Biochimica et Biophysica Acta 968 (1988) 9-16 9 Elsevier

BBA 12175

H o w d o e s d o x o r u b i e i n i n t e r f e r e wi th ae t in p o l y m e r i z a t i o n ?

R o b e r t o C o l o m b o a n d A l d o M i l z a n i

Zoology and Cytology Laboratory, Department of Biology, University of Milan, Milan (Italy)

(Received 15 July 1987) (Revised manuscript received 17 September 1987)

Key words: Doxorubicin; Anthracycline; Actin polymerization; Polymer size

It is well known that doxorubicin (adriamycin), an antibiotic with an antitumoral action, has some undesirable side effects. Among these, the most serious is, undoubtedly, damage to myocardial tissue (progressive cardiomyopathy). We have for some time focused our attention on the effect of this drug on cellular contractile systems and, more specifically, on the process of actin polymerization, which we consider to be an extremely delicate key point for the economy of most cellular motor manifestations. In the present study, using capillary viscometry, spectrofluorometry and electron microscopy, we have shown a negative action of doxorubiein on various important chemical events which contribute to the transformation of G-aetin into F-aetin. Specifically, we found that the drug mainly acts by reducing the polymer size. A possible action mechanism of the antibiotic is proposed and a plausible correlation among the events described in vitro and those observed in vivo is advanced.

Introduction

Doxorubicin (adriamycin), an anthracycline an- tibiotic, is one of the most effective drugs availa- ble for cancer chemotherapy. In spite of its wide spectrum of activity [1-4], the clinical use of doxorubicin has been greatly limited because of severe toxic side effects. Among them, progressive cardiomyopathy is undoubtedly the most char- acteristic and injurious for the patient [5-9]. After acute or chronic treatment with doxorubicin, there is unequivocal cardiopathological symptomatol- ogy, characterized by an increase in serum levels of lactate dehydrogenase and phosphocreatinine [10-13]. Chronic treatment also provokes an in- creased uptake of Ca z+ by cardiac tissue and mitochondria [14-16].

Correspondence: R. Colombo, Zoology and Cytology Labora- tory, Department of Biology, University of Milan, Via Celoria 26, 20133 Milan, Italy.

Since cardiac muscle tissue is composed of cells characterized by their contractile function, it is likely that any interference at the level of the cardiac myofibrillar apparatus, in its totality or of a single protein component, will alter the entire functional sphere of the single myocell and conse- quently of the entire myocardium. This simple observation is, apparently, not so obvious, consid- ering the small number of studies in which, more or less directly, a possible injurious action of doxorubicin on the contractile systems or on their individual protein components has been evaluated. Necco and Ferraguti [17] studied the effect of the drug on in vitro myoblast fusion and found that doxorubicin determined severe alterations (disap- pearance of the Z-line and disorganization of myofilaments) in the myofibrillar architecture of these cells. Someya et al. [18] demonstrated a direct chemical interaction among doxorubicin, actin and heavy meromyosin. Our previously re- ported results [19,20] suggested that doxorubicin

0167-4889/88/$03.50 © 1988 Elsevier Science Publishers B.V. (Biomedical Division)

10

can influence the steady state of the system of chemical events which leads to the formation of filamentous actin, thus favoring accumulation of the monomeric form (G-actin).

In the present study, it was our intention to demonstrate in detail and to quantify the negative influence of doxorubicin on actin polymerization. It is known that most actin-binding proteins inter- act with actin (and, therefore, perform their func- tional task) only when the protein is in the poly- meric form. It is, thus, easy to deduce that any perturbation in the biochemical pattern which, starting with globular actin, leads to the formation of the polymer, can make its action fel't on cellular motor manifestations. Such effects, when attribu- table to antitumoral drugs such as doxorubicin, inspire considerations that are often contrasting. Although inhibition of cellular motility in anti- tumoral therapy may have favorable implications (inhibition or slowing of metastasis), it can, parallely and by means of analogous mechanisms, negatively influence the entire physiology of organs essential for a correct functioning of the individ- ual (cardiotoxic effect).

Materials and Methods

Preparation of actin. Globular actin (M~ 43 000) was prepared from muscles of the posterior limb of rabbits, according to the method of Spudich and Watt [21] and modified by McLean-Fletcher and Pollard [22]. The protein thus obtained was further purified by gel filtration on a G-150 Seph- adex column (2.6 × 50 cm) [22]. The purity of the protein preparations was periodically tested by SDS-polyacrylamide gel electrophoresis. The pro- tein concentration was assayed using the method of Bradford [23] or by determination of the ab- sorbance at 290 nm, considering [24]:

A29 o ( C = 1 m g / m l ; p a t h l e n g t h = 1 cm) = 0 .63

(See Ref 24.) At the end of the chromatographic procedure, the protein was in the monomeric form (G-actin) dissolved in G buffer (2 mM Tris, 0.2 mM ATP, 0.5 mM dithiothreitol, 0.2 mM CaC12, 1.5 mM NAN3) at pH 8.

Fluorescent labeling of actin. Fluorescent actin was obtained by coupling the protein with N-(1- pyrenyl)iodoacetamide (pyrene) using the method

of Tellam and Frieden [25], with the following modifications: (a) actin was polymerized by ad- dition of 100 mM KC1 and 1 mM MgC12; (b) the pyrene/actin molar ratio in the reaction mixture was 2:1. The yield of the conjugation reaction (pyrene/actin) was calculated according to the indications of Cooper et al. [26] and of Kouyama and Mihashi [27]:

m M act in = (A29 o - 0 . 1 2 7 - A 3 , ~ ) / 2 6 . 6 m M - ~ . c m 1

[py rene ] = A 3 4 4 / 2 . 2 . 1 0 4 M - a . c m i

The degree of labeling ranged from 0.7 to 0.8 mol of pyrene/mol of actin.

Spectrofluorometry. A Kontron SF-25 spec- trofluorometer equipped with an excitation slit of 5 nm and an emission slit of 10 nm was used. The excitation wavelength was 365 nm and the emis- sion wavelength was 407 nm. The intensity of the excitation light was reduced to about 20% of the initial intensity by use of a neutral density filter [26]. To minimize the effect of light scattering, a cutoff filter was used on emission to eliminate all wavelengths under 390 nm. The cuvette was main- tained at 25 °C by a thermostated water jacket.

Viscometry. Capillary viscometry tests were car- ried out using Ostwald-type viscometers con- structed by SICAS (S. Giuliano Milanese, Milan, Italy) according to the specifications of Cooper and Pollard [28]. The sample volumes were 0.5 and 1.0 ml, with a buffer flow time of 30 and 60 s, respectively. All the measurements were made at 25 °C in a thermostated water bath.

Pelleting. Sedimentation of the polymer (F- actin) was obtained using a Spinco 50 L ultra- centrifuge, with a type 40 rotor, at 110000 × g for 60-90 min. The amount of monomer (G-actin) in the supernatant was determined according to Bradford, taking 100 ~tl of supernatant from the meniscus.

Drug. Doxorubicin hydrochloride (M r 580) was supplied by Farmitalia Carlo Erba (Milan, Italy). The drug was kept in the dark, in a dry atmo- sphere, at 2-3 ° C. The solutions of antibiotic were prepared just before use. The concentration of doxorubicin was calculated from the absorbance at 480 nm, considering:

A480 ( C = 1 0 / ~ g / r n l ; p a t h l e n g t h = 1 cm) = 0 .174

The drug concentration used in all the treated samples was 17.2 /~M (10 /~g/ml). The drug was used without further purification.

Electron microscopy. Negative staining of actin polymer, formed in the presence or in the absence of doxorubicin, was performed with 2% uranyl acetate in 100 mM phosphate buffer, 50 mM KC1 (pH 7). After 30 s, excess stain was partially removed with filter paper, before drying in air. Preparations were examined at 80 kV acceleration voltage with a Jeol 100B microscope.

Results

Polymerization of actin is usually induced by the addition of specific salts to the protein solu- tion (optimal conditions, 100 mM KC1 plus 2 mM MgC12). We induced polymerization of the pro- tein by adding only 50 mM KCI (suboptimal conditions) to render the process slower and thus easier to monitor. Moreover, the suboptimal con- ditions made it possible to enlarge the differences between treated samples and controls.

It is known that the numerical values of the different parameters characteristic of a solution of G-actin induced to polymerize (viscosity, critical concentration, extent and rate of polymerization) vary, sometimes greatly, in different extractions and with the age of the solution [29]. In our experiments, each treated sample, therefore, had its corresponding control composed of the same actin solution. In this way, the differences between treated samples and controls were always of the same order, even when the numerical values varied as a function of that stated above. The figures are a graphic expression of symbolic experimental tests of the phenomena described.

Critical concentration To calculate the critical concentration in the

presence of the drug, different solutions of G-actin at different concentrations, range 0.1-0.7 mg/ml (2.3-16.2 #M), were polymerized overnight, at 25 °C with 50 mM KC1. Doxorubicin, 10 #g/ml (17.2 ~M), was added to the treated samples. At the end of polymerization (steady state) the viscos- ity was determined on some samples (treated and untreated); others were centrifuged to determine the quantity of polymer present. According to the

11

pelleting tests, after doxorubicin treatment the critical concentration (Cc) increased (Co (control) = 0.42 + 0.065 /~M; C c (treated) --- 0.66 + 0.13 /~M). The viscometric tests showed similar critical concentrations for treated samples and controls (Co (control) = 0.032 + 0.05 #M; C~ (treated) = 0.2 _+ 0.44/~M).

A ctin-doxorubicin interaction According to Someya et al. [18], we confirmed,

by gel filtration experiments (data not shown), the direct actin-doxorubicin interaction. It is interest- ing to point out that, after dilution, some of the drug dissociated from the protein molecule (about 30%, as determined by dialysis of the sample for one night).

Effect of doxorubicin on actin polymerization Samples of monomeric actin (11.6 ~M) were

polymerized by the addition of 50 mM KC1, in the presence or in the absence of doxorubicin (17.2 /~M). Increasing viscosity of the solution was fol- lowed by a capillary viscometer. As shown in Fig. 1, the polymerization process was slightly in-

q S P

0 . 7

0 . 6

0 . 5

0 . 4

0 . 3

0 . 2

0,1

O ~ 5 1C) 1'5 T I M E , r a i n .

Fig. 1. Time course of actin polymerization. G-actin solutions (11.6 #M) were polymerized in 50 mM KC1 at 25 o C. Forma- tion of the polymer was followed by determination of the specific viscosity (by capillary viscometry) at 3-rain intervals. O, Control samples; i , samples treated with doxorubicin (10

/*g/ml or 17.2 ~M).

12

hibited. In fact, analysis of the curves showed that the viscosity values of treated samples at any time point were lower than those of the controls. This led to an increase in treated samples in the delay time (or lag phase), a delay in reaching steady state and a lower plateau value. These results, considering the characteristics of the technique used, can be attributed to a slower formation of the polymer throughout the process and/or to the formation of slightly shorter filaments (in treated samples).

Effect of doxorubicin on the elongation step of actin polymerization

Samples of G-actin (11.6 btM), treated and con- trols, were polymerized by the addition of 50 mM KC1 in the presence of a solution of F-actin sonicated for 15 s at 50 W. The polymerization process was followed with viscometric (Fig. 2A) and spectrofluorometric (Fig. 2B) techniques. The figures clearly show that doxorubicin markedly inhibited the elongation phase. Comparison of the slopes relative to the rectilinear portion of the curves showed a 30% inhibition (by viscometric tests) and 40-50% inhibition (by spectrofluorome- try). The differences in percentage inhibition shown by the two methods can be explained by considering the increased degree of disturbance to the system under examination during viscosity measurements. For the spectrofluorometric mea- surements, 5% pyrene/actin was added to the reaction mixture.

Effect of doxorubicin on polymer size Some of the results obtained during this study

suggested an effect of the antibiotic on filament

Fig. 2. Elongation tests. (A) G-actin solutions (11.6 vM) were polymerized in 50 mM KCI at 25 o C, with the addition of 30 vl of a solution of F-actin (11.6/tM) at steady state, sonicated for 15 s at 50 W. Formation of the polymer was followed by determination of the specific viscosity at 60-s intervals: e, Control samples; II, samples treated with doxorubicin (17.2 vM). (B) G-actin solutions (3 ml, at a concentration of 11.6 g M + 5 % pyrene/actin) were polymerized in 50 mM KC1 at 25 o C, with the addition of 100/.tl of a solution of F-actin (32.5 /~M) at steady state and were sonicated for 15 s at 50 W. Formation of the polymer was followed spectrofluorometri- cally (excitation light = 365 nm; emission light ~ 407 nm) by determination of the fluorescence intensity at 15 s intervals.

Symbols as for A.

length. In other words, it is probable that doxorubicin affects the polymerization of actin by decreasing the quantity of polymer produced (an effect of little practical importance) as well as reducing the definitive length of the filaments being formed.

To prove this working hypothesis, we used a technique described by Maclean-Fletcher and Pol-

qso

0 . 6

0 . 5

0 . 4

0 . 3

0 . 2

0 . 1

0

i

A 1 0 T I M E , r n i n . 5

i Z UJ P z W u Z LU u

W rr O

.J U.

0 5 0 1 0 0

B T I M E , s e e .

13

lard [30]. We polymerized equal quantities of the same G-actin solution (100/~1) with 50 mM KC1 in the presence or in the absence of doxorubicin (17.2 FM) at 25°C, overnight. The following morning, at the end of polymerization, 3 ml of a solution of G-actin (11.6 /~M) plus 50 mM KC1 was added to treated and to control samples. The polymerization procedure was then followed with viscometry and with spectrofluorometry. The sam- ples prepared for spectrofluorometry contained, in the 3 ml of G-actin added, 5% pyrene/act in . Viscometric (Fig. 3A) and spectrofluorometric (Fig. 3B) results showed an increase in the rate of polymerization in samples in which the germs of polymerization present (100/ t l of actin polymer- ized overnight) were formed in the presence of the antibiotic. This datum clearly indicates that in the presence of the drug formation of a greater num- ber of germs (filaments), which obviously must be smaller than those formed in control samples, occurred.

In order to understand better the philosophy of this methodology, it should be remembered that the increase in the rate of polymerization is a function, when equal quantities of polymer are added, of the number of ends added. It is then easy to understand that, when equal quantities of polymer are added (Fg /ml ) , it is possible to ob- tain different polymerization rates as a function of the number of filaments (or ends) into which the polymer is divided [30].

At this point considering also the results of Someya et al. [18], we tested the action of doxorubicin alone on solutions of monomeric

Fig. 3. Test of polymer size. (A) G-actin solutions (11.6 /tM) were polymerized in 50 mM KC1 at 25 o C, in the presence of: (®) 100 #1 of a solution of F-actin (11.6 FM) polymerized overnight in 50 mM KC1 at 25 o C, or (o) 100 FI of a solution of F-actin (11.6 FM) polymerized overnight in 50 mM KCI at 25 o C in the presence of doxorubicin (17.2 #M). Formation of the polymer was followed by determination of specific viscos- ity at 90-s intervals. (B) G-actin solutions (3 ml at concentra- tion of 11.6 FM plus 5% pyrene/actin) were polymerized in 50 mM KCI at 25 o C, in the presence of: (O) 100 #1 of a solution of F-actin (32.5/tM) polymerized overnight at 25 o C in 50 mM KCI, or (I) 100 ~tl of the same solution polymerized in the presence of doxorubicin (17.2 /~M). Formation of the polymer was followed by determination of fluorescence intensity (exci- tation light = 365 nm; emission light = 407 nm) at 15 s inter-

vals.

actin. Viscometric tests showed that doxorubicin alone cannot provoke polymerization of actin, as we previously determined (data not shown).

For further proof of the action of doxorubicin on filament size, samples of actin (11.6/~M) were incubated overnight at 25 °C under polymerizing

condit ions (50 mM KC1), in the presence or in the

q EIl=

0 .7

O.m

0 , 5

0 , 4

0 . 3

0 2

0.1

A 0 5 10 15 TIME,rn in .

z m

z

W u z u m uJ iz o 3 .I n

5 0

2 5

I l l

B 0 2 0 4 0 T I M E , s e e .

14

r

0 . 5 0

0 . 2 5

i 5 1 5 T I M l = , m i r l .

Fig. 4. Assay to reveal the presence of small oligomers in the supernatants. G-actin solutions (3 ml at the concentration of 11.6 /~M) were polymerized in 50 m M KCI at 2 5 ° C in the presence of: (e) 150 ~l of supernatant from centrifugation of an actin solution (11.6 /xm) polymerized overnight in 50 m M KCI at 25 ° C, or (n) 150/~l of supernatant from centrifugation of a solution of actin, identical to the previous one, polymer- ized under the same conditions but with the addition of doxorubicin (17.2 /tin). Formation of the polymer was fol- lowed by determination of the specific viscosity at 60-s inter-

vals.

absence of doxorubicin. At the end of polymeriza- tion (steady state) the samples were centrifuged for 150 min at 110000 x g and the supernatants were collected by turning the centrifuge tube up- side down. At this point, G-actin samples (3 ml) were polymerized by the addition of 50 mM KC1, in the presence of 150 /~1 of the collected super- natants. Formation of the polymer with time was followed by viscometry. Fig. 4 shows the graph of a characteristic experiment. In the samples in which 150 tzl of supernatants from the treated samples had been added there was a shortening of the delay time (Td). This fact clearly indicates that in the treated supernatants there were still mole- cules of the polymer which, because of their reduced size, sediment with difficulty. The reduc-

Fig. 5. Electron microscopic analysis of actin polymer. F-actin formed in the presence of doxorubicin (B) showed some shorter actin filaments (arrows) which normally are not present in

controls (A). Bar scale = 200 nm.

tion in the delay time, evaluated according to the indications of Cooper et al. [31], was 25% of that of controls.

Finally, electron microscopic analysis of F- actin, formed under doxorubicin treatment, con- firmed the presence of short filaments (Fig. 5B).

Effect of doxorubicin on the annealing, step As previously suggested, doxorubicin appears

to act negatively on the polymerization of actin by reducing the size of the filaments being formed. This result could be obtained by two distinct mechanisms which may act individually or to- gether. Actin filaments of a reduced size could be the result of perturbations of the elongation phase, or they could derive from a slowed (or inhibited) annealing. Using viscometric techniques we evaluated the capacity of F-actin solutions, after sonication, to re-establish initial viscosity [30] (in the absence or in the presence of doxorubicin). No significant differences between controls and treated samples were found (data not shown), thus suggesting a refractoriness of the annealing phase to the action of the antibiotic.

Discussion

In our previous studies [19,20], we suggested that doxorubicin negatively influenced actin monomer assembly, in contrast with the data of Someya et al. [18].

In the present study, we sought to determine, during actin polymerization, the chemical event(s) sensitive to the action of doxorubicin. First, we investigated the doxorubicin/G-actin interaction. In agreement with Someya et al. [18], we have confirmed that doxorubicin binds to monomeric actin.

The evaluation of the actin polymerization mechanism under doxorubicin treatment revealed the following facts. (1) An increase in the con- centration of monomer in equihbrium with the polymer at steady state (critical concentration), according to pelleting data. The same parameter, calculated by capillary viscometric methods, did not show significant variations between treated and control samples. (2) The time course of polymer formation, in 50 mM KCI and 17.2 #M of doxorubicin and followed viscometrically, ap- peared to be discretely inhibited. In particular, Fig. 1 shows that in the presence of the antibiotic, the delay-time (lag-phase), calculated according to Cooper et al. [31], doubled its value (T d (control) = 1.36 min; T d (treated) = 2.8 min). Moreover, in treated samples, the rate of polymerization was inhibited by about 13% and the plateau was lower than that of controls by about 10%. (3) Experi- ments carried out to evidence the effect of doxorubicin on elongation showed a marked sensitivity of this step of polymerization to the action of the antibiotic. Fluorometric as well as viscometric data (Figs. 2A and 2B) indicated that doxorubicin reduced the elongation rate. Only the percentage inhibition, determined using two dif- ferent methods, was numerically different (30% with capillary viscometry and 40-50% with spectrofluorometry). In our opinion, this fact is attributable to a strong disturbance in the system during viscometric measurements. (4) The most significant data, in our opinion, are those derived from the experiments carried out to determine the average size of the polymer. The polymer formed in the presence of the antibiotic was capable of stimulating actin polymerization to a greater ex-

15

tent (Figs. 3A and 3B). Since the increased veloc- ity in the process of aggregation of actin monomers due to the addition of preformed polymer to the system, when an equal quantity of polymer was added, is a function Of the number of ends sup- plied, we concluded that the polymer formed in the presence of the antibiotic was fractioned into shorter filements than those of control samples. Furthermore, small aliquots of supernatants from centrifugation of treated actin samples at steady- state were capable of stimulating actin polymeri- zation in 50 mM KC1 (Fig. 4). This fact clearly indicates that in the supernatants of treated sam- pies, F-actin was present (in a greater quantity than in controls) in the form of small filamentous structures (small oligomers) which fail to sediment under the centrifugation conditions used. The polymer formed under doxorubicin treatment re- vealed the presence of short filaments, when ex- amined by electron microscopy.

In view of our results, we must conclude that doxorubicin, under our experimental conditions, acts negatively on actin polymerization. The data obtained in the elongation tests (Figs. 2A and 2B) and in the evaluation of the number of ends (Figs. 3A and 3B) represent in our opinion, the key to understanding the mechanism of action of doxorubicin on actin polymerization.

Our data indicate that when doxorubicin is added to a G-actin solution, some monomers in- teract with the drug. In the solution there would thus be normal monomers and monomers altered by interaction with the antibiotic. At steady state, in the presence of doxorubicin, several species of polymer are likely to be present: (a) filaments with both ends free; (b) filaments with one end capped by altered monomer; (c) filaments with both ends capped by altered monomers.

When treated F-actin samples are strongly di- luted (e.g., in the experiences related to the Fig. 3A and B), the doxorubicin bound to the capping subunits is released from protein molecule; a sub- stantial part of the molecular classes in which the polymer formed in the presence of the drug is fractioned is, therefore, transformed in as many centers of polymerization, which gave rise to the differences observed in polymer size tests (Fig. 3A and B).

If we hypothesize an analogous shortening ac-

16

tion of the drug in vivo, it is possible to explain many of the effects we observed on cell cultures and explanted embryos [19,20]. Since gel capacity is inversely proportional to filament length (aver- age weight of the polymer) [32], short filaments could fail to support cellular phenomena (morpho- logical changes in cells of neural tube, cytoplasmic extroflession during morphogenesis of nervous cells, etc.) which are altered after treatment with doxorubicin. In contrast, to extrapolate these con- cepts to muscle cells, where actin structures show a high stability, is not so immediate.

Acknowledgements

The study was supported by grant No. 86.00502.44 from the Consiglio Nazionale delle Ricerche, Rome.

References

1 Benjamin, R.S. (1979) Recent Results Cancer Res. 63, 230-240.

2 Carter, S.K. (1975) J. Natl. Cancer Inst. 55, 1265-1274. 3 Young, R.C., Rosenoff, S. and Olson, H.M. (1975) in

Adriamycin Review (Staquet, M., Tagnon, H., Kenis, Y., Bonadonna, G., Carter, S.K., Sokal, G., Trouet, A., Ghione, M., Praga, C., Lenaz, L., Karim, O.S., eds.), pp. 149-159, European Press Medikon, Ghent.

4 Young, R.C., Ozols, R.F. and Meyers, C.E. (1981) N. Engl. J. Med. 305, 139-153.

5 Cortes, E.P., Lutman, G., Wanka, J., Wang, J.J., Pikren, J., Wallace, J. and Holland, J.F. (1975) Cancer Chemother. Rep. 6, 215-255.

6 Henderson, I.C. and Frei, E. (1979) N. Engl. J. Med. 300, 310-312.

7 Lefrak, E.A., Pitha, J., Rosenheim, S. and Gottlieb, J.A. (1973) Cancer 32, 302-314.

8 Lefrak, E.A., Pitha, J., Rosenheim, S., O'Bryan, R.M., Burgess, M.A. and Gottlieb, J.A. (1975) Cancer Chemother. Rep. 6, 203-208.

9 Praga, C., Foa, P., Necco, A. and Bertolazzi, C. (1975) in Adriamycin Review (Staquet, M., Tagnon, H., Kenis, Y., Bonadonna, G., Carter, S.K., Sokal, G., Trouet, A., Ghione, M., Praga, C., Lenaz, L., Kar]m, O.S., eds.) pp. 132-142, European Press Medikon, Ghent.

10 Domal, N., Sawada, H., Matsuyama, E., Konishi, T., Uchino, H. (1981) Cancer Treat. Rep. 65, 79-91.

11 Olson, R.D., MacDonald, J.S., Von Boxtel, C.J., Boerth, R.C., Harbison, R.D., Slonim, A.E., Freeman, R.W. and Oates, J.A. (1980) J. Pharmacol. Exp. Ther. 215, 450-454.

12 Olson, H.M. and Shannon, C.F. (1979) Cancer Treat. Rep. 63, 2057-2059.

13 Olson, H.M., Young, D.M., Prieur, D.J., LeRoy, A.F. and Reagan, R.L. (1974) Am. J. Pathol. 77, 439-454.

14 Arena, E., Arico, M., Biondo, F., D'Alessandro, N., Gerbari, F., Sanguedolce, R. and Rausa, L. (1975) in Adriamycin Review (Staquet, M., Tagnon, H., Kenis, Y., Bonadonna, G., Carter, S.K., Sokal, G., Trouet, A., Ghione, M., Praga, C., Lenaz, L., Karim, O.S., eds.), pp. 160-172, European Press Medikon, Ghent.

15 Revis, N. and Maurics, N. (1979) Exp. Mol. Pathol. 31, 440-451.

16 Nicolay, K., Fok, J.J., Voorhout, W., Post, J.A. and De Kruijff, B. (1986) Biochim. Biophys. Acta 887, 35-41.

17 Necco, A. and Ferraguti, M. (1979) Exp. Mol. Pathol. 31, 353-360.

18 Someya, A., Akima, T., Misumi, M. and Tanaka, N. (1978) Biochem. Biophys. Res. Commun. 85, 1542-1550.

19 Necco, A., Vailati, G. and Colombo, R. (1983) Cell Biol. Int. Rep. 7, 129-134.

20 Colombo, R., Necco, A., Vailati, G., Saracco, B., Milzani, A. and Scar], G. (1984) Cell Biol. Int. Rep. 8, 127-135.

21 Spttdieh, J. and Watt, S. (1971) J. Biol. Chem. 246, 4855-4871.

22 MacLean-Fletcher, S. and Pollard, T.D. (1980) Biochem. Biophys. Res. Commun. 96, 18-27.

23 Bradford, M. (1976) Anal. Biochem. 72, 248-254. 24 Houk, T.W. and Ue, K. (1974) Anal. Biochem. 62, 66-74. 25 Tellam, R. and Frieden, C. (1982) Biochemistry 21,

3207-3214. 26 Cooper, J.A., Walker, S.B. and Pollard, T.D. (1983) J.

Muscle Res. Cell Motil. 4, 253-262. 27 Kouyama, T. and Mihashi, K. (1981) Eur. J. Biochem. 114,

33-38. 28 Cooper, J.A. and Pollard, T.D. (1982) Methods Enzymol.

85, 182-210. 29 Oosawa, M. and Maruyama, K. (1986) J. Biochem. 100,

1001-1008. 30 MacLean-Fletcher, S. and Pollard, T.D. (1980) Cell 20,

329-341. 31 Cooper, J.A., Buhle, E.L., Walker, S.B., Tsong, T.Y. and

Pollard, T.D. (1983) Biochemistry 22, 2193-2202. 32 Flory, P. (1953) Principles of Polymer Chemistry, Cornell

University Press, Ithaca.