptides for their receptors have ranged from 4 11M to 0.2 nM. Research Objectives The discovery of new and potentially useful tubulin active agents could provide a new tool to identify drugs useful in the treatll1ent of cancer. The objective of this dissertation project is to discover and characterize peptides that may be useful in drug discovery oftubulin-interactive drugs. This was accomplished by studying peptides found by two separate methods. A traditional cytotoxicity screen used in our laboratory found a peptide, vitilevuamide, fronl two ll1arine organisms. After isolating the cytotoxic element from the crude extract of tissue fronl the ascidians Poly~yncranton lithostrotum and Didemnum cucculiferum, vitilevuamide was subjected to a differential cytotoxicity screen and determined to be a tubulin active compound. Further testing confirmed those results. Chapter 2 discusses the unique properties of viti levu amide and our attempts to characterize vitilevuanlide and compare it to other tubulin active agents. The second nlethod for finding tubulin interacting peptides was by screening tubulin against a phage display library. Tubulin is a very "sticky" protein and has multiple binding sites. Many of these sites are well known and well characterized while others remain unknown. Chapter 3 describes the use of phage display for the isolation of several peptides that nlay 32 associate with tubulin and the potential implications of the peptides found. Some of these peptides were investigated in tubulin assays that help characterize potential tubulin active properties. The final chapter of this dissertation is devoted to a brief discussion and comparison ofvitilevuanlide (and other peptide agents) to those peptides studied from phage libraries. Binding sites for the phage-derived peptide are theorized due to the peptide's sequence similarity to known tubulin binding proteins. CHAPTER 2 CHARACTERIZATION OF VITILEVUAMIDE Introduction Many structurally diverse molecules interact with tubulin, the major conlponent of microtubules. Disruption of tubulin dynanlics is the mechanism of action of many cancer drugs used in the clinic today. Molecules that interfere with tubulin polynlerization can cause cells to arrest in metaphase and several agents with this l11echanism of action are useful as antineoplastic drugs. Terrestrial plants have provided all of the tubulin inhibitors used in the clinic today. They include the vinca alkaloids, the taxols, and colchicine (80). However, other organisms produce tubulin inhibitors; for instance, fungi have produced the tubulin active compounds phomopsin A (81), ustiloxins A-F (82), rhizoxin (83), and the ansamitocins (84). Marine organisnls also have provided many structurally diverse antimitotic compounds. Spongistatin 1, halichondrin B, halistatin 1 and 2, honl0halichondrin D, and helniasterlin were isolated from a variety of sponges (68, 85-91). Curacin A was isolated from a blue-green algae (92). Dolastatin 10 and the depsipeptide dolastatin 15 are peptides discovered from the sea mollusk Dolabella auricularia (92-96). Vitilevuamide is an especially cytotoxic compound that was found in Dr. Chris 34 Ireland's laboratory to possess an average LCsoof 100 nM in a panel of man lIn ali an cancer cell lines (97). It was isolated from two species of marine ascidians, Didemnum cuculiferum and Polysyncranton lithostrotum. Structurally, vitilevuamide is a bicyclic, 13-amino acid peptide, ofnl0lecular fOffilula C77H114N1402JS (MrH+ 1603.8117) (Figure 2.1). Anlino acids were identified as ala, ser, val, thr, ile, 2 phe, pro and modified anlino acids as 2 homoisoluecine, lanthionine, dehydroalanine, and N-methyl methoxinine, (and an additional succinate unit). Vitilevuanlide also contains an ester linkage that is necessary for activity. Due to the cytotoxic potency of vitilevuamide, it was evaluated in a 25-cellline panel. Analysis of the varying sensitivities in these cell lines suggested a weak similarity to several taxol analogs (data not shown) (97). Further testing revealed vitilevuanlide scored strongly positive in a cell-based screen for inhibitors oftubulin polymerization (data not shown) (98). We therefore hypothesized that vitilevuamide acts as an inhibitor of tubulin polymerization. Materials and Methods Chemicals and Reagents The isolation and purification of vitilevuamide from Didemnum cuculiferum and Polysyncranton lithostrotum have been described (97). [H3 ] Vinblastine was obtained frOln Anlershatn Corp., and [H3 ] colchicine from NEN. Non-radiolabeled vinblastine, vincristine, colchicine and GTP were obtained fron1 Sigma Chemical Co. (St. Louis, MO, USA). [H3 ] Dolastatin was a generous gift of Dr. R. D. Haugwitz (Drug Synthesis and Chemistry Branch, National Cancer Institute) and nonradiolabeled dolastatin was a Figure 2.1. Structure of viti levu amide. Amino acids were identified as ala, ser, val, thr, ile, 2 phe, pro and modified aInino acids as 2 homoisoluecine (hil), lanthionine (lan), dehydroalanine (dha), and N-methyl methoxinine (nnlm), (and an additional succinate unit). Vitilevuamide also contains an ester linkage that is necessary for activity. Molecular fonnula C77Hl14N14021S (MrH+ 1603.8117) 35 36 0- 0 Nmm NHY Val lie 0 Dha Hit 37 generous gift of Dr. D. 1. Newman (Natural Products Branch, National Cancer Institute). [8_H]3 GTP was obtained from Alnershanl Phanl1acia Biotech. 4,4' -dianilino-l, 1'­binaphthyl- 5, 5' -disulfonic acid (bis-ANS) and 2-1norpholinoethanesulfonic acid (MES) was obtained from Sigma. Electrophoretically homogeneous bovine brain tubulin containing Inicrotubule associated proteins (MTP) was isolated and purified with minor nl0difications according to nlethods described elsewhere (99,100). Purified tubulin was obtained by phosphoceIlulose chromatography (PC). SDS gel electrophoresis detelTIlined that this tubulin was free of MAP protein and no contaminating proteins were detectable even when the gel was grossly overloaded (Figure 2.2). All drugs were dissolved in DMSO prior to use. Cell Cytotoxicity Cytotoxicity was established in a MTT (3-( 4,5-dinlethylthiazol-2-yl)-2,5- diphenyltetrazolium bronlide) assay as described by Mosmaml (l 0 1) and modified by others (102,103). The MTT assay assesses cellular viability based on the ability of the l11itochondrial succinate dehydrogenase to reduce the MTT. Vitilevuamide was dissolved in 1000/0 DMSO at an initial concentration of 10 mghnL and serially diluted. The final concentration of DMSO in the ceIl culture wells was 1 % or less. Each human cel1line was seeded (20,000 cells/well) in 200 JlL of growth mediunl in Coming 96-well 111icrotiter plates. Cells were routinely cultured in nlinilnal essential nledium (aMEM) (Sigma) supplemented with 10 % fetal bovine serum (Atlanta Biologicals, Atlanta, GA), 10,000 U/L penicillin, and 10,000 U/L streptomycin (Sigma). Four hours after seeding, refed with fresh nledium. Each well was then treated with 11 JlL of MTT solution (5 38 Figure 2.2 SDS-polyacrylamide gel electrophoresis of phosphocellulose-purified tubulin. Gels (100/0) were loaded with 10 J,lg, 25 J,lg and 50 J,lg (left to right respectively) of tubulin purified on a phosphocellulose column to remove high molecular weight associated proteins. Gels were stained with Coomassie blue and photographed. Molecular weights are indicated on the left and the major band is located between the 66 kDa and 45 kDa bands, indicating the presence of the 50 kDa a and P subunits of the tubulin heterodimer. Lower lTIolecular weight bands present are degradation products of the tubulin band. No high molecular weight bands were seen, indicating that the tubulin is nearly free of any associated proteins. 39 45 hDa 29 kOa 24kDa - 40 Ing/mL in phosphate buffered saline (PBS), pH 7.4) and incubated for 4 hours. Viable cells reduce the MTT to a purple forn1azan product that was solubilized by the addition of 1 00 ~L of DMSO to aspirated culture wells. The absorbance at 540 nm was measured for each well using a BIO RAD MP450 plate reader. Average absorbance for each set of drug-treated wells was compared to the average absorbance of the control wells to detelmine the fractional survival at any particular drug dose. LCso was defined as the drug concentration that yielded a fractional survival of 0.5. Cell Cycle Analysis AA8 Chinese han1ster ovary (ATCC) cells (2-3 n1illion) were seeded in 25 Inl flasks and grown 4 hours in aMEM medium. Cells were then grown for 16 h in an ICso concentration of vitilevuamide (concentration required to inhibit growth of the cells 80%) to allow all cells time to traverse at least one cell cycle. Medium was removed and the cells were trypsinized. Cells were then pel1eted and resuspended in PBS and fixed by adding 2 mL cold lnethanol drop-wise while mixing. Fixed cells were pelleted at 400 x g for 5 min and washed once with PBS. Cell suspensions were centrifuged as above and then resuspended in 0.5 ITIL of 100 J.!g/mL propidium iodide solution in PBS. One half mL of PBS containing 200 UhnL of RNase A was then added. Cells were incubated at roon1 temperature for 30 min in the dark. DNA content was determined by flow cytometry (Univ. Utah, F.A.C.S. core facility). The results were analyzed using the Modfit cell cycle analysis program (Varity Software, Topshan, NE). 41 In Vitro Inhibition ofTubulin Polymerization Inhibition of tubulin polytuerization was determined using a light-scattering assay (25, 70) modified to use a 96-well microtiter plate. Tubulin (20 J..lM) isolated as described above (99, 100) was incubated with polymerization buffer (100 tuM Pipes, pH 6.9; 1 n1M MgCh, 1 mM EGTA) and inhibitory drug. Experiments using PC purified tubulin (20 J..lM) were incubated in a gluta111ate buffer (l M glutmnate, 5 111M MgCh). Unless otherwise indicated, all drugs were preincubated for 15 min on ice with the tubulin. Incubations were transferred to wells containing 1 mM GTP (final concentration) to begin polymerization. An initial reading was taken to establish a base line and then the plate was warmed to 37° C and read every 20 seconds for 30-45 min. Measurements were done on a SpectraMax 190 96-well plate reader at 350 nm. An increase in absorbance indicated an increase in tubulin polymerization. Determination of Critical Concentration for Tubulin Polymerization The in vitro polymerization assay described previously was used to determine the effect of viti levu amide on the critical concentration oftubulin (i.e., the minimum concentration oftubulin required for polymerization). A microtubule protein (MTP) concentration range of 5 J..lM to 80 J..lM was used. All concentrations were preincubated on ice with either DMSO or drug for 15 tnin prior to reading. An incubations were done in PEM 100 (l00 111M Pipes, pH 6.9, I1nM MgCh, 1 mM EGT A). Polymerization was initiated by adding the incubation nlixture to a well containing GTP (l mM final concentration) in a 96-well plate. The plate was warmed to 37° C to promote polymerization. Concentrations of tubulin where no polymerization could be detected were not included in determining the critical concentrations. O.D. readings at 350 nm were recorded and plotted on a graph oftubulin concentration vs. O.D. Regression analysis was performed and critical concentration determined by calculating the x­intercept of the best-fit line. Induction of Tubulin Aggregation 42 Promotion oftubulin aggregation was followed turbidmetrically at 350 nn1 on a 96-well plate reader. Tubulin (20 JlM) containing MAPs in PEM 100 (100 mM Pipes, pH 6.9; 1 mM MgCb, 1 InM EGT A) were added to wens containing either 20 JlM drug or DMSO alone. In some cases, an additional drug was added for inhibition studies. No GTP was added to any of the reactions to prevent tubulin fron1 forming microtubules. The plate was kept at room temperature (22°C) and read every 20 seconds for 30-60 min to determine the extent of aggregation. Measurelnents were done on a SpectraMax 190 96-well plate reader at 350 nm. An increase in absorbance indicated an increase in amount of aggregation. Protection of Tubulin Denaturation Fluorescent measurelnents were taken on ISS PC-l spectrofluorometer. In order to quantify the rate of tubulin denaturation (73), phosphocellulose (PC)-purified tubulin (20 JlM) was incubated at 37°C in PEM 100 (l00 mM Pipes, pH 6.9; 1 mM MgCb, 1 mM EGTA) in the absence or presence oftubulin active compound. At various time points, 100 JlL oftubulin incubation mixture was diluted to 2 ~lM and nlixed with 5 JlM 43 bis-ANS. Fluorescence was determined using an excitation wavelength of 385 nm and emission wavelength of 490 nm (74). Sanlples without bis-ANS or without tubulin were used as negative controls. Tubulin Fluorescence Quenching Fluorescent nleasuren1ents were taken on ISS PC-l spectrofluoron1eter. PC tubulin (5 ~M) was incubated for 30 min in PEM 100 (100 n1M Pipes, pH 6.9, 1 mM MgCh, 1 luM EGT A), in the presence or absence of drug. After 30 nlin, the incubation was transferred to a quartz cuvette and read. The excitation wavelength was 284 nn1 and the emission spectra were recorded between 300 nn1 to 420 nm. Cuvettes containing only buffer and buffer plus drug were utilized as blanks. Emission curves of drug-bound tubulin were compared to those of control drug free tubulin. Drug Binding Displacen1ent Assay Binding of [H3 ] dolastatin and [H3 ] vinblastine to tubulin was measured using centrifugal gel filtration chromatography as described elsewhere (104, 105). All reactions contained 0.1 M MES, pH 6.9, 0.5 mM MgCh, tubulin, radiolabeled ligand, vitilevuamide, and DMSO (the drug solvent). Tubulin (1 mghuL) was added last in every experiment and allowed to incubate for 45 min. Reaction n1ixtures were then added to 1-1UL microspin columns ofBio-Gel P-30 (Bio-Rad) and centrifuged in a desktop centrifuge. The filtrate was collected and radioactivity detem1ined by liquid scintillation counting, allowing calculation of a mole ratio of drug to tubulin for each experinlent. Controls determined that no radioactivity passed through the column in the 44 absence of tubulin. Assay for GTP Binding to Tubulin Each 0.5 l11L reaction l11ixture contained 0.5 Ing/mL (5 /-lM) tubulin, 50 /-lM [S}H] GTP, 0.1 M MES, pH 6.9, 0.5 n1M MgCh, 20/0 (v/v) DMSO, and different concentrations of inhibitor. Incubation was for 10 min at 00 C. Triplicate 0.15 mL aliquots of each reaction mixture were processed by centrifugal gel filtration on syringe-columns of Sephadex G-50 (superfine) at 40 C, as described previously (67,105). Protein and radioactivity in the filtrates were quantified. Values were then nom1alized to the control reaction mixtures. Stabilization of Colchicine Binding Stabilization of colchicine binding activity to tubulin was determined using the ll1ethod described by Luduefia et al. (l06). Each O.l-ml reaction mixture contained 0.4 mglml tubulin, 0.1 M MES (pH 6.4), 0.1 M EDT A, 1 mM GTP, 0.5 lYIM MgCh, 1 mM 2- l11ercaptoethanol, 1mM EGT A, 50/0 DMSO, 60 JlM eH] colchicine, and the indicated drug at 50 ~lM. If indicated, the reaction nlixtures were preincubated with test drug for 3 h at 37 DC prior to the addition of the eH] colchicine. After the addition of eH] colchicine the samples were incubation for 2 h at 37 DC. Reaction mixtures were then filtered through a stack of three DEAE-cellulose filters to determine the amount of eH] co 1chicine bound to the tubulin. 45 In Vivo Activity Against P388 Lynlphocytic Leukemia CDFI mice were injected intraperitonially (IP) on day ° with 1 x 106 P388 lymphocytic leukenlia cells. Five animals were then rando111ly selected and placed into either of two groups, placebo or drug treated. On days 1, 5, and 90fposttunl0r implantation each mouse was treated with either vehicle (100/0 methylcellulose in Eagle's balanced pH solution) or vitilevuamide by IP injection. Animals were checked twice daily and the day of death for every mouse posttunlor implantation was recorded. A positive drug response is defined as a greater than 25% increase in the mean life span (% ILS) relative to placebo control. Curve Fitting Experinlents (in vitro polytnerization and aggregation) that nleasured the time course of either microtubule fonnation or tubulin aggregate formation were fit with a non-linear regression nlodel to determine the observed rate constant (kob) of the nlicrotubule formation or tubulin aggregate formation. This curve fit was performed to detennine whether introducing a drug (vitilevuamide) to the system caused a reduction or enhancement of the rate of polymerization or aggregation. A nl0dified fonn of the one phase exponential association curve fit was used. y= IF (X <XO, Plateau, plateau + (Top - Plateau)*(l - exp (-kob *(X-XO»» Plateau until X=XO, then exponential decay to zero 46 This equation was chosen because it best fits the functional situation seen when measuring tubulin polYl11erization. Four variables are fit with the equation. The first variable is XO and represents the X value (time in seconds), at which polymerization or aggregation begins and can be Ineasured. This variable (XO) of the equation accounts for the lag observed in the initiation of tubulin polytllerization or aggregation seen on the plots. The second variable is Plateau and represents the average of the Y (OD at 350 nm) values for all points until XO is reached. The Plateau variable accounts for the values (OD at 350 nm) seen in the lag portion of the graph and factors this value in when calculating the total increase in Y (OD at 350 nm), which is representative of the amount of polytnerization. The third variable is TOP and represents the average for the nlaximunl Y (OD at 350 nm) observed. The TOP variable represents the highest extent of polytnerization that has occurred and is the best indicator on ho\v well a drug inhibits tubulin polymerization or induces aggregation. The final variable kob in the equation is the observed rate constant, expressed in units of inverse time (S-I). It is a measure of how quickly the incubation reaches equilibriunl (in this case maxinlUln absorbance). It is found by fitting the equation to the data and the value can be considered an apparent rate of polytnerization or aggregation for purposes of the analysis. The larger the kob is for a curve, the faster the tubulin polymerized or aggregated and reached an equilibrium point. It should be noted that this is not the same as the kon rate. The curves were calculated by exporting the data measured (SoftMax Pro) on the plate reader to an Excel spreadsheet. Blanks were subtracted from the data for each well and then nomlalized to the initial (0 tinle point) time point for each \vell. These values were then exported to the GraphPad Prizm program and plotted as OD versus tinle. This 47 new plot is identical to the plot obtained from the SoftMax Pro program. Each plot was then fitted with the equation above and the four values described were calculated. ICso values were calculated using the TOP value because this variable is most relevant to the biological effect (inhibition of extent of polymerization) of these agents. The kob value is also c0l11pared in the Results section below because rate of polytnerization also has a biological consequence in the cell. Results The activity-based purification of viti levu amide showed that it was a highly cytotoxic cOlnpound in preliminary results (97). Further analysis in several human tumor cell lines revealed cytotoxic activity in the nanonlolar range. LCso values (graphs not shown) were obtained for A2780 human ovarian carcinoma cell lines (3 nM), RCT 116 hUlnan colon tumor (6 nM), A5249 lung cancer (124 nM), SK MEL-5 melanoma tumor (311 nM) and A498 kidney cancer (311 nM) cell lines. An LCso value of 3.1 JlM was obtained for Chinese halTIster ovary cells treated with vitilevuamide for only 16 h instead of 72 h in order to calculate the proper concentrations to be used in the cell cycle analysis. The nanolTIolar toxicity of viti levu amide in these cells suggested utility as an anticancer agent. Analysis in a proprietary cell panel (Wyeth Ayerst Research) suggested a nlechanism of cytotoxicity similar to taxol analogs, anticancer agents that stabilize and proll10te tubulin polymerization. 48 Effect of Vitilevuamide on Cell Cycle Progression AA8 Chinese hamster ovary cells (2-3 nlillion) were seeded in 25 mL flasks and treated with 15 J.lM vitilevuanlide for 16 h. Fluorescence-activated cell sorting (F ACS) analysis (Figure 2.3B) showed that vitilevuamide produced an accumulation of cells in the G2/M phase of the cell cycle. In the presence of 15 J.lM vitilevuaInide, 51.60/0 of the cells accumulated in G2IM, compared with 17.0%) in controls. Correspondingly, the fraction of cells in Go/G\ was reduced from 40.80/0 to 7.90/0 in treated cells. Cells treated with vitilevuamide also demonstrated an increase in polyploidy. The drug-treated population contained 190/0 diploid cells, 78% tetraploid cells and 30/0 octaploid cells. These results demonstrate that, in cultures of cycling Chinese hamster ovary cells, vitilevumnide arrests cell cycle progression in G2/M. In Vitro Inhibition of Tubulin Polymerization The ability ofvitilevuaInide to inhibit the polymerization oftubulin was confirmed in vitro in a purified protein system. Vitilevuamide was examined for its ability to inhibit tubulin both in the presence and the absence of microtubule associated proteins (MAPs). MAP induced polymerization was checked first and vitilevuanlide was found to be an effective inhibitor, with an in vitro ICso of 2 J.lM (data not shown). Inhibition of polymerization of MAP tubulin does not formally exclude the possibility that drug inhibition of MAPs is the mechanism of action. In order to confiml that tubulin (and not MAPs) was the target of viti levuamide, it was tested in an in vitro tubulin polynlerization assay using PC tubulin. Figure 2.4A shows the concentration-dependent inhibition of PC tubulin by vitilevuaInide. It was determined that the ICso concentration Figure 2.3. Effect ofvitilevuanlide on cell cycle progression. Chinese hamster ovary cells were plated in growth rnediu111 in 25 mL flasks at a density of 2-3 nlilliol1 49 cells/flask. The cultures were then treated with vehicle (A) or 15 ~lM vitilevuamide (B) for 16 h. Cells were then processed for propidiunl iodide staining and F ACS analysis. Cell distribution in Go/GJ , S, and Gz/M, was determined using the Varity Modfit progral11. The areas under the left solid peak, the middle stripped peak, and the right solid peak represent fraction of cells in Go/G I, S, and Gz/M, respectively. Cells were treated in triplicate and pooled prior to F ACS analysis. 50 A. GO/Gl G21l\t1 so 100 150 20f.J ::,)0 ChalUlel'ii B. G2~1 SO 100 150 .:l00 250 ChalUlels 51 for the inhibition of the extent ofpolyrnerization was approxilnately 6.7 ~lM (Figure 2.5). Figure 2.4B is the curve fit analysis of the plot in Figure 2.4A that was used to determine the rates for tubulin polymerization. Table 2.1 summarizes both the effects of vitilevuanlide on both the extent and the rate oftubulin polymerization and compares each concentration of vitilevuamide to untreated control with at-test. Vitilevuamide significantly affected the extent ofpolynlerization of PC tubulin at all concentrations above 2.67 ~M (P < 0.05). Vitilevuamide also significantly affected the rates of PC tubulin polynlerization at all concentrations above 1.33 ~lM (P < 0.05). Effect ofVitilevuanlide on the Critical Concentration ofTubulin Polymerization In vitro tubulin polymerization that is monitored by the light scattering method described above requires a Ininimum concentration of protein to produce an observable result. This is conlnl0nly referred as the critical concentration of tubulin polymerization. When tubulin is polymerized without any inhibitors the critical concentration oftubulin nonnally ranges from 2-7 ~M. Most compounds that inhibit tubulin polytnerization will cause an increase in the critical concentration oftubulin. Vitilcvuamide was found to increase the critical concentration of tubulin polymerization (Figure 2.6). The extent of tubulin polynlerization (increase in absorbance) was used to calculate the points used on the graph. When only DMSO is added, the critical concentration for tubulin polytllerization is approximately 2 ~lM. The addition of2 ~M vitilevuamide increases the critical concentration oftubulin polymerization to 12 ~M. Addition of 5 ~M vitilevuamide further increased the critical concentration of tubulin polymerization to 19 52 Figure 2.4. Dose-dependent in vitro inhibition of glutanlate-induced phosphocellulose purified tubulin polynlerization. A. 45 nlin time course (X-axis) oftubulin polynlerization nleasured by turbidity at 350 nm (Y-axis). Concentration of vitilevuamide for each time course as follows: 0 0 ~M (control), .... 667 ~M, • 1.33 ~LM, • 2.67 ~M, 0 4 j1M, .6. 6.67 ~M, and D 8 ~M. All reactions contained 20 ~M tubulin, 1 M glutanlate, 5mM MgCh, and 4(Yo DMSO. All reactions were preincubated with drug for 15 min on ice prior to the addition of GTP. Tubulin without GTP was used as a blank and subtracted from experilnental values. B. Curve fit of data in A used to calculate rates. Concentration ofvitilevuanlide for each time course as follows: 0 0 ~M (control), .... 667 ~M, • 1.33 ~M, • 2.67 ~M, 0 4 ~M, .6. 6.67 ~M, and D 8 ~M. These results are froln one of two identical experinlents. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. A. Co 0 B. E c: 0 LC') M ci 0 53 0.25r-------------------------------------------------------------------~ 0.2 cP:& 0.15 oo •• 0 0 .~ 0 ••• 0.1 0 •••• o • ~ 0.05 0 0 0.25 0.20 0.15 0.10 0.05 0.00 o 500 1000 1500 Time (secs) ~ ....... I 500 1000 1500 Time (s) 2000 444444 I 2000 4 2500 I. 44 4 2500 54 Figure 2.5. ICso calculation of the inhibition of PC tubulin polyn1erization by vitilevuan1ide. The total extent of polymerization (total increase in absorbance) was used to calculate the ICso. The ICso value was calculated by plotting the concentration of vitilevuaInide (X-axis) versus percent of the extent of control polymerization (Y-axis) (vitiJevuaInide alone). Each point represents the average of at least three individual experiments. ICso for inhibition of the extent of polymerization was calculated to be 6.68 ~lM. Curves were fit using a sigmoidal dose-response curve from GraphPad Prism version 3.02, GraphPad Software, San Diego CA USA. 55 c: 1.00 0 +i CO "QNt:) 0.75 E ~ 0 a.. 0.50 e.., c: 0 U 0.25 ~ 0 0.00 +-------,-------.----,--------.-------.--------, 0.0 2.5 5.0 7.5 10.0 12.5 15.0 Vitilevuamide (IJM) I 56 Table 2.1 Best-fit Values for the Extent of Polymerization and Rate ofPolytllerization of Phosphocel1ulose Purified Tubulin Treated with Vitilevuamide Concentration Extent Different fron1 Rate of Di fferent from Vitilevuanlide Polymerization Control Polymerization Control (~M) (O.D.) (Mean:±: (t-test) (s-l) (Mean (t-test) SD) SD) I o (Control) 0.2310 ± 0.006998 ± 0.004243 0.0006725 0.333 0.2160 ± No 0.006604 ± No 0.02687 P= 0.5171 0.0004130 0.5538 0.667 0.2165 ± No 0.004581 No 0.03889 P= 0.6525 0.001338 0.1500 1.33 0.2235 ± No 0.004344 ± No 0.01626 0.5925 0.001987 P= 0.2155 2.67 0.2190 ± No 0.002491 ± Yes 0.02404 P= 0.5589 0.0003041 0.0131 4.00 0.1615± Yes 0.0006299 ± Yes 0.006364 p= 0.0060 ! 0.0001524 p= 0.0058 6.67 0.1250 ± Yes I 0.0004363 ± Yes 0.02687 0.0314 0.0006056 0.0094 8.00 0.0380 ± Yes 6.4850e-006 ± Yes 0.01980 P= 0.0055 2.5580e-006 0.0046 13.33 0.0390 Yes 0.001095 ± Yes 0.01838 0.0048 0.0009009 P= 0.0176 ! . I I I I 57 Figure 2.6. Effect of vitilevuanlide on the critical concentration of tubulin. Tubulin (5-80 ).!M) was preincubated with vitilevuamide for 15 min prior to the addition of GTP to start polymerization. Control polymerization contained only DMSO. Polymerization was nlonitored at 37°C for 30 nlin by UV spectrometry at 350 nm. Results were plotted as tubulin concentration (X-axis) versus O.D. at 350 nm (Y-axis). Reactions that produced no detectable polymerization were not used in regression analysis. The point where the regression line crosses the x-axis was determined to be the critical concentration. Critical concentrations (cc) were determined for. Control (cc=2 JlM), • 2 ).!M Vitilevuamide (cc=12 JlM), ... 5 JlM Vitilevuanlide (cc=19 JlM), and. 10 JlM Vitilevuanlide (cc=34 JlM). co N o ("') N o WU OS£ le ao co o o ("') o o o CJ) N o o I c: :::s .c :::s I- ::E :t 58 59 J.lM and the addition of 10 J.lM ofvitilevuamide increased the critical concentration of tubulin polynlerization to 34 J.lM. Comparison of the best-fit lines used to calculate the critical concentrations found that the slopes of each line were not significantly different (P= 0.06319) but that the x-intercepts (value for critical concentration) were significantly different (P < 0.0001). Effects on Tubulin Aggregation Drug-induced tubulin aggregation results fronl structurally induced changes in the tubulin heterodimer that is GTP and glutamate independent and results in the formation of either ring or spiral structures conlposed of dnlg bound tubulin dimers. Many tubulin inhibitors can cause tubulin aggregation when incubated with tubulin at stoichiometric or superstoichionletric concentrations, but this aggregation varies in its degree and structure (71,104). The ability ofvitilevualnide to induce tubulin aggregation was examined by measuring spectrophotometrically (350 nm) the increase in turbidity oftubulin incubated with stoichionletric or superstoichiometric amounts of vitilevualnide at room temperature. For comparison purposes, the extent of aggregation (indicated by the total increase in absorbance) was used to determine how well a conlpound was able to induce tubulin aggregation. Vitilevuanlide's ability to induce aggregation was cOlllpared to two other cOlnpounds, dolastatin 10 and vinblastine. Both dolastatin 10 and vinblastine have been shown to induce tubulin aggregation (104). Comparison of vitilevuanlide to dolastatin 10 and vinblastine was done to evaluate if vitilevuamide is more effective at inducing tubulin aggregation than these other two agents. Figure 2.7 A shows the tin1e­dependent increase in turbidity of tubulin (20 ~M) in the presence of 20 ~LM and 40 11M 60 vitilevuan1ide, 20 JlM vinblastine, and 20 JlM dolastatin 10 without GTP and at roon1 temperature. The increase in the extent of turbidity indicates that vitilevuamide induces tubulin aggregation. At 20 JlM, vitilevuan1ide (0.05367 ± 0.0029) was a better inducer of the extent (total increase in absorbance) of aggregation than both 20 JlM vinblastine (0.0343 ± 0.0029) (P 0.0001) and 20 JlM dolastatin 10 (0.021 ± 0.0014) (P < 0.0001). EXaIllining the rate of aggregation (Figure 2.7B) found that 20 JlM vitilevuamide (0.001472 ± 0.0004692) pron10ted aggregation at a slower rate than 20 JlM vinblastine (0.002298 S-l 0.0002773) (P= 0.0138). In contrast, 20 ~M vitilevuaIl1ide pron10ted aggregation at a faster rate than dolastatin 10 (0.0005221 S-I ± 8.591 Oe-005) 0.0353). The type of aggregate (spiral or ring) fortned by vitilevuamide has not been detennined. Electron n1icroscopy studies will be required to determine what type of aggregate is fonned by vitilevuan1ide. Inhibition of Drug Induced Tubulin Aggregation Tubulin aggregate fortnation results from structural changes in the tubulin heterodilner that are induced by high concentrations of a single tubulin drug. Tubulin aggregation induced by stoichiometric or superstoichiometric anlounts of a tubulin inhibitor can be decreased by coincubation of a second tubulin inhibitor at lower concentrations (71). Not all tubulin drugs will inhibit the aggregation of tubulin induce by another drug. These experiments were done to detem1ine the ability of vinblastine and dolastatin 10 to inhibit tubulin aggregation induced by vitilevuan1ide. The extent of vitilevuan1ide-induced tubulin aggregation (maxilnum increase in absorbance) was inhibited by dolastatin 10 (Figure 2.8A). The rates for vitilevuamide-induced tubulin 61 Figure 2.7. Induction oftubulin aggregation by vitilevuamide, dolastatin 10, and vinblastine. A. 20).lM tubulin was incubated either alone 0 or with. 40 ).lM vitilevuamide, 0 20 ).lM vitilevuamide, .6. 20 ).lM vinblastine, and 0 20 ).lM dolastatin 10 at room tenlperature (~22°C) in the absence ofGTP. Aggregation was measured as a change in absorbance at 350 nm (Y-axis) over time (X-axis). At 20 ).lM, vitilevuamide (0.05367 0.0029) was a better inducer of extent (total increase in absorbance) of aggregation than both vinblastine (0.0343 ± 0.0029) (P < 0.0001) and dolastatin 10 (0.021 ± 0.0014) (P < 0.0001). All reactions were done in PEM 100 (100 mM Pipes, pH 6.9; 1 InM MgCb, 1 mM EGT A). B. Curve fit for plot in A used to detemline aggregation rates. 20 ).lM tubulin was incubated either alone 0 or with. 40 ).lM vitilevuan1ide, 0 20 ).lM vitilevuamide, .6. 20 ).lM vinblastine, and 0 20 ).lM dolastatin 10 at roon1 temperature (~22°C) in the absence ofGTP. ). Examining the rate of aggregation found that 20 ).lM vitilevualnide (0.001472 S-I ± 0.0004692) promoted aggregation at a slower rate than 20 ).lM vinblastine (0.002298 S-I ± 0.0002773) (P= 0.0138). In contrast, 20 ).lM vitilevuanlide promoted aggregation at a faster rate than 20 ~tM dolastatin 10 (0.0005221 S-I ± 8.5910e-005) 0.0353). This graph represents one of two identical experilnents done in duplicate. A. o o B. E c: o It) M ci o 0.01 o.oo_~w::I"-' 500 1500 Time (s) 62 -.-.. - .-. 63 aggregation were also calculated using the curve fits for dolastatin 10 (Figure 2.8B). The extent of vitilevuanlide-induced tubulin aggregation (maximum increase in absorbance) was inhibited by vinblastine (Figure 2.9A). The rates for vitilevuamide-induced tubulin aggregation were also calculated using the curve fits for by vinblastine (Figure 2.9B). values for the extent of inhibition (maxinlum increase in absorbance) was subnlicromolar for dolastatin 10 (Figure 2.1 OA) and 2.28 ~M for vinblastine (Figure 2.1 OB). The ICso for dolastatin lOis an estinlate because all of the concentrations of dolastatin 10 tested resulted in a greater than 50% reduction in vitilevuamide-induced tubulin aggregation. The ICso for dolastatin lOis below 1.25 ~lM however. Both dnlgs resulted in an inconlplete reduction of the extent of vitilevuanlide-induced tubulin aggregation. At 10 ~lM, dolastatin was able to reduce vitilevuamide-induced tubulin aggregation by 70 %. At 1 0 ~lM, vinblastine was able to reduce vitilevuanlide-induced tubulin aggregation by 60 %. This residual aggregation is due to the inability of dolastatin 10 and vinblastine to conlpletely displace vitilevuamide from the tubulin heterodimer and prevent any aggregation froln occurring. Higher concentrations of dolastatin 10 and vinblastine may reduce vitilevuamide-induced tubulin aggregation even further, although at higher concentrations 1110re aggregation nlay be seen. An increase in aggregation may result because both dolastatin 10 and vinblastine cause aggregation at higher concentrations. Table 2.2 sU111marizes the inhibition ofvitilevuanlide-induced tubulin aggregation by vinblastine and dolastatin 10. All concentrations of dolastatin lOused significantly reduced (P< 0.005) the extent of vitilevuarnide-induced tubulin aggregation. Exanlining the rate of viti levu amide-induced tubulin aggregation revealed dolastatin 10 reduced the rate significantly (P< 0.005) at all concentration above 1.25 ~M. All concentrations of 64 Figure 2.8. Inhibition ofvitilevuatnide-induced tubulin aggregation by low doses of dolastatin 10. A. Twenty ~M tubulin were incubated either alone 0 or with 20 ~lM vitilevuaJnide ° at room temperature in the absence of GTP and turbidity measured on a 96-well plate reader at 350 nm (Y-axis) for 45 nlin (X-axis). Tinle course of aggregation with 20 ~M vitilevuamide 0, plus 2.5 ~M dolastatin 10 ., 5~lM dolastatinl0 0, and 10 ~M dolastatin 10 L\. This graph represents one of three identical experiments. B. Curve fit for plot in A used to determine aggregation rates. Twenty ~lM tubulin were incubated either alone 0 or with 20 ~M vitilevuamide ° at room temperature in the absence of GTP and turbidity lneasured on a 96-well plate reader at 350 nm (Y-axis) for 45 min (X-axis). Time course of aggregation with 20 ~M vitilevuamide 0, plus 2.5 ~lM dolastatin 10 ., 5~M dolastatinl0 0, and I 0 ~M dolastatin 10 L\. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. A. 0.05 0.04 0.03 a0 0.02 0.01 0 -0.01 0 B. 0.050 0.045 0.040 0.035 Er:: 0.030 0 L() 0.025 M ci 0.020 0 0.015 0.010 0.005 -0.000 500 0 500 1000 1500 Time (sees) 1000 1500 Time (5) 65 2000 2500 .... - .... - ••• 2000 2500 66 Figure 2.9. Inhibition of viti levu amide-induced tubulin aggregation by low doses of vinblastine. A. Twenty JlM tubulin were incubated either alone or with 20 JlM vitilevuamide ° at room temperature in the absence of GTP and turbidity measured on a 96-well plate reader at 350 nn1 (Y-axis) for 45 min (X-axis). Time course of aggregation with 20 JlM vitilevuamide 0, plus 2.5 JlM vinblastine ., 5 JlM vinblastine 6, and 10 JlM vinblastine O. This graph represents one of three identical experiments. B. Curve fit for plot in A used to determine aggregation rates. Twenty JlM tubulin were incubated either alone 0 or with 20 JlM vitilevuamide ° at room ten1perature in the absence of GTP and turbidity n1easured on a 96-well plate reader at 350 nm (Y-axis) for 45 min (X-axis). Tin1e course of aggregation with 20 JlM vitilevuan1ide 0, plus 2.5 JlM vinblastine ., 5 ~LM vinblastine 6, and 10 JlM vinblastine O. This graph represents one of three identical experilnents. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. A. o o B. E I:: o It) M c:i o 0.000 -0.005 o 500 1000 1500 Time (s) 2000 67 <> 2500 68 Figure 2.10. ICso calculation of dolastatin 10 and vinblastine inhibition of viti levu amide induced tubulin aggregation. A. ICso calculation of dolastatin 10 inhibition of vitilevuan1ide induced tubulin aggregation. The total extent of aggregation (total increase in absorbance) was used to calculate the ICso. The ICso value was calculated by plotting the concentration of dolastatin 10 versus percent of the extent of control aggregation (vitilevuamide alone). Each point represents the average of at least three individual experiments. B. ICso calculation of vinblastine inhibition of viti levu amide induced tubulin aggregation. The total extent of aggregation (total increase in absorbance) was used to calculate the ICso. The ICso value was calculated by plotting the concentration of vinblastine (X-axis) versus percent of the extent of control aggregation (Y-axis) (vitilevuamide alone). Each point represents the average of at least three individual experiments. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. A. s::: ...2..., 100 ctJ C) Q.) '­C) C) ex: o 50 ..'..­., s::: o () ~ o~~~~~~~~~~~~~~~~ 0.0 2.5 5.0 7.5 10.0 Dolastatin 10 (IJM) B. s::: .2 100 ~ fa C) .Q..) C) C) « .0.. 50 ~ s::: 0 0 ~ 0 o~~~~~~~~~~~~~~~~ 0.0 2.5 5.0 7.5 10.0 Vinblastine (IJM) 69 I 70 Table 2.2 Reduction in the Extent and Rate ofVitilevuatTIide-Induced Tubulin Aggregation by Dolastatin 10 and Vinblastine Concentration Extent Di fferent frOlTI Rate of Different from Dolastatin 10 Aggregation Control Aggregation Control (~lM) (O.D.) (Mean (t-test) (S-I) (Mean (t-test) SD) SD) o (Control) 0.06288 ± 0.001562 ± 0.01127 0.0001461 1.25 0.0350 ± Yes 0.001357 ± No 0.007071 P= 0.0116 0.0001273 0.1092 2.5 0.02733 ± Yes 0.001011 Yes 0.007506 p= 0.0008 0.0003092 p= 0.0024 5.0 0.02067 ± Yes 0.001090 ± Yes 0.004082 P<O.OOOI 0.0001018 P<O.OOOl 10.0 0.01233 ± Yes 0.001109 ± Yes 0.005750 P<O.OOOI 0.0002005 P= 0.0004 Concentration Extent Di fferent fronl Rate of Different from Vinblastine Aggregation Control Aggregation Control (~tM) (O.D.) (Mean ± (t-test) (S-I) (Mean ± (t-test) SD) SD) o (Control) 0.06288 ± I 0.001562 ± 0.01127 I 0.0001461 1.25 0.0486 ± Yes 0.0007163 ± Yes 0.002510 0.0190 0.0002572 P<O.OOOI 2.5 0.03983 ± Yes 0.0005076 ± Yes 0.005707 p= 0.0007 6.3690e-005 P<O.OOOI 5.0 0.0290 ± Yes 0.0002512 ± Yes 0.003674 P<0.0001 0.0001560 P<O.OOOl 10.0 0.02333 ± Yes 0.0001915 ± Yes 0.003933 P<0.0001 0.0001263 • P<O.OOOI i 71 vinblastine used significantly reduced (P< 0.005) the extent of vitilevuamide-induced tubulin aggregation. Examining the rate of vitilevuamide-induced tubulin aggregation revealed vinblastine significantly (P< 0.005) reduced the rate at all concentrations tested. Dolastatin 10 aggregation has been well characterized (71, 104) and results in the fom1ation ofwell-defined ring structures. Many tubulin active compounds either completely inhibit or partially inhibit this aggregation when coincubated at low concentrations (68, 71). Vitilevuamide was examined for its ability to inhibit dolastatin 10 induced tubulin aggregation. A time course of dolastatin 10 induced aggregation \vas nl0nitored by turbidity and Figure 2.11 A represents a study of aggregation of 20 JlM tubulin at room temperature (22°C). When no drug was added there was not an increase in tubulin turbidity. Addition of 20 JlM dolastatin 10 resulted in an increase in turbidity. When lower concentrations of viti levuanlide were added, do lastatin 10 induced tubuli n aggregation was inhibited (Figure 2.11 A). Table 2.3 summarizes the inhibition of dolastatin-induced tubulin aggregation by vitilevuamide. All concentrations of vitilevuaInide above 0.5 JlM significantly reduced (P< 0.005) the extent of vitilevuamide­induced tubulin aggregation. When examining the rate of dolastatin 1 O-induced tubulin aggregation (Figure 2.11 B), vitilevualnide reduced the rate significantly (P< 0.005) at all concentrations above 0.5 JlM. A rate was not calculated at 5.0 JlM vitilevuamide because aggregation was virtually undetectable and the equation for determining the rates could not be fit. Dolastatin 10 induced tubulin aggregation was completely inhibited by 5 JlM vitilevuanlide and the IC50 for this inhibition was 745 nM (Figure 2.12). 72 Figure 2.11. Inhibition of dolastatin 10 induced tubulin aggregation by low doses of vitilevuaI11ide. A. Twenty J-LM tubulin were incubated either alone or with 20 ~lM dolastatin 10 ° at rOOln temperature in the absence of GTP and turbidity measured on a 96-well plate reader at 350nln (Y-axis) for 45 min (X-axis). Time course of aggregation with 20 J-LM dolastatin 10 0, plus 0.1 J-LM vitilevuamide ... , 0.5 J-LM vitilevuaIllide ., 1 ~lM vitilevuamide 0, 2 J-LM vitilevuaInide 0, and 5 J-LM vitilevuamide L\. This graph represents one of three virtually identical experiments. B. Curve fit for plot in A used to detennine aggregation rates. Twenty JlM tubulin were incubated either alone 0 or with 20 J-LM dolastatin 10 ° at rOOl11 temperature in the absence of GTP and turbidity measured on a 96-well plate reader at 350nm (Y-axis) for 45 min (X-axis). Time course of aggregation with 20 J-LM dolastatin 10 0, plus 0.1 JlM vitilevuamide ... , 0.5 J-LM vitilevUaIllide ., 1 J-LM vitilevuamide 2 J-LM vitilevuamide 0, and 5 J-LM vitilevuamide L\. This graph represents one of three experiments. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. o o A. 0.01 B. E c: o II) M c::i o o 0.075 0.000 73 500 1000 3500 o~~::::=::: r--- 1000 2000 Time (5) 3000 I 74 Table 2.3 Reduction in the Extent and Rate ofDolastatinlO-Induced Tubulin Aggregation by Vitilevuamide Concentration I Extent Di fferent from I Rate of Di fferent from Dolastatin 10 Aggregation Control Aggregation Control (~LM) (O.D.) (Mean ± (t-test) (S-I) (Mean ± (t-test) SD) SD) o (Control) 0.0772 ± I 0.001028 ± 0.01404 I 0.0001457 0.1 0.0795 ± No I 0.001038 ± No 0.01008 0.7918 I 3.9940e-005 P= 0.8979 0.5 0.05875 ± Yes I 0.0006297 ± Yes 0.006131 P= 0.0459 I 7.7380e-005 0.0018 1.0 0.0372 ± Yes I 0.0002336 ± Yes 0.004147 p= 0.0003 I 0.0001115 P<0.0001 2.0 0.0142 ± Yes I 0.0003214 ± Yes 0.009066 P<0.0001 0.0004238 P= 0.0078 5.0 0.00246 ± Yes I Not detemlined Not Determined 0.003310 P<O.OOOI I 75 Figure 2.12 ICso calculation of vitilevuamide inhibition of dolastatin 10 induced tubulin aggregation. The total extent of aggregation (total increase in absorbance) was used to ca1culate the ICso. The ICso value \vas calculated by plotting the concentration of vitilevualnide (X-axis) versus percent of control aggregation (Y-axis) (dolastatin 10 alone). Each point represents the average of at least three individual experilnents. Curves were fit using GraphPad Prism version. 3.02, GraphPad Software, San Diego CA USA. 125 5 100 ~ CO C) Q) C, 75 C) <C o J:; 50 c o () ~ 25 ICso= 745nM • O+-------~------~------~------~------~----~ o 234 5 6 Vitilevuamide (IJM) 76 77 Vitilevuanlide's Effect on Tubulin Denaturation Tubulin in solution at roonl temperature denatures over time. Cooling the solution on ice slows down tubulin denaturation while heating tubulin to 37°C speeds up the denaturation. As tubulin denatures, hydrophobic regions on the tubulin heterodimer become nlore exposed and can be bound by the tluorescent lnolecule bis-ANS (73, 74). Upon binding hydrophobic sites on tubulin, bis-ANS fluorescence is greatly enhanced. Drugs that bind tubulin can slow this denaturation, which is analyzed by measuring bis­ANS binding. Vitilevuamide was examined for its ability to inhibit tubulin denaturation over time. Twenty J.lM tubulin were incubated with 12 J-lM vitilevuamide over the course of240 min at 37°C. Controls only contained DMSO «1 %). Every 60 nlin, an aliquot was renl0ved and the tubulin concentration diluted to 2 J.lM. Bis-ANS (5 mM final) was added and fluorescence was ilnmediately read using an excitation wavelength of 385 nm and an el1lissions wavelength of 490 mn. An increase in fluorescence compared to the 0 nlin tinle point was an indication of tubulin denaturation. Figure 2.13 denlonstrates vitilevumnide's ability to partially inhibit the time-induced and tenlperature-induced denaturation oftubulin. A t-test was perfonned and found that at each tinle point the anlount of denaturation ofvitilevualnide-treated tubulin was significantly different fronl control (P< 0.05). Dolastatin 10 was also tested in this assay and was found to cOJnpletely inhibit the denaturation (data not shown). Quenching ofTubulin Fluorescence Comparison of vitilevuanlide with known agents that protect surface-exposed tryptophans and tyrosines will help detennine if vitilevuanlide binds in the same region 78 Figure 2.13. Prevention of time and temperature dependent denaturation oftubulin by vitilevuaInide. In order to quantify the rate oftubulin denaturation, PC tubulin (20 ~lM) was incubated at 37°C in PEM 100 (lOa mM Pipes, pH 6.9,1 mM MgCb, ImM EGTA) in the absence. or presence _ of 12 ~lM vitilevuamide. Every 60 min, 100 ~L oftubulin incubation was diluted to 2 ~M and mixed with 5 ~M bis-ANS. Fluorescence was determined using an excitation wavelength of 385 nM and emissions wavelength of 490 nM. Samples without bis-ANS or without tubulin were used as blanks. Each time point was subtracted from the 0 min. time point and plotted on the graph as incubation time (X­axis) versus fluorescence intensity at 490 nm (Y-axis). Each point represents the average of three individual experin1ents. * Indicates a significant difference (P< 0.005) between control. and vitilevualnide _ treated groups. 79 20000 * * 18000 16000 * 14000 e = Q "0I!'!\t' 12000 -.-DMSO (Control) ".... ~ ">..... -ll-Vitilevuamide (12 ~M) .~ r:I'l == 10000 ~ ".... ,.=..=. * .=0~= 8000 r:I'l r:I'l .~ e ~ 6000 4000 2000 o 50 100 150 200 250 Time (Minutes) 80 as these agents. Vinca site agents have been shown to quench approximately three surface-exposed tryptophans on tubulin regardless of the ability to form tubulin aggregates. Colchicine site agents do not have this property (75). Vitilevuamide \vas incubated \vith 5 J.lM tubulin for 30 min and tryptophan and tyrosine fluorescence was lneasured by exciting at 284 nm and reading the emission from 300 nm to 420 nm. Figure 2.14 represents the tryptophan and tyrosine emission of tubulin ., tubulin incubated with 5 ~tM ., 10 J.lM ., 15 J.lM x, 20 ~tM * vitilevuamide, and 16 ~M • vinblastine. This plot illustrates that vitilevuamide does not quench any solvent-exposed tryptophans or tyrosines on the tubulin heterodinler. Vinblastine was very effective in its ability to quench tryptophans and tyrosines on tubulin. This result is unusual since it is believed that vitilevuamide binds in a region close to the vinblastine site (discussed below). Further work needs to be done to determine vitilevuamide's exact site of binding. Effects of Vitilevuanlide on the Drug Binding DOlnains of Tubulin Several sites of drug interaction with tubulin have been described. Ligand binding assays using colchicine, vinblastine and dolastatin 10 were employed to determine ifvitilevuamide occupied the colchicine, vinca or dolastatin 10 sites. Vinblastine binding to tubulin was inhibited by vitilevuamide. Lineweaver-Burke analysis yielded lines that intercepted at the negative abscissa, indicating that this inhibition was noncompetitive (Figure 2.15A). An apparent Ki was calculated for vinblastine from the equation V max (inhibitor) = V max (no inhibitor)/(l +([l]/Ki». V max 81 Figure 2.14. Effects ofvitilevuanlide on the fluorescence oftubulin. Vitilevuamide was incubated with 5 ~tM tubulin for 30 min. and the tryptophan fluorescence was nleasured by exciting the solution at 284 nm and reading the emissions frOln 300nm to 420nm. Figure 2.11 represents the tryptophan and tyrosine elnissions oftubulin +, tubulin incubated with 5 ~M ., 1 0 ~M ~, 15 ~M x, 20 ~M * vitilevuamide, and 16 ~M • vinblastine. The graph illustrates that vitilevuamide does not quench any tryptophans or tyrosines on the tubuhn dimeI'. 12000 10000 8000 Vi ."":::: = ~ QJ ~ = QJ ~ Vi QJ :.. 6000 0 = ~ QJ .:: ....... «I Q) ~ 4000 2000 t ex ,::. . ~ . • 295 .. ._ e. 315 • • • .. 335 -. • e. e 355 • Control (DMSO) ~ Vitilevualnide 5 JlM Vitilevualnide 10 JlM x Vitilevualnide 15 JlM >K Vitilevuamide 20 JlM • Vinblastine 16 JlM 375 395 415 Eluission Wavelength 82 83 values were calculated from the y-intercepts found from the regression analysis. This equation yielded an apparent Ki value for inhibition of [3H] vinblastine binding of 2.4 7 ~M. Vitilevuamide did not inhibit the binding of dolastatin 10 to tubulin at low concentrations. Inhibition of binding was seen at higher concentrations; however, no clear dose-dependent relationship could be detem1ined (Figure 2.1SB). Since a simple binding relationship could not be determined between vitilevuamide and dolastatin 10, an apparent Ki value was not calculated. It was noted that higher concentrations of vitilevuan1ide and dolastatin 10 pron10te tubulin aggregation. It is probable that tubulin aggregation resulted in the decreased binding of dolastatin 10 seen at higher concentrations of viti levu amide. A similar observation with dolastatin has been reported in the literature (104). Since all known agents that bind in the vinca domain also prevent GTP exchange on tubulin (107), the effects of vitilevuamide on the binding of radio labeled GTP to tubulin were studied in comparison to vinblastine and hemiasterlin. Vinblastine and hen1iasterlin are known to prevent GTP binding (79). Figure 2.16 illustrates the percent of GTP bound per ~g of tubulin in the presence of vitilevuamide, vinblastine and hemiasterlin. Vitilevuamide inhibited GTP binding only weakly (20%), even at concentrations as high as 80 ).lM. This result was similar to that for vinblastine, but neither vinblastine nor vitilevuamide was as effective as hen1iasterlin at inhibiting GTP binding. The ability ofvitilevumnide to prevent colchicine binding was exan1ined. Ligand binding experin1ents revealed vitilevuamide did not inhibit the binding of colchicine to tubulin (data not shown). Many tubulin active drugs, especially those that bind in the 84 Figure 2.15. of vitilevuamide on vinblastine or dolastatin 10 binding to tubulin. Each 0.5 mL reaction contained 0.1 M MES, 0.5 mM MgCh, 1.0 mg/mL tubulin, 20/0 (v/v) dimethyl sulfoxide, the indicated concentration of eH] vinblastine (A) or eH] dolastatin 10 (B), and indicated concentration of viti levu amide summarized below. Tubulin was added last in every experiment and allowed to incubate at 37°C for 45 min. Reaction mixtures were then added to 1 mL microspin columns ofBio-Gel P-30 and processed as described previously (28,29). The following concentrations of vitilevuamide in graph A were used: ., none; +, 1.56 /-lM; ., 3.4 /-lM. The following concentrations of vitilevuamide in graph B were used: ., none; ., 3.1 J.lM; + 7.8 J.lM; x, 15.8 J.lM. All lines were drawn by linear regression and R2 values for all fits were above 0.96. Each regression line represents data from a minimum of three individual experiments. 85 A. ~ 8 ~ W-l Z :-< :-< C/l ~ 6 03 ;?; > ..-1 0 4 :E 0.. ~ .>.--1, -- ~ 2 ~ f--< Oi) ::::1. o I I I ! I j I I I I I I I j I I -0 .2 o 0.2 0.4 0.6 0.8 1.2 ~ M -I V IN B LA ST IN E (1 IS) B. ~ 10 ~ 0 z f= 8 <r: f--< C/l -< ......l 0 6 Cl ......l 0 -~-e - 4 z :J >--, -' CO 2 >--, -' f--< bJ) :::i. 0 0 0.5 1 1.5 2 ~M-I DOLASTATIN 10 (liS) 86 Figure 2.16. Inhibitory effects of vitilevuamide, hemiasterlin, and vinblastine on the binding of radiolabeled GTP to tubulin. Each 0.5 mL reaction mixture contained 0.5 InghnL (5 )lM) tubulin, 50 ~tM [8}H] GTP, 0.1 M MES (l M stock solution adjusted to pH 6.9 with NaOH), 0.5 mM MgCb, 2 %( v/v) DMSO, and the indicated concentration of inhibitor (., vinblastine; ., vitilevuamide; ... , hemiasterlin). Incubation was for 10 min at 0° C. Triplicate 0.15 n1L aliquots of each reaction mixture were processed by centrifugal gel filtration on syringe-columns of Sephadex G-50 (superfine) at 4° C, as described previously (29,30). Protein and radioactivity in the filtrates were quantified. Values were then nonnalized to the control reaction mixtures. 87 100 -"...0..... 80 C Q U <~= ."5- 60 "3 .:::. = ~ OJ) e ........ ~ c = 40 Q .:c ~ ~ ~ 20 o -5 15 35 55 75 11M Drug 88 vinca domain actually strengthen the binding of colchicine to tubulin and can prevent the natural decay of the colchicine tubulin cOlllplex (67, 106, 107). In addition, all tubulin active drugs described to date that promote the aggregation of tubulin also stabilize the binding of colchicine to tubulin. Since vitilevuamide was shown to induce tubulin aggregation it was tested for its ability to stabilize colchicine binding. The results of these experiments are sunlmarized in Table 2.4. Vitilevuamide stabilized colchicine binding to tubulin when tubulin was preincubated with the vitilevuamide for 3 h prior to colchicine addition, as did vinblastine and hemiasterlin. Without a preincubation we did not observe the stabilization of colchicine binding by vitilevuamide. This requirenlent for preincubation with tubulin would indicate that vitilevuanlide needs to attain equilibriunl with tubulin in order to stabilize colchicine binding. Simply adding the two compounds together does not allow enough time for vitilevuamide to stabilize the heterodimer to allow for optimal colchicine binding. Increased Life Span of Mice with Lymphocytic Leukenlia Nude mice injected with P388 lymphocytic leukenlia cells that were treated with vitilevuanlide lived longer than mice receiving no treatment. Vitilevuanlide was able to produce a highly significant 70%) ILS (Increased Life Span) in mice treated with 30 ~g/kg/dose (Table 2.5). At doses of 12 ~g/kg/dose and 6 ~g/kg/dose, vitilevuamide also showed an ILS of 20% and 80/0, respectively. At higher doses, vitilevuamide was toxic and actually decreased the life span of these mice. These in vivo experinlents demonstrated that vitilevuaIllide is an active antineoplastic agent. Table 2.4: Stabilization of Colchicine Binding Activity of Tubulin by Vitilevuamide, Vinblastine, and Hemiasterilina DRUG None Hemiasterlin Vinblastine Vitilevuamide pmol [3H] Colchicine bound per pmol tubulin NOT PREINCUBA TED 0.22 0.24 0.21 0.20 PREINCLTBA TED 0.09 0.22 0.19 0.20 89 a Each O.l-rrll reaction mixture contained 0.4 mghnl tubulin, 0.1 M MES (pH 6.4),0.1 M EDTA, 1 mM GTP, 0.5 mM MgCh, 1 mM 2-mercaptoethanol, ImM EGTA, YYO DMSO, 60 ~M [3H] colchicine, and the indicated drug at 50 ~M. If indicated, the reaction n1ixtures were preincubated for 3 hours at 37°C prior to the addition of the [3H] colchicine. Incubation was for 2 h at 37°C, after the addition of [3H] colchicine. The data presented in the table represent average values obtained in two independent experiments, each of which contained triplicate samples. Table 2.5 Antitumor Activity of Viti levu amide Against Murine P388 Lymphocytic Leukemiaa Dose (J.lg/kgl dose) 130 60 30 12 6 ILS max b (%) 45(Toxic) 13(Toxic) 70 20 8 a P388 lYI11Phocytic leukemia (1 x l06/mouse) cells were inoculated i.p. on day O. Drug was injected on days 1, 5, and 9. b ILSmax, nlaximal increase in life span over control 90 91 Discussion In this chapter, it was shown that vitilevuamide is cytotoxic to several human tunl0r cell types and CHO cells in culture. The LCso for cell cytotoxicity was in the range of 3-300 nM. The potent antimitotic activity of natural products is strongly suggestive of an interaction oftubulin (107); however other mechanisms of action are possible. Subsequent experilnents have shown that this toxicity results from vitilevuamide's ability to bind and inhibit tubulin polymerization. Using FACS and cell cycle analysis, it was found that vitilevuamide produced a selective accumulation of CHO cells in the G2/M phase of the cell cycle. In cultures treated for 16 h with 15 I-lM vitilevuamide, 51.60/0 of the cells accumulated in G2/M, compared with 17.00/0 in controls. The increase in G2/M cells was accompanied by a decrease in Go/G] cells. The results of F ACS analysis indicated that vitilevuamide inhibited cell cycle progression at mitosis. Further analysis needs to be performed in order to determine the mitotic index of cells treated with vitilevuamide. Based on its ability to disrupt microtubule function in rat glioma cells (data not shown) (98), experiments were done to investigate whether vitilevuamide could directly affect microtubule polymerization. In a systeln using isolated bovine brain microtubule protein consisting oftubulin and MAPs, vitilevuamide inhibited the extent of microtubule asselnbly in vitro. The ICso concentration was approxinlately 2 I-lM. Vitilevuanlide was also tested in a MAP-free polyn1erization assay and had an ICso of approximately 6.7 I-lM. The ICso found in the MAP-free system was approximately three times the ICso value found in the MAP enriched systen1. This higher ICso is due to glutamate's ability to strongly polytnerize tubulin, more than MAP-induced polymerization. The ICso for 92 vitilevuamide was similar to rhizoxin (6.9 J.lM) and halichondrin B (7.2 J.lM). Vitilevuamide was a less potent inhibitor oftubulin polymerization than maytansine (3.4 JlM), dolastatin 10 (1.2 J.lM), phomopsin A (1.4 J.lM), and vinblastine (1.5 J.lM) (67, 87, 95). The rate of nlicrotubule polymerization was also affected by vitilevualnide. At concentrations above 1 JlM a significant (P< 0.05) decrease in the rate of microtubule polytllerization was observed. This result is different from the analysis looking at the extent of polymerization. A significant (P< 0.05) decrease in the extent of polynlerization was observed at concentrations of vitilevuamide above 2.67 JlM. The difference in vitilevuamide concentrations would indicate that vitilevuamide has an effect on the microtubule dynamics below the eoncentration needed to see a significant decrease in the extent of polymerization. At these lower concentrations, vitilevuamide is able to disrupt the dynamics of the microtubule but is unable to inhibit the complete formation of nlicrotubules. It may be that by sinlply disnlpting the rate of polymerization in the cell vitilevuatnide is able to prevent normal cellular processes, such as proper separation of chromosomes during mitosis. The effect of viti levu amide on the critical concentration oftubulin polymerization was also established. Nucleation is a key step in the assembly of nlicrotubules and occurs above a nlininlum tubulin concentration called the critical concentration. No polytnerization will occur when the concentration of available tubulin falls below this concentration. Above the critical concentration a certain amount of tubulin polymerizes and the nlicrotubules are at a steady state with a constant concentration of tubulin equal to the critical concentration. Vitilevuamide increased the critical concentration in a 93 substoichiometric range. This is similar to many other tubulin-binding agents, but unlike cryptophycin 1, which causes substoichiometric reductions in the maxirnum extent of assenlbly (l08). Vitilevuamide was a strong inducer of the extent oftubulin aggregation at both stoichiometric (20 J.lM) and superstoichiometric (40 ~lM) concentrations. In these experiments, vitilevuamide significantly (P< 0.05) induced aggregation to a greater extent than both vinblastine and dolastatin 10 at a concentration of 20 J.lM. Interestingly, 20 ~lM vitilevuamide induced aggregation at a slower rate than 20 J.lM vinblastine induced aggregation. The rate of dolastatin 10 induced tubulin aggregation was slower than both vitilevuanlide and vinblastine. It has not yet been determined whether the turbidity increase caused by vitilevuamide represents distinct structures, like those formed by either vinblastine (spiral) or dolastatin 10 (rings). Electron microscopy is needed to detenl1ine the structure of aggregates fonned by vitilevuamide. The aggregation of tubulin in this study was performed at room temperature (22°C). Other studies may be performed to determine if incubation at 4 °C affects vitilevuamide-induced tubulin aggregation. Dolastatin 10 turbidity is greatly decreased at cold temperatures while vinblastine-induced tubulin aggregation is less affected (95). Reactions done at 37°C enhanced tubulin aggregation of both vinblastine and dolastatin 10 and aggregate fonl1ation were not cold reversible. Looking at how aggregation oftubulin occurs at these alternative tenlperatures may give some insight into the nature of the aggregates fonned by tubulin. One could speculate that vitilevuamide will behave lTIOre like dolastatin 10 and less like vinblastine since both vitilevuamide and dolastatin 10 are peptide agents. Testing the temperature effects on vitilevuamide-induced tubulin aggregation will deternline if vitilevuamide functions more like dolastatin 10 or vinblastine. 94 As mentioned earlier, drug induced aggregation can be inhibited by substoichiometric amounts of a second tubulin active compound. Inhibition of the extent oftubulin aggregation induced by vitilevuamide was tested using both vinblastine and dolastatin 10. At all the concentrations tested, dolastatin 10 significantly (P< 0.05) inhibited the extent of viti levu amide-induced tubulin aggregation. However, dolastatin 10 did not significantly inhibit the rate of viti levu amide-induced tubulin aggregation at 1.25 ~M, indicating that although the amount of aggregates fonned was reduced, the rate at which they were produced was not changed. Vinblastine significantly (P< 0.05) reduced both the rate and extent of vitilevuamide-induced tubulin aggregation at all concentrations of 1.25 ~M and above. Dolastatin 10 (subnlicronlolar) was a much more potent inhibitor of the extent of vitilevuaInide-induced tubulin aggregation than vinblastine (ICso 2.28 ~M). This may be due to the positioning of the binding sites of these compounds on ~-tubulin. Vitilevuamide is a nonconlpetitive inhibitor of vinblastine binding, as is dolastatin 10. Although no binding relationship could be determined for vitilevuanlide and dolastatin 10, one may speCUlate their binding sites may be in closer proximately to each other than the vinblastine and vitilevuamide binding sites are. If so, then lower concentrations of either vitilevuamide or dolastatin 10 would be required to inhibit the aggregation induced by either drug when compared to vinblastine. When vitilevuanlide was tested for its ability to inhibit dolastatin 10 induced aggregation, it was found to have nearly identical ICso values to dolastatin 10 inhibition ofvitilevuamide induced aggregation (754 nM and 95 529 nM (estimate) respectively). Vinblastine was a nluch weaker inhibitor (2.28 /-lM) of vitilevuamide-induced aggregation, nearly five tinles less potent than dolastatin 10. Again this may indicate relative positioning of binding of these three compounds on p tubulin. Studies with two spongistatin compounds, spongistatin 1 and 6, conlpared well to vitilevuaIllide in their ability to inhibit dolastatin 10 aggregation. Spongistatin 1 was found to maximally inhibit dolastatin aggregation at 2.5 /-lM and spongistatin 6 at 1 0 ~M (68). Vitilevuamide nlaximally inhibited at 5 /-lM, making it less potent than spongistatin 1 and l110re potent than spongistatin 6. Another compound, cryptophycin 1, also inhibits dolastatin 10 aggregation with a maximal inhibitory concentration of 2 /-lM (71). Many compounds that bind to tubulin prevent the time- and temperature­dependent denaturation of tubulin. These include phomopsin A (l09), dolastatin 10 (110), cryptophycin 1 (108), and vinblastine (110). It was shown that vitilevuamide also possessed this property, but that it was less effective than dolastatin 1 0 (data not shown). Another study exanlining both vinblastine and dolastatin 10 also found no detectable increase in denaturation when incubated with dolastatin 10 but a slight increase was observed with vinblastine (110). This increase in tubulin denaturation was very similar to vitilevuamide although the vinblastine concentration used was slightly higher (20 ~lM) than vitilevuamide (12 /-lM). Another paper (111) found that cryptophycin 52 was sil11ilar to vinblastine in preventing denaturation, while phomopsin A was as potent as dolastatin 10 (106). Thus it appears that tubulin-binding drugs can be classi fied into three groups according to their effects on tubulin denaturation, those that are strong inhibitors, weak inhibitors and those with no inhibition. In terms of inhibition of tubulin denaturation, pholl1opsin A = dolastatin 10> vitilevuamide = vinblastine cryptophycin 52. 96 Many compounds that bind tubulin cause a quenching of solvent-exposed tryptophans and tyrosines on the tubulin dimer. Vinca site agents, such as vinblastine and nlaystatine, quench tryptophans with the greatest efficacy while others are less effective. It was interesting that vitilevuamide did not quench any surface tryptophans or tyrosines on tubulin because it is thought to bind in the region close to the vinblastine site. It is possible that vitilevuanlide binds to a site near vinblastine, but far enough away that it has no contact with the tryptophans protected by vinblastine. Until further tests can be perfonned, the exact nature of this binding cannot be deternlined. Vitilevuamide is a new member of the group of drugs that inhibit vinca alkaloid binding to tubulin. Kinetic analysis of viti levu amide binding revealed that inhibition of vinblastine binding was noncompetitive. It has previously been shown that the natural products dolastatin 10, halichondrin B, phonlopsin A, spongistatin 1 and cryptophycin 1 are also noncompetitive inhibitors of the vinca alkaloid binding site, while the macrolides maytansine and rhizoxin inhibit in a conlpetitive fashion (8, 13, 30, and 34). The current results suggest that vitilevuamide binds in the vinca donlain in a site distinct from, but in close proximity to the vinca site. Using the V max values obtained from the Lineweaver­Burke plot, an apparent Ki of 2.4 7 ~M for vitilevuanlide inhibition of eH] vinblastine binding was calculated. This result compares well to Ki values reported for other noncompetitive inhibitors of vinblastine binding. Vitilevuamide is a weaker inhibitor of vinblastine binding than spongistatin 1 (Ki, 1.3 ~M) (13) and stronger than halichondrin B (Ki, 5.0 ~M) (8). The competitive inhibitor Inaytansine (Ki, 0.9 ~M) (8) and vincristine (Ki, 1.8 ~M) (13) are stronger inhibitors of vinblastine binding. The effect of vitilevuanlide on dolastatin 10, another tubulin-binding nlarine 97 peptide with anticancer potential, was probed. Our results show that vitilevuamide inhibited dolastatin binding only at concentrations that induced tubulin aggregation. It is possible that an inhibitory binding relationship exists between vitilevuamide and dolastatin 10, however experimental conditions allowing aggregation prevented its detection. Thus, no clear relationship for inhibition of dolastatin 10 binding to tubulin could be determined. Bai et al. have shown that spongistatin 1 noncompetitively inhibits the binding of both vinblastine and dolastatin to tubulin (13), while cryptophycin 1 cOinpetitively inhibits the binding of dolastatin 10 to tubulin (34). It may be possible that vitilevuatnide binds in the same site as spongistatin 1 because there exists a sinlilarity in the cyclic ring size of spongistatin 1 and vitilevualnide. Future studies will require a radio labeled form of spongistatin or vitilevuamide to determine this. Like phomopsin A, dolastatin 10, and vinblastine; vitilevuanlide enhanced colchicine binding to tubulin, although vitilevuamide required a preincubation period in order to exhibit this enhancement. Stabilization of colchicine binding is a property conlnl0n to drugs that modulate the vinca alkaloid domain of tubulin. All agents that prevent the tinle dependent decay of colchicine binding to tubulin (vinblastine, dolastatin 10, pholnopsin A, and cryptophycin 1) have also been shown to induce tubulin aggregation. Vitilevuanlide falls within this class, producing tubulin aggregation at high concentrations, as seen by light microscopy. The exact structure of these aggregates is unclear and further investigation is needed to determine if the aggregates are similar in structure to those produce by dolastatin 10, cryptophycin 1 or vinblastine. Binding of GTP to tubulin is weakly inhibited by viti levuamide, vinblastine and cryptophycin 1 (35), but is strongly inhibited by other natural products such as 98 hemiasterlin, spongistatin 1, phomopsin A and dolastatin 10. Future studies will determine if preincubation of vitilevuamide with tubulin will enhance its inhibition of GTP binding, as seen with cryptophycin 1 (34). The data presented here suggest that vitilevuanlide binds tubulin close to the so-called "peptide site" within the vinca donlain, and that the precise interaction Inay be unique fronl other tubulin active conlpounds. The potent cytotoxic profile of vitilevuamide nlakes it a viable candidate for lise as an anticancer agent in vivo. Results obtained in mice harboring P388 lymphocytic leukemia suggest that vitilevuamide does indeed have anticancer potentia1. Vitilevuamide was potent in its ability to increase the life span of leukemic nlice and further in vivo testing is warranted. In conclusion, vitilevuatnide is a structurally unique cytotoxic nlarine metabolite that inhibits tubulin polynlerization as its mechanism of action. Some of vitilevuamide' s properties relevant to this study are cOlnpared with other tubulin agents in Table 2.6. It pronl0tes the aggregation of tubulin at roonl temperature and inhibits the aggregation of tubulin induced by dolastatin 10. Both vinblastine and dolastatin 10 inhibit vitilevuamide induced tubulin aggregation. Like many of the complex group of natural products, vitilevuamide inhibits the binding of vinca alkaloids to tubulin and inhibits nucleotide exchange (although weakly). It was shown that vitilevuamide does not inhibit colchicine binding, but rather stabilizes it when preincubated with tubulin. This coincides with vitilevuamide's ability to prevent tubulin denaturation. Ligand binding assays determined that vitilevuamide interacts with tubulin at a site that is distinct fr01TI vinblastine, colchicine, dolastatin 10 or GTP. This is partially confirmed by tubulin quenching assays that showed vitilevuamide does not quench surface exposed Drug Inhibition of Tubulin polymerization (ICso) Dolastatin 10 .59 ~M Phomopsin A 2.8 ~M Cryptophycins 1-5 ~M Hemiasterlins 0.98 ~M Spongistatin 3.6~M Taxol Promotes Vitilevuamide 2.0 ~M-6.7 ~M Table 2.6 Major Characteristics of Tubulin Binding Agents Compared to Vitilevuamide Induces Inhibits Inhibits Stabilizes Aggregation Vinblastine Dolastatin 10 Tubulin Induced Tubulin Induced Denaturation Aggregation Tubulin Aggregation Yes Yes -- Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes ND Enhances N/D No ND Yes ND Yes ND ND ND Yes Yes Yes Yes Protects Tryptophan Fluorescence Yes Yes Yes N/D ND ND No \0 \0 100 tryptophans or tyrosines, like many other vinca site agents. Unfortunately, vitilevuamide is structurally cOlnplex, preventing easy chemical synthesis. Also, the low yield of vitilevuamide obtained from the ascidians in which it is found makes large-scale collection inefficient. However, the isolation of viti levu amide from two different genera oftunicates suggests the possibility that it is the product of a microbial symbiont. If this is the case, vitilevuamide could be produced in fermentation culture, and n1ay fonn the prototype for a new class of anticancer agents. CHAPTER 3 PHAGE DISPLAY SCREENING OF TLTBULIN Introduction The inlportance oftubulin as a drug target Inake agents that bind tubulin and either inhibit or modulate tubulin-protein interactions an area of great scientific interest. Peptides with high affinity for tubulin may help characterize tubulin-protein interactions or provide new targets for tubulin inhibiting drugs. In order to find tubulin interactive peptides, tubulin was subjected to several rounds of affinity purification using an M13 bacteriophage library. The following are the results of this screen and several other experinlents used to determine if the selected peptides had activity in simplified biological systems that test a compound's effects on tubulin function and biochemical properties. Materials and Methods Chelnicals and Reagents M13 bacteriophage library was a generous gift from George P. Smith (University of Missouri). Nonradiolabeled vinblastine, colchicine and GTP were obtained from Signla Chemical Co. (St. Louis, MO, USA). Nonradiolabeled dolastatin 10 was a 102 generous gift of Dr. D. J. Newman (Natural Products Branch, National Cancer Institute). 4,4' -dianilino-1, l' -binaphthyl-5, 5' -disunfonic acid (bis-ANS) was obtained from Siglna. Electrophoretically homogeneous bovine brain tubulin containing microtubule associated proteins (MTP) was isolated and purified with Ininor nl0difications according to methods described elsewhere (99,100). Purified tubulin was obtained by phosphocellulose chromatography (PC). SDS gel electrophoresis determined that PC tubulin was free of MAP protein and no contanlinating proteins were detectable even when the gel was grossly overloaded (Figure 2.2). Biopanning by Method Described by Slnlth Initial atte