Estramustine Depolymerizes Microtnbules by Binding to Tubulin 1

[CANCER RESEARCH 53, 4573-4581, October 1, 1993] Estramustine Depolymerizes Microtnbules by Binding to Tubulin 1

Bjfrn Dahll6f, 2 Anita Bilistr6m, Fernando Cabral, and Beryl Hartley-Asp Kabi Pharmacia Oncology, S-223 63 Lund, Sweden lB. D., A. B., B. HA.], and Department of Pharmacology, University of Texas Medical School, Houston, Texas 77225 [F. C.]

ABSTRACT Experiments using purified microtubule proteins from rat brain and from DU 145 human prostate cancer cells suggested that the antimi- To investigate the mechanism of action of the antineoplastic drug es- totic effects of estramustine were caused by binding of the drug to tramustine, we compared its effects on human prostate cancer cells with those of vinblastine. At their respective concentrations that result in 50% MAPs (9, 10). Such an interaction, revealed by binding of radiola- inhibition of clonogenic growth, both drugs caused an accumulation of beled EM to MAPs and by the ability of EM to strip MAPs from cells blocked at mitosis and similar dose- and time-dependent depolymer- taxol-stabilized microtubules in vitro, was believed to lead to disas- ization of interphase microtubules. Also, coicemid-resistant and colcemid- sembly of microtubules. However, it should be noted that these au- hypersensitive Chinese hamster ovary cells with tubulin mutations were thors also showed binding of EM to tubulin but interpreted this as collaterally cross-resistant or -sensitive to estramustine. Thus, the cyto- being nonspecific. Investigators (11) from another laboratory also toxicity of estramustine is due to its microtubule depolymerization prop- reported binding of EM to both tubulin and MAPs but could not detect erties. This could be caused by interaction with tubulin and/or with mi- any depolymerizing activity of EM in vitro. Thus, the actual target for crotubule-associated proteins (MAPs). Previous investigations have shown the antimicrotubule activity of EM has not yet been unequivocally that high concentrations of estramustine phosphate can inhibit microtubule polymerization in vitro by binding to MAPs. However, estramustine determined. phosphate is the clinical prodrug to estramustine, the intracellular active Curiously, it was found that the prodrug EMP also possessed anti- compound. In this study, we investigated the effects of estramustine on the microtubule activity, but only in microtubule polymerization experi- binding of MAPs to taxoi-stabilized microtubulesin vivo. In contrast to ments in vitro (12, 13). In experiments using cultured cells, EMP previous reports, no effect of estramustine on the binding of MAPs to either is inactive because it cannot penetrate the plasma membrane microtubules was found. Furthermore, we found that polymerization of (see "Results") or demonstrates activity identical with that of EM. The purified tubulin could be inhibited by estramustine in vitro. Taken to- latter result was found in glioma cells and was shown to be due to gether, these results demonstrate that estramnstine causes depolymeriza- dephosphorylation of EMP by cellular phosphatases (14). The micro- tion of microtubules by direct interaction with tubulin. tubule-depolymerizing activity of EMP in vitro was shown to be due to interactions between the negatively charged estramustine phosphate INTRODUCTION molecule with the positively charged tubulin-binding domains of EM 3 is an antineoplastic drug used in the treatment of advanced MAPs (15, 16). It is important to remember that this interaction is prostate cancer. The active compound consists of a 17/3-estradiol purely an in vitro phenomenon which is irrelevant for the intracellular group linked to nitrogen mustard via a carbamate bridge (Fig. 1). mode of action of EM. Unfortunately, in the vast majority of the Clinically, EM is given as the prodrug EMP, which is subsequently literature concerning EM, the proper distinction between the modes of dephosphorylated when administered both p.o. and i.v. (1). action of EM and EMP has not been made. EM was designed to function as an alkylating drug targeted to In order to study the mechanism of action of EM, we have com- estrogen receptor-bearing cells (e.g., breast cancer cells). There, the pared the cytotoxic and antimicrotubule effects of EM with those of carbamate bridge would be cleaved to release nor-nitrogen mustard. VLB, a well-characterized antimitotic drug with affinity for tubulin, However, the compound was found to be devoid of alkylating activity which recently has been combined with EM for the treatment of due to unanticipated stability of the carbamate bridge and lack of advanced prostate cancer (17, 18). Our results show that the two drugs alkylating activity of the intact molecule (2, 3). Also, the intact estra- are qualitatively similar with respect to cellular effects, which sug- mustine molecule did not show any significant binding to estrogen gests a common mode of action. This prompted us to carry out further receptors (3, 4). For a review of these properties of EM, see Ref. 5. experiments to challenge the hypothesis that EM exerts its action via Although EM did not exhibit the predicted properties for which the binding to MAPs. Evidence is presented which shows that EM, both drug was developed, the intact molecule had pronounced cytotoxicity in vivo (cell culture) and in vitro, causes microtubule disassembly by independent of any hormonal or alkylating activities (3, 5, 6). It was interaction with tubulin. shown that EM belongs to the group of antimitotic drugs, such as the Vinca alkaloids and colchicine, which can disrupt the microtubule MATERIALS AND METHODS network of cells in tissue culture and block cells at mitosis (7). Also, EM has recently been shown to block cells at mitosis in prostate tumor Cell Culture. The DU 145 cell line (purchased from American Type Cul- xenografts (8). ture Collection) was derived from a metastatic brain lesion of a human prostate adenocarcinoma and cultured as previously described (6). Growth conditions and characteristics of the CHO cell lines have previously been described (19). Received 3/8/93; accepted 7/19/93. Drugs. EM (estradiol-3-N-bis(2-chloroethyl)carbamate) and EMP were The costs of publication of this article were defrayed in part by the payment of page synthesized by Kabi Pharmacia (Lund, Sweden). VLB was purchased from charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Lilly, and taxol was from Sigma or as a gift to F. C. from Dr. Matthew Suffness 1 This work was funded in part by Grant CA52962 from the National Institutes of of the National Cancer Institute. For each experiment, fresh stock solutions Health to E C. were made in DMSO and diluted in medium to a final concentration of 0.1% 2 To whom requests for reprints should be addressed, at Departmentof Cell Biology, DMSO. Kabi Pharmacia Oncology,Scheelevagen 22, S-223 63 Lund, Sweden. 3 The abbreviations used are: EM, estramustine; BSA, bovine serum albumin; CHO, Antibodies. Mouse anti-/3-tubulin (N357) was purchased from Amersham; Chinese hamster ovary; DMSO, dimethylsulfoxide; EMP, estramustinephosphate; IC5o, alkaline phosphatase-conjugated rabbit anti-mouse antibody ($3721) was from concentration which results in 50% inhibitionof clonogenicgrowth; MAP, microtubule- Promega; fluorescein isothiocyanate-conjugated goat anti-mouse antibody associated protein; MTP, microtubuleproteins; PBS, phosphate-bufferedsaline; PC-tubu- (F479), and tetramethylrhodamine B isothiocyanate-conjugated swine anti- lin, phosphocellulose-purifiedtubulin; Pipes, 1,4-piperazinediethanesulfonicacid; SDS, sodium dodecylsulfate; PAGE, polyacrylamidegel electrophoresis;TBST, Tris-buffered rabbit antibody (R156) were from from Dako A/S, Denmark. Rabbit antiserum saline + Tween 20; VLB, vinblastine. recognizing CHO-MAPs was raised against a heat-stable protein (Mr 210,000) 4573

OR (4% SDS-30% sucrose-10 mM dithiothreitoi-100 mM Tris-HC1, pH 6.8-0.01% bromophenol blue) and then combined with cytoskeletal pellets suspended in

nor-nitrogen mustard 20 p,1 of water. To verify that the method resulted in identical amounts of protein in corresponding fractions from different wells, the cells were labeled CICH2CH2. ~ with [35S]methionine prior to treatment with drug, and protein in each fraction CICH2CH2~ N-C-O was quantitated by liquid scintillation counting (data not shown). Protein Electrophoresis and Immunodetection. The cytoskeletal frac- R = H; estrarnustine tions were taken through two rounds of freezing and thawing in order to reduce R = phosphate; estramustine phosphate the viscosity caused by high molecular weight DNA. An equal volume of both Fig. 1. Chemical structure of estramustine and estramustine phosphate. Nor-nitrogen fractions were mixed with the same volume of SDS-PAGE sample buffer and mustard has been linked via a carbamate bridge to estradiol (see "Introduction"). heated to 95~ for 5 min. The samples were alkylated by the addition of a 0.1 volume of 0.5 M iodoacetamide and incubated for 10 min at room temperature. SDS-PAGE was run in a Protean II xi cell (length, 20 cm; Bio-Rad) using a standard buffer system and 14% homogeneous gels, T = 0.37%, at 10 mA of 100 EM constant current overnight without cooling. Each lane contained 40 ~1 of sample. Rainbow molecular weight markers from Amersham (United King- 80 \ dom) were used for determination of molecular weights. The gels were blotted onto nitrocellulose filters using a semidry blotting device (Semidry electrob- > 'E. 60 lotter A; Ancos, Denmark) in 39 mM glycine, 48 mM Trizma-base, and 20% =,,,. methanol for 75 min at room temperature and 100 mA. The wet filters were 09 40 blocked in 5% fat-free dry milk in TBST for 30 min, incubated for 30 min with the anti-tubulin antibody diluted 1:500 in TBST, washed 3 times for 10 min in 20 TBST, incubated with alkaline phosphatase-conjugated secondary antibody diluted in TBST for 30 min, washed 3 times for 10 min in TBST, and finally developed in 100 mM Tris-HC1, pH 9.5-100 mM NaCI-5 mM MgCI2, using the i 10-9 lO-a lO-r 10-6 10-5 I 10-~ color reagents nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phos- Concentration (M) phate from Promega as described by the manufacturer. Indirect Immunofluorescence. Cells were grown on 8-well plastic cham- Fig. 2. Cytotoxicity of EM (&) and VLB (0) on DU 145 human prostate cancer cells ber slides (Permanox; Nunc, Denmark) for 2 days. After drug treatment, the measured as the number of clones with >50 cells 11 days after a 24-h exposure to drug. cells were fixed for 5 min in 100% methanol at -20~ followed by 5 rain in Points (bars), means (• of three experiments give IC5o values of approximately 3 nM for VLB and 16 ~M for EM. 100% acetone at -20~ The remaining steps were carried out at room temperature. Following fixation, the cells were washed for 10 s in PBS and 5 min in PBS/BSA (PBS with 0.2% BSA). Incubation with the primary antibodies (1:250 dilution) was carried out for 30 min and was followed by 3 washes for that copurifies with microtubules from CHO cells and is believed to be the 10 min in PBS/BSA. Incubation with secondary antibodies (fluorescein hamster MAP4 (20). isothiocyanate- and/or tetramethylrhodamine B isothiocyanate-conjugated an- Assays for Cytotoxicity. The cytotoxic effects of EM, EMP, and VLB on tibodies diluted 1:20) was for 30 min and was followed by 3 washes for 10 min survival of DU 145 cells were determined by using a colony-forming assay. in PBS/BSA. Coverslips were mounted using 1 mg/ml of p-phenylendiamine Logarithmically growing cells were suspended using 85 mM sodium citrate, in PBS with 80% glycerol. diluted, and seeded in 5-ml Petri dishes at a concentration of 210 cells/ml with Purification of Bovine Brain Tubulin. Microtubule proteins were pre- a plating efficiency of approximately 20%; 24 h later, the dishes were randomly pared from fresh bovine brain cortex in the absence of glycerol by two cycles divided into groups of 5 prior to treatment with drug for 24 h. Thereafter, the of polymerization-depolymerization in polymerization buffer (100 mM Pipes-1 cells were incubated for another 11 days with a change of medium at day 7, mM GTP-0.5 mM MgSO4-1 mM ethyleneglycoi bis03-aminoethyl ether)- fixed in 100% methanol, and stained with 1.0% methylene blue, and colonies composed of >50 cells were scored. Results are presented as percentages of surviving cells compared to nontreated cells. Drug resistance of CHO cells was Table 1 Effect of VLB and EM on the mitotic index of human prostate cancer DU 145 cells determined by seeding approximately 100 cells into each well of a 24-well dish. Cells in replicate wells were then exposed to various concentrations of the Concentration Mitotic index a drug for 5-7 days, and surviving colonies were stained with 0.5% methylene 1 nM VLB 1.0 • 0.4 blue as previously described (21). 4 nM VLB 9.2 • 1.3 Mitotic Index. DU 145 cells were suspended as described above and 4 ~M EM 1.7 • 0.4 16/.tM EM 9.8 • 1.2 diluted to 105 cells/ml. Culture flasks were inoculated with 5 ml of the cell a Percentage of mitotic cells in treated cultures divided by percentage of mitotic cells suspension and incubated for 2 days. Thereafter, cells were exposed to drug for in nontreated controls after 26 h of treatment (mean of three experiments ___ SEM). 26 h, and preparation for mitotic index analysis was carried out as previously described, except that colcemid was omitted (22). At each drug concentration, 4000 cells (for the control, 8000 cells) were counted. The results are given as P S P S P S P S P S P S P S percentages of mitotic cells in treated cultures divided by the percentages of mitotic cells in the control. Determination of Depolymerization in Vivo. DU 145 cells were grown to confluency in 12-well tissue culture plates, and duplicate wells were treated with drug for 3 h. Thereafter, wells were washed briefly in PBS without Ca 2+ l and Mg 2+ and lysed in 300 /~1 of microtubule-stabilizing buffer containing C 4 16 64 1 4 I 16 taxol (20 mi Tris-HCl, pH 6.8-140 mM NaC1-2 mM ethyleneglycol bis(/3- I aminoethyl ether)-N,N,N',N'-tetraacetic acid-1 mM MgC12-0.5% Nonidet estramustine nM vinblastine P-40-5 t~M taxol) as described earlier (19). The lysates were transferred to Fig. 3. Depolymerization of microtubules in DU 145 cells by EM and VLB. Cells were tubes, and the wells were washed once with 200 pJ of the same buffer and treated with drug for 3 h, and the cytoskeletal pellets (lanes P) were separated from pooled with the previous solution. Tubes were spun for 10 min in an Eppendorf cytosolic supernatants (lanes S), followed by SDS-PAGE, Western blotting, and immu- centrifuge at 14,000 rpm at 4~ and the supernatants, containing the unpoly- nodetection of /3-tubulin (see "Materials and Methods"). For each concentration, the staining intensity in lane P should be compared to that in lane S. In the concentration range merized tubulin, were transferred to separate tubes. Residual cytoskeletal ma- of 4-64 WMEM and 1-16 nM VLB, dose-dependent depolymerization of microtubules can terial in each well was dissolved in 480 p,1 of hot SDS-PAGE sample buffer be seen, which is correlated to the cytotoxicities of the two drugs (see Fig. 2). 4574

N,N,N',N'-tetraacetic acid, pH 6.8) (23). Tubulin was separated from MAPs by a suboptimal system containing 0.8 u sodium glutamate (pH 6.6) at 30~ in the ion-exchange chromatography on MgSO4-pretreated phosphocellulose in 20 absence of Mg 2+ (25, 26). mM Pipes-0.5 mM MgSO4 at pH 6.8 (24). Tubulin was eluted in the void RESULTS volume, and the buffer concentration was adjusted to 100 m~ Pipes-1 mM GTP-0.5 mM MgSO4, pH 6.8. Cytotoxicity and Mitotic Index. A clonogenic assay (DU 145) Assembly of Microtubules. Polymerization of microtubule proteins was used to compare the cytotoxicities of EM, EMP, and VLB. EMP (2 mg/ml) in vitro was performed in polymerization buffer containing 10% caused no cell death, and high-performance liquid chromatography glycerol. The glycerol had to be added in order to diminish EM binding to the analysis revealed that no dephosphorylation to EM, the active com- test tubes and pipettes. To further minimize adsorption, stock solutions and pound, had occurred (data not shown). EM had an ICso value of dilutions of EM were made in g/ass test tubes using glass pipettes. The reaction was initiated by increasing the temperature to 37~ and was monitored by approximately 16 p,M, while the ICso for VLB was approximately 3 nM apparent change in A3so. Tested drugs were dissolved in DMSO at 1000-fold (Fig. 2). As would be predicted for antimitotic drugs, both VLB and the final concentration. Thus, the final DMSO concentration was 0.1% in all EM caused mitotic arrest at concentrations that produced cytotoxicity experiments (including control). Phosphocellulose-purified tubulin (2 mg/ml) (Table 1). either (a) was polymerized in the same way except that the polymerization Depolymerization of Microtubules. At drug concentrations buffer contained an additional 8% DMSO (12) or (b) was polymerized using which represented the different parts of the cytotoxicity curves of EM

control control

EM: nm VLB:

Fig. 4. Indirect immunofluorcsccnccof microtubules in DU 145 cells. A and B. control cells. (, E, and (;, 3-h ttcatmcnt with 4. 1r ~tnd 04 p,v EM. respectively, l), K and H. 3-h Ircattllcnt x,.ilh I, 4, and 1(~ nv VLB. rc~,pccrvcl>. (omplctc disasscm- bh was perceived eli h4 ~txl EM '

64 16

4575

0 rain.

Fig. 5. Dcpol3mcrizationof nficn~tubulcsin DU 145 cells is rcversihlc'. Treatment v,'as for 1 h using 64 /J.M EM (A, (', and A) or 1() nv VI,B (B, D, and F) followed b\ I indirect immunofh.~rcscence at (1 (A and B). I()(12 and D). or b()rain (/f and F) after washotH of drug, As can be ",ceil ill l', ~tm.t /-, tlli- 10 min. crotubule nc'tx',orks x;crc rk'lbrn/cd within I h ['or both drugs. In .t, bar III ,uv.

60 min.

and VLB. cells \\crc harxcsted and the cytoskeletons were separated microtubulcs as revealed by immunocvtochemistrv. However, it from soluble components using a taxol-containing buffer (19). The shouh, i bc noted that concentrations of EM and VI,B ',,,ere compared fractions wcrc suhjcclcd to SDS-PAGE analvsis followed by Western which x~crc cquipotent in inhibiting the invasion of l)t! 145 cells into blotting and immunodctection of/3-tubulin to determine the rclali\c chick heart spheroids r'~thcr lhall cquiloxic. Usino indirect immuno- ratio of polvmerized tubulin [Fig. 3, h,w.s P (pellet) l rcr.sus tmpol>- fluorescence. \\c studied tile microtubule ncl\,~orks of I)U 145 cells at mcrized free tubulin dimers [lapw,sS (supcrnatant)l. In the control the same c~mccntratiot> :> those tined t~)r deternlillaliOll Of dcorcc of samples (Fig. 3, h,w.s C). the supernatant fraction shows apparently microtubulc dcpol)mcdzation. Again. ,ac found similar dose-depen- slightly higher staining intensity than the pellet fraction. For cells dent effects on the microtubulcs (Fig. 4). At 4/a.~, [!M and 1 nxt VI,B. treated with 4, l(x and 64/,tv EM, a comparison of each h,w P with the naosl peripheral microtubulcs seemed to bc retracted, and the the corresponding/am' A indicates a dosc-depcndcnt depolymerization ccnlral ncl;vorks shmved a slighlly 'criss-crossed' pauern. AI 16/z~t of the microtubule network. The same rcsull \,,as obtained using 1.4. EM and 4 nx~ VLB. most microtubules were concentrated in the and 16 nv VLB. Although exact measurements of the degree of pcrinuclear regions, ~nd at (~4 /,tx~ EM and 1(~ nv VLB, Ihc microtu- depolymerization at each drug concentration have not been madc bulc nctx,~orks xvcrc apparently completel.v depolymerized. Blebs in (because of technical difficulties), wc conclude that both EM and VLB the plasma membranes filled with lubulin were frequently found cause microlubulc depolymerization at cvtotoxic concentrations. This which is a common cffcc{ of antimicrolubule drugs. All immunofluo- indicates a similar mode of action. rescence experiments were anahzcd by two investigators without Indirect Immunotluorescence. In a previous investigation (27), it knowledge of the conditions involved, and we repeatedly failed to was found that EM and VLB caused qualitatively different patterns of determine whcthcr cells had bcen treated with EM or VLB. Thus, in 4576

Cmd 0 0.005 0.01 0.02 0.03 0.04 showed that the three cell lines responded to challenge to EM in a manner similar to their response to colcemid, a microtubule-depoly- WI" merizing drug (Fig. 6). Since mutations in tubulin genes affect sensitivity to EM, we conclude that the dominant contribution to the cytotoxicity of EM is its microtubule-depolymerizing activity. Cmd4 EM Treatment of Taxol-stabilized Microtubules in Vivo. The results presented thus far indicate that EM, although being less potent, does not differ significantly from other microtubule-depolymerizing Tax 5-6 drugs in its effect on cells in culture. This outcome is compatible with either a tubulin-binding activity, similar to other microtubule drugs, or with binding to MAPs. For the latter possibility, the assumption has to be made that breaking MAP-tubulin interactions in vivo results in EST 0 0.5 1 2 4 6 depolymerization of the microtubule network (the physiological role of the different MAP proteins is not yet fully understood). Previous Wl" ~" ~: ...... experiments from others showed that EM had affinity for MAPs in vitro and that EM could dissociate MAPs from microtubules presta-

...... bilized with taxol in vitro and in vivo (9, 10). Therefore, using a similar approach, we tested whether EM could affect the interaction between MAPs and tubulin in vivo in CHO cells. J CHO cells were treated for 1 h with 1 ~M taxol to stabilize the microtubule network and then with taxol plus 64 /.ZM EM for another 3 h. The cells were fixed and stained for indirect immunofluorescence .... B using antibodies to both tubulin and an Mr 210,000 heat-stable CHO Fig. 6. Cross-resistance of CHO tubulin mutants to estramustine. Wild-type ~,FF) CHO cells, colcemid (Cmd 4), a colcemid-resistant mutant with altered/3-tubulin, and Tax 5--6, MAP. The results were compared to those from untreated cells and a taxol-resistant mutant with altered a-tubulin, were seeded into individual wells of from ceils treated with either taxol or EM alone for the same time (Fig. 24-well dishes as described in "Materials and Methods" and exposed to varying concen- 7). Occasionally, significantly larger cells were found (Fig. 7, A, B, E, trations of colcemid (Cmd;A) or estramustine (EST; B). After 5 d, survivingcolonies were stained with methylene blue and photographed. In A, concentrations of colcemid are and F), and these were selected for photography, when possible, indicated above each well in/xg/ml; in B, concentrationsof estramustine are indicated in because of their extensive microtubule networks. However, the results p.u. Note that Cmd 4 is resistant to both colcemid and estramustine, whileTax 5-6 is more sensitive to both agents compared to wild-type cells. of the experiment were independent of the cell size. In control cells, the MAP antibody decorated all the microtubules recognized by the tubulin antibody. When cells were treated with 64 /z~ EM, significant depolymerization of the microtubule network was DU 145 cells, EM and VLB affected the microtubule network in a seen. However, some residual microtubules persisted, and these ap- qualitatively similar manner that correlated to the degree of microtu- peared to be preferentially stained by the MAP antisera. bule depolymerization and cytotoxicity. Prestabilization of cellular microtubules by taxol permitted us to Time-course experiments using 64/xM EM and 16 nM VLB revealed ask whether EM, at concentrations sufficient to cause depolymeriza- that the microtubule networks were affected as early as 5-10 min after tion in the absence of taxol, would change the MAP decoration of treatment. Maximal effects were seen at these concentrations after 1 h microtubules. As can be seen in the two lower panels in Fig. 7 of treatment with both drugs (data not shown). (compare E and F with G and H), the addition of EM to taxol- Reversible Effects. The modes of action of EM and VLB were stabilized microtubules did not release MAPs from the microtubule further compared by studying the reversibility of rnicrotubule depo- network. This indicates that EM does not cause microtubule depoly- lymerization. DU 145 cells were treated with 64/xM EM or 16 nM VLB merization in vivo by binding to MAPs. Thus, our results contrast with for 1 h, concentrations sufficient to cause significant and apparently those of others regarding the effects of EM on taxol-stabilized micro- similar effects on the microtubule networks (Fig. 5). After washout, tubules (9, 10). microtubule networks began to reform within 10 min for both drugs. In Vitro Polymerization of Microtubules. To further test whether After 1 h, background staining due to unpolymerized tubulin dimers MAPs mediate the ability of EM to produce microtubule depolymer- was greatly reduced, and microtubules extended to the cell periphery. ization, the effect of EM on assembly of microtubules in vitro was Thus, neither EM nor VLB caused irreversible depolymerization of explored. MTP (tubulin plus MAPs) and PC-tubulin (free from MAPs) microtubules at the concentrations used. were purified from bovine brain by standard procedures (see "Mate- Mutations in Tubulin Genes Result in Resistance to EM. Sev- rials and Methods"). The purity of each fraction was verified by eral CHO cell mutants have been isolated with mutations in tubulin SDS-PAGE analysis. With this method, MAPs were not detected in genes and an altered sensitivity to microtubule drugs (19, 21, 28). For the PC-tubulin preparation (Fig. 8A, lane b). example, cells with a mutation resulting in more stable microtubules We first tried to reproduce the results of others (9, 10) i.e., EM was are less sensitive to depolymerizing drugs such as colcemid and VLB found to detach MAPs from in vitro polymerized microtubules stabi- but more sensitive to taxol compared to wild-type cells. In contrast, lized with taxol. However, SDS-PAGE analysis of precipitated micro- mutants with less stable microtubules are more sensitive to depoly- tubule pellets showed no such effects on the MAP content using merizing drugs and less sensitive to taxol. For a further discussion of similar conditions (data not shown). In the absence of taxol, EM could the implication and utility of these mutants, see the report by Cabral partially inhibit microtubule polymerization in a dose-dependent man- and Barlow (28). ner (data not shown). Since the latter result could be due to binding of In this study, we used mutant cells to assay whether mutations in EM to tubulin or MAPs, next we explored the effects of EM on tubulin genes could also affect sensitivity to EM. Wild-type cells, polymerization of PC-tubulin using similar conditions for polymer- Cmd4, a colcemid-resistant mutant, and Tax5--6, a taxol-resistant mu- ization except that 8% DMSO was added to promote assembly in the tant, were tested for their sensitivities to colcemid and EM. The results absence of MAPs (12). Importantly, we found that EM could partially 4577

anti-tubulin anti-MAPs

control

EM Fi<.z,. 7. I)oul>lr i ). IlCLillllCnl ',.', ith f~4 t(',l t{\I It~l .~ h c. HIl'--c'<, IlliCr<>IUt~ulc cli,,:i<,,,cmhl,, ~l', I,',,c.'c~t<.'ct h', (ul~tllil/ -I,cu~i)/m (( I. t](w,c',cr, hc')illC mic'i~v, ubulc.',, I'c'lll;ti;] l I i' t,+ -t,+l+ili/~: thin I/llCil)Itll*itilu [lC{\t.t~ii~.. ]ollo\xl di,,,,<>ci,tc \1\1',. Ilcq Li,,,Q-,,Lihilizcd mic'IOltl bulL:,, i/l /7/<~. lu 1. I,,? II 7(,.1 taxol

taxol + EM

inhibit tile l~,i ', )~_r)/,~ti(m ~)I l)(-Itlbtilin in ~i clcr-,c-cJcl,cnclcilt [l'b;illi~L') )cdt~c!i<)n in Ihc ,.'x1~'~l ,>I F),)J','mcriz~Ik)ll m vm~ ushl~ I H-',I VI.B :mcl ~.t q L'()IL'CI]Iid (~.Iill,l 11 I '~]II~\\ II). \\ hcI-Clm lhc C(,)IICCIIIT'Alit')II nCCdL'CI It)

led/iCe l)(~[>mCIi/',iil~l ;LI){Ul #,()<~ //l I'11'~) ',',',IS 'dI')131"(.')XilllLI'tL'l} ~- I1\1 prccipil'aic~ u: c, mccntr,ition~ 4() i(\,, ,\lih~)u~h the c',Icnl (~I inhi (1 EL',. 3 ;lnd 4)~,ml !,:, n\l !m>i ',l~(~x\ n). rc,,pccli', L'I\'. Fhu>,. comp;ncd i~ilion ai)l~c_"r cx~lmt)lc. ~c <>l,tciincd Cll~prc)xim'

A abc 0,5 ~ ~ -MAP1 0,4 C 0 C \MAP2 = 0,3 12

o 5 ffl 0,2 '~ 0,1 ~-tubuJin.... } tau Fi~, 8. Purification and polymerization of mi- -tubulin -0,1 crotubules in vitro. MTP and PC-tubulin were pre- 0 5 10 15 pared from bovine brains as described in "Materials minutes and Methods." A, Coomassie Brilliant Blue-stained SDS polyacrylamide gel of (a) MTP, (b) PC-tubulin, and (c) MAPs. The identities of the bands are 0,5 i 9 i ii i tentatively indicated. No staining of high M~ MAPs can be found in the PC-tubulin fraction. Detection 0,4 D of tau in lanes a and b is obscured by the high amounts of tubulin in the same region of the get. B, 0,3 ~0 polymerization of PC-tubulin (2 mg/ml) in 10 and 40 ~M EM compared to control. C, polymerization 0,2 of PC-tubulin at suboptimal conditions in 1, 2, and 5 /XM EM compared to the control. In this system, 5 ~M EM results in a similar relative degree of 0,1 inhibition compared to the control as 40/xM EM did in the standard assay (B). D, PC-tubulin incubated 0 at suboptimal conditions in the presence of 40/xM EM compared to the control. E, the material in D -0,1 - | .... it .... incubated for another 15 rain on ice and the absor- 0 5 10 bance measured at 5, 10, and 15 min following start minutes of cold treatment. Thus, the structures induced by high concentrations of EM are qualitatively differ~ ent compared to the control sample. 0,5 0,5 C E 0,4 B 10 0,4 8 0,3 40 0,3

o 0,2 0,2 40 0,1 0,1

0 0 ...... C -0,1 -0,1 ! ,,I 0 5 10 15 0 5 10 15 minutes minutes

By altering the temperature, buffer composition, and concentration polymerization conditions, induce the formation of cold-stable struc- of Mg 2+, others have shown increased activity of weak tubulin bind- tures similarly to what is found using other synthetic estrogens. ers in such a suboptimal system compared to the activity in the standard polymerization assay (25, 26). Therefore, we polymerized DISCUSSION PC-tubulin in a similar manner and found dose-dependent inhibition of microtubule polymerization at considerably lower EM concentra- In this paper we compared the cellular effects of EM with those of tions (1-5 /XM EM; Fig. 8C). However, at higher EM concentrations VLB and described the following results: (a) both drugs induced (10-40 /XM), the microtubule polymerization (as detected by an in- mitotic arrest at their respective ICso concentrations; (b) at compa- crease of apparent absorbance) was not totally abolished. Rather, the rable cytotoxic concentrations, both drugs caused indistinguishable shapes of the polymerization curves were gradually changed. This and reversible depolymerization of microtubules, as detected by cyto- effect was most pronounced at 40/XM EM (Fig. 8D). Several reports skeleton extraction and indirect immunofluorescence; (c) a panel of have previously shown that estrogenic substances, which inhibit bind- CHO cell lines with tubulin mutations exhibited similar patterns of ing of radiolabeled colchicine to tubulin, can induce formation of resistance or hypersensitivity to EM and to colcemid, a well-charac- cold-stable tubulin structures (for a review, see Ref. 29). Therefore, terized microtubule inhibitor that binds to tubulin; (d) EM was unable we examined the cold stability of control microtubules and the ma- to displace MAPs from taxol-stabilized microtubules in vitro and in terial obtained using 40/x~ EM. As can be seen in Fig. 8E, control vivo; and (e) EM was shown to inhibit in vitro polymerization of microtubules were completely depolymerized already following 5 rain purified tubulin in the absence of MAPs. at 4~ whereas a substantial absorbance was recorded for the sample For several years, EM has been known in the microtubule field for containing EM even after 15 min of cold treatment. Thus, we conclude its unique mode of action. While all other microtubule drugs interact that EM can efficiently inhibit polymerization of pure tubulin in the with sites on tubulin molecules, EM was thought to bind to the MAPs absence of MAPs and that high concentrations, using suboptimal and thereby disrupt microtubule assembly (9, 10, 30). In this context, 4579

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 1993 American Association for Cancer Research. MODE OF ACTION OF ESTRAMUSTINE it is important to remember the different properties of EM, which is ACKNOWLEDGMENTS the intracellular active metabolite, and the prodrug EMP. For EMP, We would like to thank Christina Ekstr6m, Anette Gummesson, Ingrid work from another laboratory clearly showed that high micromolar Nordh, and Mats A.kesson for excellent technical assistance. Furthermore, we concentrations inhibited polymerization in vitro by binding to MAPs are grateful to Dr. Margareta Wallin and her collaborators for introducing us to (12, 15, 16), while EM failed to inhibit microtubule polymerization preparing bovine brain tubulin and to Robert Kristofferson and Jan-Otto Holm (11). Thus, while EMP was shown to have a MAP-mediated mode of for their darkroom expertise. Lennart Persson at SKANEK (K~ivlinge) is ac- action in vitro, these papers did not reveal the intracellular mode of knowledged for providing bovine brains. We wish to thank Dr. Matthew action of EM. Subsequently, others showed that EM also could dis- Suffness of the National Institutes of Health for the generous gift of taxol. sociate MAPs from taxol-stabilized microtubules in vitro, and this was taken as strong evidence for a MAP-mediated mode of action of EM REFERENCES in vivo (9, 10). However, the results presented in this paper demon- 1. Gunnarsson, P. O., Andersson, S-B., Johansson, S-/~., Nilsson, T., and Plym-Forshell, strate that EM does not displace MAPs from either (a) taxol-stabilized G. Pharmacokinetics of estramustine phosphate (Estracyt) in prostatic cancer patients. microtubules in vitro or (b) microtubules in living cells. These results Eur. J. Clin. Pharmacol., 26: 113-119, 1984. 2. Kruse, E., and Hartley, A. B. The effect of estramustine, nor-nitrogen mustard and question the significance of the earlier observations on EM. Further- tauromustine on macromolecular labelling in the human prostatic tumour cell line more, our in vitro polymerization experiments demonstrate a direct 1013L. PharmacoI. Toxicol., 64: 9-13, 1989. functional inhibition by EM of microtubule polymerization via tubulin 3. Tew, K. D., Eriksson, L. C., White, G., Wang, A. L., Schein, P. S., and Hartley-Asp, B. Cytotoxicity of estramustine, a steroid-nitrogen mustard derivative through non- interactions. Thus, we conclude that EM (but not EMP) acts via a DNA targets. Mol. Pharmacol., 24: 324-328, 1983. tubulin-mediated mechanism. It is of interest to note that this is not the 4. Leclercq, G., Heuson, J-C., and Deboel, M-C. Estrogen receptor interaction with Estracyt and degradation products, a biochemical study on a potential agent in the first time an antimitotic agent was first considered to interact via treatment of breast cancer. Eur. J. Drug Metab. Pharmacokinet., 2: 77-84, 1976. MAP-binding, as was the case for griseofulvin (31), but later shown 5. Tew, K. D. The mechanism of action of estramustine. Semin. Oncol., 10 (Suppl 3): to interact with tubulin (32, 33). 6. 21-26, 1983. Hartley-Asp, B., and Gunnarsson, P. O. Growth and cell survival following treatment Synthetic estrogens (e.g., diethylstilbestrol) as well as estradiol are with estramustine, nor-nitrogen mustard, estradiol and testosterone of a human pros- known to have antimicrotubule as well as antimitotic activities (34- 7. tatic cancer cell line (DU 145). J. Urol., 127: 818--822, 1982. Hartley-Asp, B. Estramustine-induced mitotic arrest in two human prostatic carci- 36). Diethylstilbestrol has been shown by two groups to compete with noma cell lines DU 145 and PC-3. Prostate, 5: 93-100, 1984. colchicine for binding to tubulin (37, 38). Therefore, it is possible that 8. Eld6v, S., Nilsson, S., Larson, A., Bj6rk, P., and Hartley-Asp, B. Evidence for a the functional and structural similarities among diethylstilbestrol, es- non-estrogenic cytostatic effect of estramustine on human prostatic carcinoma cells in vivo. Prostate, 20: 43-50, 1992. tradiol, and EM provide the basis for an interaction with the same 9. Stearns, M., and Tew, K. D. Estramustine binds MAP-2 to inhibit microtubule as- tubulin site. Our results showing induction by EM of cold-stable sembly in vitro. J. Cell Sci., 89: 331-342, 1988. 10. Steams, M., Wang, M., Tew, K. D., and Binder, L. I. Estramustine binds a MAP- l-like tubulin structures indicate other possible common features of EM and protein to inhibit microtubule assembly in vitro and disrupt microtubule organization synthetic estrogens. Furthermore, analysis of the molecular confor- in DU 145 cells. J. Cell Biol., 1988: 2647-2656, 1988. mation of EM has indicated similarities to other drugs known to bind 11. Ffid6n, B., Rutberg, M., Deinum, J., and Wallin, M. The effect of estramustine derivatives on microtubule assembly in vitro depends on the charge of the substituent. to the colchicine site (39). Binding studies are now under way in order Biochem. Pharmacol., 42: 997-1006, 1991. to determine more precisely the nature of the EM-tubulin interaction. 12. Frid6n, B., Wallin, M., Deinum, J., Prasad, V., and Luduena, R. Effect of estramustine phosphate on the assembly of trypsin-treated microtubules reconstituted from purified In a recent review, it was suggested that EM would preferentially tubulin with either tan, MAP2, or the tubulin-binding fragment of MAP2. Arch. affect the mitotic spindle (30). We found that EM and VLB affect both Biochem. Biophys., 257: 1987. interphase and spindle microtubules (measured as mitotic arrest) in the 13. Kanje, M., Deinum, J., Wallin, M., Ekstr6m, P., Edstr6m, A., and Hartley-Asp, B. Effect of estramustine phosphate on the assembly of isolated bovine brain microtu- same concentration range. Of course, the consequences of depolymer- bules and fast axonal transport in the frog sciatic nerve. Cancer Res., 45: 2234-2239, ization of the spindle microtubules are much more devastating for the 1985. 14. von Schoultz, E., Gunnarsson, P. O., and Henriksson, R. Uptake, metabolism and dividing cell than depolymerization of interphase microtubules, but antiproliferative effect of estramustine phosphate in human glioma cell lines. Anti- this is true for all microtubule drugs. cancer Res., 9: 1713-1716, 1989. Recently, clinical trials have indicated that a combination of EM 15. Wallin, M., Deinum, J., and Frid6n, B. Interaction of estramustine phosphate with microtubule-associated proteins. FEBS Lett., 179: 289-293, 1985. and VLB is beneficial for patients with advanced prostate cancer (17, 16. Moraga, D., Rivasberrios, A., Farias, G., WaUin, M., and Maccioni, R. B. Estramus- 18). Part of the rationale for combining these two drugs was their tine-phosphate binds to a tubulin binding domain on microtubule-associated proteins MAP-2 and tau. Biochim. Biophys. Acta, 1121: 97-103, 1992. distinct molecular targets. The other reason for this combination was 17. Seidman, A. D., Scher, H. I., Petrylak, D., Dershaw, D. D., and Curley, T. Estramus- their different toxicity profiles. Based on similar arguments, it was fine and vinblastine--use of prostate specific antigen as a clinical trial end point for also suggested that the combination of taxol and EM would be a 18. hormone refractory prostatic cancer. J. Urol., 147: 931-934, 1992. Hudes, G. R., Greenberg, R., Krigel, R. L., Fox, S., Scher, R., Litwin, S., Watts, P., logical combination therapy approach (30). Since our study shows that Speicher, L., Tew, K., and Corals, R. Phase-II study of estramustine and vinblastine, EM is a tubulin-binding drug, it does not, based on mechanistic two microtubule inhibitors, in hormone-refractory prostate cancer. J. Clin. Oncol., 10: 1754-1761, 1992. arguments, support the combinations of EM with other microtubule 19. Minotti, A. M., Badow, S. B., and Cabral, E Resistance to antimitotic drugs in drugs. However, the combination of EM with drugs such as VLB may Chinese hamster ovary cells correlates with changes in the level of polymerized still be clinically beneficial. This may be the result of additive anti- 20. tubulin. J. Biol. Chem., 266: 3987-3994, 1991. Brady, R. C., and Cabral, E Identification of a 210K microtubule associated protein microtubule effects without similar enhancement of the side effects. in Chinese hamster ovary cells (abstract). J. Cell Biol., 101: 30a, 1985. The dose-limiting toxicity of VLB is bone marrow depression, 21. Cabral, E, Sobel, M. E., and Gottesmann, M. M. CHO mutants resistant to colchicine, colcemid and griseofulvin have an altered/3-tubulin. Cell, 20: 29-36, 1980. whereas cardiovascular effects are the dose-limiting complications of 22. Hartley-Asp, B., Billstr6m, A., and Kruse, E. Identification by C-banding of two EM. Thus, higher concentrations of drugs with antimitotic properties, prostate tumour cell lines, 1013L and DU 145. Int. J. Cancer, 44: 161-164, 1989. resulting in better response rates, may be obtained by giving them in 23. Wallin, M., Frid6n, B., and Billger, M. Studies of the interaction of chemicals with microtubule assembly in vitro can be used as an assay for detection of cytotoxic combination rather than by giving a single drug. chemicals and possible inducers of aneuploidy. Mutat. Res., 201: 303-311, 1988. In conclusion, we have investigated the mechanism of action of EM 24. Williams, R. C., and Detrich, H. W., III. Separation of tubulin from microtubule- associated proteins on phosphocellulose. Accompanying alterations in concentrations and compared its effect on human prostate cancer cells in culture with of buffer components. Biochemistry, 18: 2499-2503, 1979. those of VLB. We conclude that both drugs cause mitotic arrest and 25. Paull, K. D., Lin, C. M., Malspeis, L., and Hamel, E. Identification of novel antimi- microtubule depolymerization at cytotoxic concentrations and that totic agents acting at the tubulin level by computer-assisted evaluation of differential cytotoxicity data. Cancer Res., 52: 3892-3900, 1992. EM, like VLB, acts by an interaction with tubulin. 26. Getahun, Z., Jurd, L., Chu, P. S., Lin, C. M., and Hamel, E. Synthesis of alkoxy- 4580

substituted diaryl compounds and separation with inhibition of tubulin polymeriza- 33. Wehland, J., Herzog, W., and Weber, K. Interaction of griseofulvin with microtubules, tion: enhancement of inhibitory effects under suboptimal reaction conditions. J. Med. microtubule protein and tubulin. J. Mol. Biol., 111: 329-342, 1977. Chem., 35: 1058--1067, 1992. 34. Hartley-Asp, B., Deinum, J., and Wallin, M. Diethylstilbestrol induces metaphase 27. Mareel, M. M., Storme, G. A., Dragonetti, C. H., De Bruyne, G, K., Hartley-Asp, B., arrest and inhibits microtubule assembly. Mutat. Res., 143: 231-235, 1985. Segers, J. L., and Rabaey, M. L. Antiinvasive activity of estramustine on malignant 35. Rao, P. N., and Engelberg, J. Structural specificity of estrogens in the induction of MO4 mouse cells and on DU-145 human prostate carcinoma cells in vitro. Cancer mitotic chromatid non-disjunction in HeLa cells. Exp. Cell Res., 48: 71-81, 1967. Res., 48: 1842-1849, 1988. 36. Sato, Y., Sakakibara, Y., Oda, T., Aizuyokota, E., and Ichinoseki, K. Effect of estradiol 28. Cabral, F., and Barlow, S. B. Resistance to antimitotic agents as genetic probes of and ethynylestradiol on microtubule distribution in Chinese hamster V79 cells. Chem. microtubule structure and function. Pharmacol. Ther., 52: 159-171, 1991. Pharmaceut. Bull., 40: 182-184, 1992. 29. Correia, J. J. Effects of antimitotic agents on tubulin-nucleotide interactions. Phar- 37. Sharp, D. C., and Parry, J. M. Diethylstilbestrol: the binding and effects of diethyl- macol. Ther., 52: 127-147, 1992. stilbestrol upon polymerization and depolymerization of purified microtubule proteins 30. Speicher, L. A., Barone, L., and Tew, K. D. Combined antimicrotubule activity of in vitro. Carcinogenesis (Lond.), 6: 865-871, 1985. estramustine and taxol in human prostatic carcinoma cell lines. Cancer Res., 52: 38. Chaudoreille, M. M., Peyrot, V., Braguer, D., Codaccioni, E, and Crevat, A. Quali- 4433--4440, 1992. tative study of the interaction mechanism of estrogenic drugs with tubulin. Biochem. 31. Roobol, A., Gull, K., and Pogson, C. I. Inhibition by griseofulvin of microtubule Pharmacoi., 41: 685-693, 1991. assembly in vitro. FEBS Lett., 67: 248-251, 1976. 32. Sloboda, R. D., Van Blaricom, G., Creasey, W. A., Rosenbaum, J. L., and Malawista, 39. Punzi, J. S., Duax, W. L., Strong, P., Griffin, J. E, Flocco, M. M., Zacharias, D. E., S. E. Griseofulvin: association with tubulin and inhibition of in vitro microtubule Carrell, H. L., Tew, K. D., and Glusker, J. P. Molecular conformation of estramustine assembly. Biochem. Biophys. Res. Commun., 105: 882---888, 1982. and two analogues. Mol. Pharmacol., 41: 569-576, 1992.

4581

Björn Dahllöf, Anita Billström, Fernando Cabral, et al.

Cancer Res 1993;53:4573-4581.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/53/19/4573

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/53/19/4573. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.