Pharmac. Ther. Vol. 55, pp. 31-51, 1992
`Printed in Great Britain. All rights reserved
`
`Specialist Subject Editor: E. HAMEL
`
`0163-7258/92 $15.00
`© 1992 Pergamon Press Ltd
`
`NATURAL PRODUCTS WHICH INTERACT WITH
`TUBULIN IN THE VINCA DOMAIN:
`MAYTANSINE, RHIZOXIN, PHOMOPSIN A,
`DOLASTATINS 10 AND 15 AND
`HALICHONDRIN B
`
`ERNEST HAMEL
`
`Laboratory of Molecular Pharmacology, Developmental Therapeutics Program, Division of
`Cancer Treatment, National Cancer Institute, National Institutes of Health, Bethesda,
`MD 20892, U.S.A.
`
`Abstract-This paper summarizes published data on the interactions of tubulin with antimi(cid:173)
`totic compounds that inhibit the binding of vinca alkaloids to the protein. These are all
`relatively complex natural products isolated from higher plants, fungi and marine invertebrate
`animals. These agents are maytansine, rhizoxin, phomopsin A, dolastatins lO and 15 and
`halichondrin B and their congeners. Effects on tubulin polymerization, ligand binding
`interactions and structure-activity relationships are emphasized.
`
`CONTENTS
`
`I. Introduction
`2. The Maytansinoids
`3. Rhizoxin
`4. Phomopsin A
`5. Dolastatin 10
`6. Dolastatin 15
`7. Halichondrin B, Homohalichondrin B and Halistatins 1 and 2
`8. One Laboratory's Comparative Data
`9. A Preliminary Model of the 'Vinca Domain' of Tubulin
`Acknowledgements
`References
`
`31
`32
`35
`37
`38
`41
`42
`44
`44
`47
`47
`
`I. INTRODUCTION
`
`Antimitotic agents almost universally alter microtubule assembly reactions. With the exception
`of taxol and its congeners (Schiff et al., 1979), the primary effect of these compounds, the one
`observed at lower drug concentrations, is inhibition of the reaction. With the exception of
`estramustine and its phosphate derivative (Stearns and Tew, 1985; Wallin et al., 1985), these
`inhibitors all interact with tubulin, the predominant protein component of micro tubules. Although
`tubulin molecules are not uniform, since there is minor primary structure variability (Krauhs et
`al., 1981; Ponstingl et al., 1981) and substantial variability resulting from post-translational
`modification (Eipper, 1972; Arce et al., 1975; Maruta et al., 1986; Edde et al., 1990; Alexander et
`al., 1991), each tubulin unit in a microtubule consists of two highly similar polypeptide chains (ex(cid:173)
`and p -tubulin) and two guanosine nucleotides. The two polypeptide chains are nearly identical in
`molecular weight (about 50 kDa) and have approximately 40% absolute sequence homology and
`about 70% homology if conservative amino acid substitutions are considered (Krauhs et al., 1981;
`Ponstingl et al., 1981). One guanosine nucleotide is in the form of GOP in the microtubule while
`in unpolymerized tubulin it can be in the form of either GOP or GTP. This nucleotide is
`
`31
`
`IMMUNOGEN 2001, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`32
`
`E. HAMEL
`
`exchangeable with GOP or GTP in solution in unpolymerized tubulin (Weisenberg eta/., 1968;
`Kobayashi, 1974; Levi et a/., 1974; Caplow and Zeeberg, 1980; Lin and Hamel, 1987), but it
`becomes inaccessible in polymerized tubulin (Weisenberg et a/., 1976; Arai and Kaziro, 1977;
`David-Pfeuty et a/., 1977). The exchangeable nucleotide has been localized to the P-subunit of
`tubulin (Geahlen and Haley, 1977; Nath eta/., 1985). The second guanosine nucleotide is always
`in the form of GTP and is always inaccessible to solution nucleotides. Even in cells its half-life
`appears to be equal to that of tubulin itself (Spiegelman et a/., 1977). The location of the
`nonexchangeable nucleotide has not been determined, but the sequence homology of the tubulin
`subunits suggests it may be on cc-tubulin.
`Inhibitors of microtubule assembly fall into two broad classes, which have been essentially
`defined by their effects on the binding to tubulin of two commercially available radio labeled drugs.
`Most inhibitors of polymerization, including virtually all synthetic compounds, inhibit the binding
`of radiolabeled colchicine to tubulin and do not affect the binding of radiolabeled vinblastine to
`tubulin. Generally, despite rather diverse but relatively uncomplicated molecular structures, when
`carefully examined such compounds show a competitive pattern of inhibition against colchicine.
`Inhibitors of radiolabeled vinblastine binding are much less common. Thus far, all such
`inhibitors are natural products (or closely related synthetic analogs) and their structures are
`complex. These agents are the subject of this paper. None of them inhibit the binding of
`radiolabeled colchicine to tubulin. Their effects on vinblastine and vincristine binding fall into both
`competitive and noncompetitive patterns. In addition, all interfere with interactions of guanosine
`nucleotides at the exchangeable site. Finally, while no colchicine-site drug has a role in the
`treatment of neoplastic diseases, the efficacy of the vinca alkaloids in cancer chemotherapy (see
`Rowinsky and Donehower, 1992) lends particular interest to the diverse group of natural products
`that interfere with the tubulin-vinblastine interaction.
`
`0
`"-c-R
`I
`
`0
`
`0
`
`II
`
`R=CH-N-C-~
`
`CH
`
`I 3
`
`I
`CH3
`
`MaytanSine:
`
`Ansamitocin P-3:
`
`Ansamitocin P-4:
`
`FIG. I. Structural formulae of maytansine, ansamitocin P-3 and ansamitocin P-4.
`
`IMMUNOGEN 2001, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Interactions of tubulin in the vinca domain
`
`33
`
`2. THE MAYTANSINOIDS
`
`Maytansine (NSC 153858; structure presented in Fig. I) is the most thoroughly studied member
`of a series of ansa macrolide compounds isolated and characterized by Kupchan and his
`collaborators (Kupchan eta/., 1972, 1974, 1975, 1977, 1978). These agents were obtained from the
`plants Maytenus ovatus, M. buchananii, M. serrata and the related Putterlickia verrucosa.
`Chemically related compounds have also been isolated from Colubrina texensis, a member of a
`different plant family (Wani eta/., 1973) and from a Nocardia species (Higashide eta/., 1977; Asai
`et a/., 1979). The best studied compounds from the microorganism are ansamitocin P-3 and
`ansamitocin P-4 (structures in Fig. 1). Maytansine (Remillard eta/., 1975; Wolpert-DeFilippes et
`a/., 1975a,b; Kupchan et al., 1978; Bai eta/., 1990b) and the ansamitocins (Higashide eta/., 1977;
`Ootsu et a/., 1980) are highly cytotoxic compounds (see below) which inhibit mitosis, with cells
`accumulating in apparent metaphase arrest (condensed chromosomes; absent nuclear membrane;
`no spindle or poorly defined spindle with few if any microtubules). Except for taxol (Rowinsky
`and Donehower, 1992), maytansine is the only antimitotic agent to have gone through extensive
`clinical trials for the treatment of neoplastic diseases (e.g. Cabanillas eta/., 1979; Rosenthal eta/.,
`1980), but thus far no useful role for the drug has been established in clinical practice. Recently
`maytansine derivatives have been coupled to tumor-specific antibodies in a new approach to clinical
`use of this agent (Chari et a/., 1992).
`Radiolabeled maytansine binds to tubulin in a reversible reaction which is relatively fast, occurs
`at ooc and is inhibited by vincristine (Mandelbaum-Shavit eta/., 1976) and vinblastine (Batra et
`a/., 1986). The apparent Kd value for the reaction at 37°C is 0.7 JI.M (Mandelbaum-Shavit eta/.,
`1976). Vincristine was found to act as a competitive inhibitor of maytansine with an apparent K;
`value of 10 JI.M (Mandelbaum-Shavit et al., 1976). Neither vincristine (Mandelbaum-Shavit eta/.,
`1976) nor vinblastine (Batra eta/., 1986) displaces all radiolabeled maytansine bound to tubulin.
`This phenomenon has not been studied in detail, but may be related to observations of Takahashi
`et a/. (1987b) on apparently contradictory effects of vinblastine on the binding of radiolabeled
`rhizoxin to tubulin (competitive) as compared with the effects of rhizoxin on the binding of
`radiolabeled vinblastine (not purely competitive) (see below).
`In contrast, nonradiolabeled maytansine almost totally displaces radiolabeled vinblastine or
`vincristine from tubulin (Mandelbaum-Shavit e.t al., 1976; Batra et a/., 1986; Safa et al., 1987).
`Maytansine has been found to competitively inhibit the binding of both vinca agents to tubulin.
`Apparent K; values vs vincristine have been reported as 0.4 JI.M (Mandelbaum-Shavit eta/., 1976;
`York eta/., 1981) and 3.1 JI.M (Bai eta/., 1990a) and versus vinblastine as 0.5 JI.M (Bhattacharyya
`and Wolff, 1977) and 0.9 JI.M (Bai eta/., 1991). Lacey eta!. (1987) found that the IC50 value of
`maytansine versus radiolabeled vinblastine was 1.6 Ji.M. All workers have concluded that may(cid:173)
`tansine binds to tubulin with greater affinity than these two vinca agents. In addition, ansamitocin
`P-3 inhibits the binding of radiolabeled vinblastine to tubulin, but the data indicated that the
`inhibition was not competitive (Takahashi et al., 1987b).
`Maytansine, like vinblastine, has only minor effects on the binding of radiolabeled colchicine to
`tubulin (Mandelbaum-Shavit eta/., 1976; Batra et al., 1986; Lacey eta/., 1987). Unlike vinblastine,
`however, maytansine does not stabilize the colchicine binding activity of tubulin (Bhattacharyya
`and Wolff, 1977; Bai eta/., 1990a). Tubulin stabilization by vinblastine, but not maytansine, has
`also been demonstrated by the ability of the former, but not the latter, to reduce the rate at which
`bis(8-anilinonaphthalene-l-sulfonate) interacts nonspecifically with denatured hydrophobic regions
`of tubulin (Prasad et a/., 1986).
`Maytansine has specific effects on the alkylation of tubulin sulfhydryl groups, which have been
`studied by Luduefia and his colleagues (Luduefia and Roach, 198la,b; Luduefia eta/., 1982; Roach
`and Ludueiia, 1984). This work has been reviewed in detail in this series (Luduefia and
`Roach, 1991). In brief, these workers have evaluated the effects of a large variety of tubulin
`ligands, including many antimitotic agents, on the pattern of alkylation observed with
`iodoacetamide and on the formation of intrapolypeptide chain crosslinks in p -tubulin following
`reaction with the bifunctional agent N,N'-ethylenebis(iodoacetamide). There are two potential
`major crosslinks. The first, between cys 239 and cys 354 (Little and Ludueiia, 1985) (referred
`to throughout this paper as the 'first crosslink'), occurs in all tubulin preparations. The second,
`
`IMMUNOGEN 2001, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`34
`
`E. HAMEL
`
`between cys 12 and either cys 201 or cys 211 (Little and Ludueiia, 1987) (referred to as the
`'second crosslink'), only occurs in nucleotide-depleted tubulin with its formation strongly inhibited
`by GTP. Maytansine had minimal effects on alkylation by iodoacetamide, in contrast to strong
`inhibition by vinblastine. Formation of the first crosslink was enhanced in the presence of both
`maytansine and vinblastine, while formation of the second crosslink was almost completely
`inhibited by maytansine and weakly inhibited by vinblastine. Since the exchangeable nucleotide site
`has been localized to the f3 -subunit, this finding suggests that the maytansine binding site may also
`be on f3 -tubulin.
`Our own studies also led to this conclusion. Maytansine potently inhibits tubulin-dependent GTP
`hydrolysis (Lin and Hamel, 1981; Bai et al., 1990b) and nucleotide binding to tubulin, particularly
`at lower reaction temperatures (Huang et al., 1985; Bai et al., l990a, 1991). Maytansine does not
`displace nucleotide bound in the exchangeable site (Huang et al., 1985; Bai et al., 1990a), and,
`moreover, the drug prevents nucleotide loss from tubulin during gel filtration chromatography (Lin
`and Hamel, 1987). The most reasonable explanation for these observations is that maytansine binds
`to f3 -tubulin in a manner that prevents entry and exit of guanosine nucleotides at the exchangeable
`site. We also have evaluated vinblastine as an inhibitor of nucleotide exchange on tubulin (Huang
`et al., 1985; Bai et al., 1990a, 1991). Although the vinca alkaloid has some effect on the reaction,
`significant inhibition is only observed at high concentrations (at least 100 /lM).
`In contrast to results from the alkylation and GTP exchange experiments with maytansine,
`however, are genetic results. The only maytansine-resistant mutant reported thus far has been a
`Chinese hamster ovary cell line with an altered oc-tubulin (Schibler and Cabral, 1985). Takahashi
`eta/. (1990) have, however, described an interesting manipulation by site-directed mutagenesis in
`a /3-tubulin gene of Schizosaccharomyces pombe. Ileu 100 was converted to asn 100 (analogous to
`the situation in mammalian f3 -tubulin) and an organism with increased sensitivity to ansamitocin
`P-3, as well as rhizoxin (see below), was produced.
`Maytansine and the ansamitocins inhibit microtubule assembly requiring microtubule-associated
`proteins (Remillard eta/., 1975; Bhattacharyya and Wolff, 1977; Kupchan eta/., 1978; Ootsu eta/.,
`1980; York eta/., 1981; Fellous eta/., 1985; Huang eta/., 1985; Lacey eta/., 1987; Takahashi et
`a/., 1987a) and the glutamate-induced polymerization of purified tubulin (Bai eta/., 1990b). With
`microtubule-associated proteins, a higher concentration of maytansine is required with micro(cid:173)
`tubule-associated protein 2 as compared with tau factor to obtain equivalent inhibition (Fellous
`et a/., 1985). With the exception of one report where an IC50 value for polymerization of 0.4 /lM
`was obtained (Bhattacharyya and Wolff, 1977), most workers have obtained IC50 values for
`maytansine and the ansamitocins in the 1-5 /lM range (Remillard eta/., 1975; Ootsu eta/., 1980;
`Huang et al., 1985; Lacey eta/., 1987; Takahashi eta/., 1987a; Bai et al., 1990b). Despite binding
`more avidly to tubulin than vinblastine (see above), in direct comparisons maytansine (and the
`ansamitocins) has always been found to be less effective than vinblastine as an inhibitor of
`polymerization (Bhattacharyya and Wolff, 1977; Ootsu et al., 1980; Huang et al., 1985; Lacey et al.,
`1987; Takahashi eta/., 1987a; Bai et a/., 1990b).
`/lM cause extensive
`Concentrations of both maytansine and the ansamitocins below 20
`disassembly of preformed microtubules (Remillard et al., 1975; Ootsu et a/., 1980) and the
`ansamitocins cause intracellular microtubules to disappear (Ootsu et a/., 1980).
`Unlike the vinca alkaloids (reviewed by Himes, 1991), neither maytansine nor the ansamitocins
`induce formation of spiral aggregates oftubulin (Bhattacharyya and Wolff, 1977; Ootsu eta/., 1980;
`Fellous et al., 1985; Takahashi et al., 1987a). Instead maytansine potently inhibits this reaction,
`for maytansine at concentrations substoichiometric to those of both tubulin and vinblastine (e.g.
`2 /lM maytansine, 10 /lM tubulin, 140 /lM vinblastine) totally prevents tubulin spiralization (Fellous
`et al., 1985). In addition, maytansine causes dissolution of preformed vinblastine-induced
`aggregates (Fellous et al., 1985).
`There is minimal information about tubulin interactions with any maytansinoids other than the
`ansamitocins, which differ solely in the ester substituent at position C(3) (see Fig. 1). York et al.
`(1981) examined four analogs as inhibitors of vincristine binding to tubulin. A compound with an
`altered ester substituent at position C(3) was essentially equivalent to maytansine, but three analogs
`with no substituent at this position and a 2-3 double bond were significantly less potent as
`inhibitors. One of these (which also lacked the N-methyl group and the 4-5 epoxide) had an
`
`IMMUNOGEN 2001, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Interactions of tubulin in the vinca domain
`
`35
`
`apparent noncompetitive pattern of inhibition, but its K; value was 5 J.lM as opposed to the 0.4 J.lM
`value obtained for maytansine. Kupchan et al. (1978) obtained initial results with a large series of
`analogs. They compared inhibitory effects on microtubule assembly, mitosis of fertilized sea urchin
`eggs and growth of KB carcinoma cells in culture. There was little correlation between effects
`observed in the three assays and they only described compounds with significant inhibitory effects
`on the assembly reaction. The following modifications (see Fig. I) did not seem to reduce
`maytansine's effectiveness as an inhibitor of assembly: altered ester at position C(3), hydroxyl group
`at position C(3), etherification of the hydroxyl at position C(9), thioether instead of hydroxy group
`at position C(9), no substituent at position C(3) together with introduction of a 2-3 double bond
`(in contrast to the finding of York eta/., 1981), noN-methyl group and acetyl group at position
`C(l5).
`The crystal structure of maytansine 3-bromopropyl ether has been summarized (Kupchan et a/.,
`1972, 1974; Bryan et al., 1973). Substituents are oriented in a manner that minimizes intramolecular
`repulsions. The two longer sides of the 19-member ring are approximately parallel and about 5.4
`A apart. The ring appears to be open in its center. Thr face of the ring with the ester group is
`relatively hydrophilic, the opposite face hydrophobic.
`
`3. RHIZOXIN
`
`The fungus Rhizopus chinensis is the etiologic agent for a disease known as rice seedling blight
`in which there is an abnormal swelling of plant roots secondary to the failure of cell division
`(Ibaragi, 1973). A number of compounds which reproduced the disease process were isolated and
`characterized by Iwasaki and coworkers (Iwasaki eta/., 1984, 1986a,b; Kobayashi eta/., 1986). The
`most important of these agents was called rhizoxin (NSC 332598; structure in Fig. 2). The most
`prominent structural feature of rhizoxin is its 16-member macrolide ring. Besides its toxicity for
`plant tissues, rhizoxin also has antitumor (Tsuruo eta/., 1986) and antifungal (Iwasaki eta/., 1984)
`activity and it causes the accumulation of cells arrested in mitosis (Tsuruo eta/., 1986; Bai eta/.,
`1990b). Rhizoxin is in the early stages of clinical evaluation in human cancer patients (Bisset eta/.,
`1992).
`Radiolabeled rhizoxin binds rapidly to tubulin at 37°C and the reaction is reversible. Scatchard
`analysis of binding data indicated one high affinity binding site with a Kd value of0.2 J.lM (Takahashi
`et a/., 1987b). Binding of the drug to tubulin was inhibited by both vinblastine and ansamitocin
`P-3 (Takahashi eta/., 1987b). Lineweaver-Burk analysis indicated that both inhibitors were acting
`in a competitive manner, with apparent K; values of 0.1 J.lM for ansamitocin P-3 and 3 J.lM for
`vinblastine (Takahashi et al., 1987b). The binding of radiolabeled rhizoxin to tubulin is also
`strongly inhibited by phomopsin A (Li eta/., 1992). The IC50 value obtained with 3 J.lM rhizoxin
`was just over 0.1 J.lM for phomopsin A, but the type of inhibition (competitive or noncompetitive)
`was not defined in this study.
`
`0
`
`FIG. 2. Structural formula of rhizoxin.
`
`IMMUNOGEN 2001, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`36
`
`E. HAMEL
`
`Takahashi et a/. (1987b) reported that inhibition by rhizoxin of the binding of radiolabeled
`vinblastine to tubulin did not appear to be purely competitive, although the apparent Ki value for
`rhizoxin was quite low, about 0.01 JlM. In my laboratory we have found that rhizoxin was a classic
`competitive inhibitor of the binding ofradiolabeled vincristine to tubulin, with an apparent~ value
`of 12 JlM (Bai et a/., 1990a).
`Rhizoxin does not inhibit the binding of colchicine to tubulin (Takahashi eta/., 1987b) nor does
`it stabilize the colchicine site (Bai et a/., 1990a). Further, it does not alter the rate of interaction
`of bis(S-anilinonaphthalene-1-sulfonate) with tubulin (Sullivan et a/., 1990).
`The effect of rhizoxin on the alkylation of tubulin sulfhydryls in terms of crosslink formation
`was examined and qualitatively it was found to behave identically to maytansine, with increased
`formation of the first crosslink and markedly reduced formation of the second crosslink (Sullivan
`eta/., 1990).
`Rhizoxin, like maytansine, inhibits tubulin-dependent GTP hydrolysis (Bai et a/., 1990b) and
`guanosine nucleotide exchange at the exchangeable site (Bai eta/., 1991). It is, however, less potent
`than maytansine as an inhibitor in both reactions (see below).
`In an elegant series of experiments Takahashi and coworkers (Takahashi et a/., 1989,
`1990) isolated rhizoxin-resistant mutants of Aspergillus nidulans and found an altered {3 -tubulin.
`This was demonstrated to result from replacement of asn 100 with an isoleucine moiety.
`Further, they found that both Schizosaccharomyces pombe and Sacchoromyces cerevisiae were
`naturally resistant to rhizoxin and lacked asparagine at position 100. When site-directed muta(cid:173)
`genesis was employed to place asparagine at position 100 in both yeasts, rhizoxin-sensitive strains
`were obtained.
`Microtubule assembly was inhibited by rhizoxin with an IC50 value of 5 J,LM; the agent was less
`potent than vinblastine and equivalent to ansamitocin P-3 (Takahashi et a/., 1987a). It was
`somewhat less potent as an inhibitor of the glutamate-induced polymerization of purified tubulin.
`Its IC 50 value was about 7 JlM, approximately 6 times that of vinblastine and twice that of
`maytansine (Bai eta/., 1990b, 1991). Rhizoxin also causes extensive depolymerization of preformed
`microtubules (Takahashi et a/., 1987a).
`There is no evidence that rhizoxin induces formation of spiral aggregates (Takahashi et al.,
`1987a,b). In fact, like maytansine, rhizoxin potently inhibits the vinblastine-induced reaction, for
`the drug at 2 JlM eliminated spiral formation with 10 JlM tubulin and 100 JlM vinblastine (Takahashi
`eta/., 1987b).
`Several analogs of rhizoxin have been evaluated for their effects on microtubule assembly
`(Takahashi et al., 1987a). Equivalent activity was observed in compounds in which the meth(cid:173)
`oxy substituent at position C(l7) was replaced with an hydroxyl group, in which the 2-3 epoxide
`was replaced with a double bond, or in which both epoxide groups were replaced with double
`bonds. A four-fold loss of activity was observed when the lactone group was disrupted. Major loss
`of activity occurred when either epoxide was hydrolyzed and replaced with an hydroxyl group at
`position C(3) or C(12), or if the position C(13) hydroxyl group was esterified.
`The side chain of rhizoxin has been studied for its role in the compound's antitubulin
`properties (Kato et a/., 1991). Significant loss of activity in inhibiting microtubule assembly
`occurred with all compounds with less than six carbon atoms in the side chain. That is, C(16)
`through C(21) are required for good antitubulin activity. Replacement of the distal side chain with
`a carbonyl oxygen at C(20) led to complete loss of activity, while its replacement with a methylene
`group (that is, restoring C(21) alone) yielded a compound with almost full activity. Reduction
`of the inactive C(20) ketone to the alcohol did not restore activity, but several esters of this
`alcohol had partial activity. All changes made in the distal portion of the side chain, beyond
`C(21), had relatively minor effects on inhibitory properties of the derivatives relative to that of
`rhizoxin. These changes included conversion of configuration at C(22}-C(23) from trans to cis,
`reduction of the C(20}-C(21) double bond, introduction of an ester linkage at C(22) and
`several derivatives in both cis and trans configuration at C(21). This work led Sawada et a/.
`(1991) to prepare fluorescent and photoaffinity analogs of rhizoxin derivatized at position
`C(22). Both analogs bound to tubulin and their binding was substantially reduced in the
`presence of rhizoxin. They may prove useful as probes for the mechanism of binding of the drug
`to tubulin.
`
`IMMUNOGEN 2001, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Interactions of tubulin in the vinca domain
`¥
`:,;c,
`CO-NH-C
`CHz
`H...,.
`7 'co
`ru_NJ.l_._._c/ 9
`110
`8
`61
`s I
`,.en
`'
`NH
`HO'
`
`/CH3
`
`.... '"]'.&&
`
`37
`
`Cl
`
`OH
`
`2S
`24
`23
`22
`21
`3 /C..........17 019 20
`2
`O-C
`CO-N-c-CO-NH-C-CO-NH-C-~H
`18
`I "CH3CHz
`II
`II
`CH3
`C
`C26
`/\
`/\
`H ~H
`CH3 C~~
`27
`
`H
`
`\
`
`FIG. 3. ~ 1ctural formula of phomopsin A.
`
`The crystal structure of one relatively inactive analog has been evaluated (Iwasaki eta/., 1986b).
`In the compound studied the side chain projects out from the 16-member ring.
`
`4. PHOMOPSIN A
`
`The fungus Phomopsis leptostomiformis grows on lupins and produces toxins which cause
`lupinosis, a severe liver disease of grazing animals. One of the pathologic changes observed in
`lupinosis is hepatocytes arrested in mitosis. Culvenor and coworkers (Culvenor eta/., 1977, 1983;
`Mackay et a/., 1986; Lacey et a/., 1987) isolated and characterized active components in fungal
`extracts. The predominant agent was named phomopsin A (NSC 381839; structure presented in
`Fig. 3). It is a hexapeptide formed from modified amino acids. Its structure includes a 13-member
`ring containing an ether linkage. Phomopsin A was used to produce experimental lupinosis,
`including the characteristic mitotic hepatocytes (Culvenor et a/., 1977). In addition, it caused
`mitotic arrest of cultured cells (Petterson eta/., 1979; Tonsing et al., 1984; Bai eta/., 1990b ), leading
`tothe prediction that it would interfere with microtubule assembly (Petterson et al., 1979).1t should
`be noted that phomopsin A in tissue culture studies has generally been used at micromolar
`concentrations (Petterson eta/., 1979; Tonsing eta/., 1984; Bai eta/., 1990b) and its IC50 value for
`the growth (at 24 hr) of Ll210 murine leukemia cells is 7 J.lM (Bai et a/., 1990b). This IC50 value
`is about 100-10,000-fold higher than that obtained with other natural product antimitotic agents
`(see below).
`Li eta/. (1992) recently reported the preparation of radiolabeled phomopsin A by fermentation
`with C4C)isoleucine. The peptide bound rapidly and stably to tubulin at 37°C and Scatchard
`analysis indicated two classes of binding site with Kd values of 10 and 300 nM. Rhizoxin and
`ansamitocin P-3 weakly inhibited the binding of the radiolabeled phomopsin A to tubulin, with
`IC50 values of about I 00 J.lM, while the inhibitory effect of vinblastine was still more feeble. Whether
`these inhibitory effects were competitive or noncompetitive was not determined. Although Li
`et a/. (1992) reported that they found no evidence for phomopsin A-induced aggregation of
`tubulin, other workers have described such aggregates (see below). This is of importance in
`evaluation of binding data. Ligand-induced aggregation of a protein can result in biphasic
`Scatchard plots even though there is actually only a single binding site for the ligand (Timasheff
`et al., 1991).
`Phomopsin A was initially reported to inhibit the binding of radio labeled vinblastine to tubulin
`more effectively than either maytansine or nonradiolabeled vinblastine, with an IC50 value of 0.8
`J.lM. The type of inhibition was not determined in this study (Lacey et a/., 1987). Subsequently,
`phomopsin A was demonstrated to be a classic noncompetitive inhibitor of the binding of
`radiolabeled vincristine to tubulin, with an apparent Ki value of 2.8 J.lM (Bai et a/., 1990a).
`Phomopsin A moderately enhances the binding ofradiolabeled colchicine to tubulin (Lacey eta/.,
`1987; Luduefia et a/., 1989; Bai et a/., 1990a). Moreover, it seems to completely stabilize tubulin.
`
`JPT SS/1-D
`
`IMMUNOGEN 2001, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`38
`
`E. HAMEL
`
`There is little or no decay of colchicine binding in its presence (Luduefia et al., 1989, 1990; Bai
`eta/., 1990a), nor any increase in the amount of binding of bis(8-ani1inonaphthalene-l-sulfonate)
`(Ludueiia et a/., 1989).
`Alkylation of tubulin sulfhydryl groups is altered by phomopsin A in a manner that resembles
`effects of both vinblastine and maytansine (Luduefia et al., 1990). Like vinblastine, phomopsin A
`strongly inhibits the reaction of tubulin with iodoacetamide and enhances formation of the first
`crosslink. Like maytansine, phomopsin A strongly inhibits formation of the second crosslink.
`Tubulin-dependent GTP hydrolysis is potently inhibited by phomopsin A and it is slightly more
`potent than maytansine (Bai et al., 1990b). The peptide is also a powerful inhibitor of nucleotide
`exchange on tubulin, although, like maytansine, phomopsin A does not displace GDP bound in
`the exchangeable site (Bai et al., 1990a). In fact, phomopsin A is the most potent inhibitor of
`nucleotide exchange yet examined (Bai et al., 1991). It is particularly notable in that exchange is
`inhibited almost as extensively at 37°C as at ooc. We have interpreted the loss of inhibition of
`nucleotide exchange by maytansine at 37°C as resulting from the reversibility of maytansine's
`binding to tubulin (Huang et al., 1985). If correct, this explanation implies that phomopsin A will
`dissociate very slowly, if at all, once bound to the protein. Initial studies with radiolabeled
`phomopsin A support this hypothesis (Li et al., 1992), as does the peptide's potent stabilization
`of tubulin (see above).
`In direct comparisons to other drugs phomopsin A was equivalent to vinblastine and more
`effective than maytansine as an inhibitor of microtubule assembly from microtubule protein (Lacey
`et al., 1987) and of the glutamate-induced polymerization of purified tubulin (Bai et al., 1990b).
`IC50 values of 0.6 J1M for inhibition of microtubule assembly (Tonsing et al., 1984; Lacey et al.,
`1987), of 0.4 J1M for inhibition of glycerol-induced polymerization of purified tubulin (Tonsing
`et al., 1984) and of 1.4 J1M for inhibition of glutamate-induced polymerization (Bai et al., 1990b)
`have been reported. Phomopsin A causes the disappearance of microtubules in cultured cells and
`the complete in vitro depolymerization of preformed microtubules (Tonsing et al., 1984).
`The disappearance of preformed microtubules was not associated with a complete loss of
`turbidity (Tonsing et al., 1984). Instead, the polymer was replaced by aggregates consisting of rings
`and spirals. These appeared to be much smaller and morphologically distinct (see below) from those
`induced by the vinca alkaloids (see Himes, 1991 and Fig. 4A). Rings and spirals were formed from
`microtubules derived from both microtubule protein and purified tubulin in glycerol (Tonsing et al.,
`1984). In our laboratory we have observed formation of morphologically similar oligomeric
`structures from unpolymerized tubulin in the presence of superstoichiometric concentrations of
`phomopsin A (Fig. 4B). These structures have been observed in reaction mixtures containing 0.1 M
`4-morpholine ethanesulfonate and heat-treated microtubule-associated proteins (Fig. 4B). In
`reaction mixtures containing only purified tubulin our initial studies indicate that rings, both single
`and in apparently aggregated clusters, are the predominant structure, with relatively few spirals.
`Substoichiometric concentrations of phomopsin A inhibit the formation of the vinblastine(cid:173)
`induced aggregate (4 J1M phomopsin A, 10 J1M tubulin, 40-150 J1M vinblastine), but the peptide is
`less potent than maytansine in this regard (Ludueiia et al., 1989).
`Lacey eta/. (1987) examined three analogs of phomopsin A for effects on microtubule assembly
`and vinblastine binding and they found little difference between the four compounds. The equally
`active agents were phomopsin B, which lacks the chloride atom at position C(l4), octahydropho(cid:173)
`mopsin A, which has the four carbon-carbon double bonds reduced, and phomopsinamine A,
`which lacks the substituent at position N(24).
`The crystal structure of phomopsin A (Mackay et al., 1986) has two molecules per unit, with
`the phenyl ring having somewhat different orientations. The side chain projects out from the
`13-member ring in both molecules.
`
`Pettit and his collaborators have been isolating cytotoxic agents produced by the marine animal
`Dolabella auricularia, a shell-less mollusc known as a sea hare (Pettit et al., 1987a,b, l989a,b,c,
`
`5. DOLASTATIN 10
`
`IMMUNOGEN 2001, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`-::l c> .,
`
`0
`::l
`
`~ :::.
`"' 0 ....,
`2
`0"
`E.. :;·
`:;·
`:;.
`0
`
`FIG. 4. Spiral and ring oligomers formed in the presence of superstoichiometric concentrations of vinblastine (panel A), phomopsin A (panel B) and dolastatin
`I 0 (panel C). The reaction mixture used for the micrograph presented in panel A contained 1.0 mg/ml tubulin, 0.5 mg/mL heat-treated microtubule-associated
`proteins, 0.1 M 4-morpholineethanesulfonate (pH 6.9 with NaOH), 0.5 mM MgC12, 0.1 M GTP and 50 f1M vinblastine and was incubated