(CANCER RESEARCH 50. 7686-7696. December I. 1990]

An in Vivo Study of the Role of the Tumor Cell Cytoskeleton in Tumor Cell-Platelet-Endothelial Cell Interactions'

Hemi Chopra, Suzanne E. G. Fligiel, James S. Hatfield, Kevin K. Nelson, Clement A. Diglio, John D. Taylor, andKenneth V. Honn2Departments of Biological Sciences ¡J.D. T.¡,Radialion Oncology [H. C., K. K. N., J. D. T., K. \'. H.], and Pathology [C. A. D., S. E. G. F.], Wayne State University,

Detroit, Michigan 48202; Gershenson Radiation Oncology Center [J. D. T., K. V. H.], Detroit, Michigan 48201; and Laboratory Service, VA Medical Center fj. S. H.,S. E. (i. F.J, Allen Park, Michigan 48IOI

ABSTRACT

We recently reported that disruption of tumor cell microfilaments orintermediate filaments resulted in an inhibition of the ability of tumorcells to induce the aggregation of homologous platelets in vitro (H.Chopra et al.. Cancer Res., 48:3787-3800, 1988). Previous investigatorsdemonstrated that disruption of the tumor cell cytoskeleton decreasesthe ability of these cells to form lung colonies. \Ve proposed that thislatter effect is due, in part, to decreased interaction of tumor cells withplatelets, following their arrest in the microvasculature. To test thishypothesis, B16 amelanotic melanoma cell microtubules, microfilaments,or vimentin intermediate filaments were disrupted with colchicine (50Mm),cytochalasin D (50 /mi), or cycloheximide (50 urn), respectively,and then cells were tail vein injected into syngeneic mice. Both cytochalasin D- and cycloheximide-treated cells formed fewer lung colonies thandid control cells. Colchicine, however, failed to inhibit lung colony formation. Neither colchicine nor cycloheximide treatment altered initialpulmonary arrest; however, fewer cycloheximide-treated cells remainedin the lungs 8 h postinjection. Greater than 90% of control or colchicine-treated cells were found to be associated with activated platelets, andthey also demonstrated typical cell membrane process formation 10 minand 8 h post-tumor cell injection. In contrast, less than 10% of cycloheximide-treated cells were in contact with activated platelets 10 min post-injection. However, by 8 h =90% of cycloheximide-treated cells were incontact with activated platelets. This recovery coincided with the reformation of the B16 amelanotic melanoma vimentin intermediate filamentnetwork and the reacquisition of the ability to induce platelet aggregationin vitro. Neither colchicine nor cycloheximide treatment altered initialB16 amelanotic melanoma cell adhesion to murine microvessel-derivedendothelial cells. This study provides in vivo evidence in support of ourprevious findings that disruption of certain cytoskeletal elements (i.e..vimentin intermediate filaments) inhibits the tumor cell ability to activateplatelets. This study also suggests that platelet activation may stabilizethe initial tumor cell arrest in the microvasculature.

INTRODUCTION

Tumor cell arrest in the microvasculature is a critical steprequired for the formation of metastatic foci (1, 2). Followingarrest, tumor cells must establish stable contacts with the en-dothelium, induce endothelial cell retraction, migrate, and attach to the subendothelial matrix. Finally, they must proteolyt-ically digest the subendothelial matrix, extravasate, and proliferate (1-9). The successful completion of this series of eventsrequires the coordinated action of phenotypic characteristics(i.e., adhesion receptors, proteinases, etc.) which determinesthe metastatic phenotype. One phenotypic characteristic whichmay regulate and coordinate this response is the organizationof the various tumor cell cytoskeletal components (10). Hagmarand Ryd (11) provided the first direct evidence for a role of thetumor cell cytoskeleton in metastasis. They observed a marked

Received 5/17/90; accepted 8/30/90.The costs of publication of this article were defrayed in pan by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1This study was supported by NIH Grant CA 47115-02. a grant from Harper

Grace Hospitals, and a grant from the Veterans Administration.2To whom requests for reprints should be addressed.

alteration in the organ distribution of metastatic foci followingmicrofilament disruption in TA3 tumor cells and concludedthat this was due to alterations in tumor cell locomotion.Following this observation, Hart et al. (12) demonstrated thatthe disruption of B16F10 tumor cell microtubules or microfilaments prior to tail vein injection resulted in a decreasednumber of pulmonary tumor colonies, an alteration in thedistribution pattern of metastasis, and a decreased adhesion tobovine endothelial cells. These effects were not the result ofalterations in initial pulmonary arrest, cell volume, or growthrate. Since this pioneering work, several investigators havedemonstrated the importance of the organization of actin filaments (13), intermediate filaments (14), and microtubules (15)in several aspects of metastasis (i.e., invasion, adhesion, lungcolony formation, etc.). A variety of mechanisms have beenproposed to explain the effects of cytoskeletal disruption ontumor cell metastasis (10). These include cell deformability,decreased homotypic and heterotypic aggregation, decreasedmotility, and decreased adhesion to endothelial cells.

Using W256' cells, previously, we demonstrated that disrup

tion of microtubules, microfilaments, or vimentin intermediatefilaments resulted in the inhibition of the ability of tumor cellsto induce macroscopic platelet aggregation (16). However, atthe ultrastructural level, there was evidence of platelet-tumorcell interactions in colchicine (i.e., microtubule disrupted)-treated cells, but not in those cells in which the microfilamentsor vimentin intermediate filaments were disrupted. Subsequentstudies, using a cell line (i.e., B16a) in which colchicine did notdisrupt intermediate filaments, demonstrated that inhibition ofmacroscopic TCIPA by colchicine was due to a secondary effectof disrupting vimentin intermediate filaments (Ref. 16; Footnote 4). The observation that TCIPA is dependent upon cytoskeletal integrity suggests that the in vivo results of previousinvestigators using cytoskeletal disrupted tumor cells (11, 12,14) may be due, in part, to an inhibition of tumor cell-plateletinteraction following arrest, as platelets are proposed to stabilize the bond between the tumor cell and the vessel wall (17-19) in addition to facilitating endothelial cell retraction (20).Therefore, in this study, the effects of cytoskeletal disruptionon lung colony formation by B16a cells were examined. Inaddition, morphometric and ultrastructural evidence for cytoskeletal dependent tumor cell-platelet interactions in vivo is

presented.

' The abbreviations used are: W256. Walker 256 carcinosarcoma cells; B16a,B16 amelanotic melanoma; FBS. fetal bovine scrum; GpIlb/IIIa. platelet glyco-protein llh 111.>complex: HBSS. Hanks' basic salt solution; IRGplIb/IIIa. ¡m-

munoreactive glycoprotcin llb/IIIa; MEM. minimal essential medium; pAbllb/Ilia, polyclonal antibody raised against human platelet lib Ilia complex; PBS,phosphate-buffered saline; TCIPA, tumor cell-induced platelet aggregation;T\B3, hydrolysis product of thromboxane A2.

4H. Chopra. J. S. Hatfield. S. E. G. Fligiel, 1. M. Grossi, Y. Chen. C. A.Finch. L. A. Fitzgerald. J. D. Taylor, and K. V. Honn. An in vitro study of roleof the tumor cell cytoskeleton and integrin >rMb/j3in tumor cell-platelet interactions, submitted for publication.

7686

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CVTOSKELETON AND METASTASIS

MATERIALS AND METHODS

Tumor Lines

The B16a tumor cell line was obtained from the Division of CancerTreatment, NIH (Frederick, MD) and was passaged in syngeneic maleC57BL6/J mice (Jackson Laboratories, Bar Harbor, ME) as previouslydescribed (Ref. 21; Footnote 4).

Tumor Cell Culture

B16a tumors were removed aseptically. They were disaggregated andplated in tissue culture flasks containing MEM (Gibco, Grand Island.NY) supplemented with 5% FBS. Adherent tumor cells were grown toconfluency and serially passaged as previously described (Ref. 21;Footnote 4). Low passage (i.e., second to fifth) cell cultures were usedin this study. All cells were at 55% to 65% confluency when used inthe experiments described below. Cells were harvested using EDTA(0.45 mmol), washed, and resuspended in the appropriate medium orbuffer.

Cytoskeletal Inhibitors

Cytochalasin D (Sigma; St. Louis, MO), a microfilament inhibitor(22), was dissolved in dimethyl sulfoxide. Colchicine (Sigma), a micro-tubule inhibitor (23), was dissolved in glass-distilled water. Cyclohex-imide (Sigma), an inhibitor of the intermediate filament network (24),was dissolved in absolute ethanol. Treatment of tumor cells withcolchicine (50 ^mol, 15 min, 25°C)or cytochalasin D (50 pmo\, 15min, 25°C)led to the disruption of their microtubules and microfila-

ments, respectively. Treatment of tumor cells with cycloheximide (50¿¿mol,8 h, 37°C)disrupted their vimentin-containing intermediate

filaments. All treatments were conducted as described previously (Ref.16; Footnote 4).

Platelet-rich Plasma and Aggregation Assays

Blood was drawn from the inferior vena cava of anesthetized miceinto a solution of heparin sulfate (25 units/ml) and 4.8% dextrose in0.9% saline solution (heparimblood ratio = 1:9, v/v). Platelet-richplasma and platelet-poor plasma were prepared as previously described

1 50-

1 40-

1 30-

£ 120-

oEO

O*

100-

90-

80-

70-

60-

50-

40-

30-

20-

1 0-

0

X

l

nL

CTL CL CD CXFig. 1. Effect of disrupting individual elements of B16a cytoskcleton on lung-

colonizing ability. Colchicine (CL, 50 nmol) failed to inhibit colony formation,whereas cytochalasin D (CD, 50 ¿imol)or cycloheximide (C.V. 50 jimol) inhibitedlung colony formation compared with control (CTL). Columns, mean; bars. SD.

Fig. 2. Light micrographs to demonstrate absence or presence of TCIPA inthe lung microvasculature following tail vein injection of B16a cells. In a, acontrol B16a cell (arrow) is completely surrounded and partially covered bynumerous clumps of platelets (*). In h, a cycloheximide-treated tumor cell (arrow)is arrested in a pulmonary vessel and is in close proximity to numerous erythro-cytes. However, there are no platelet clusters demonstrable anywhere near thetumor cell (l>). These images are typical of those used to score tumor cellassociation with platelet aggregates (Fig. 3). a, x 1,200; A, x 16,000.

(Ref. 25; Footnote 4). Aggregometry studies were performed with aModel DP-247E dual channel aggregometer (Seinco, Morrison, CO)as described previously (Ref. 25; Footnote 4). Tumor cells were pre-treated with various cytoskeletal disrupting agents or solvent controls,washed with MEM, assayed for their ability to aggregate platelets, andthen used for the in vivo or in vitro assays described below.

Experiments were performed to test if the effect of cycloheximide onTCIPA was reversible. Tumor cells were treated with cycloheximide asdescribed above. At the end of 8 h the cells were washed (3 times) withsterile MEM containing 5% FBS and allowed to recover for 8 h at37°C.The cells were then harvested with EDTA (0.45 mmol), washed,

and tested for their ability to aggregate platelets as described above.

Experimental Metastasis Assay

Tumor cells were harvested and treated with colchicine as describedabove or treated with cycloheximide (50 ^mol) for 8 h at 37°Cand then

7687

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

<o11° 1 <n11°—H<100—

C3illa:

goesO<

80—tuÜJ

70—h™¿

60Xt

50~~in

.-J40m

0soCEO

220—1-

ií 10 —JLj_J^—L-tup0LU0

90—13l"-ÜJ

70—¿

60—xt

SO3

40—ai^30

-O|20

—

SÕ1 0—iàCTL

CL CTL CL CTLCXa1.41.2EEf

0.8—o0.4

—0.2—a.a_10'

10' 8h 8hb1.41.2

—EE2"

0.8u0.40.2.0.010'

10'1CTL8hCX8h

CTL10- CLIO' CTLSh CLBh CTL101 CXI O1 CTLSh CX8h

Fig. 3. Morphometric evaluation of the effects of disrupting the tumor cell microtubules (colchicine, CL, 50 //muli or the vimentin intermediate filaments(cycloheximide, CX, 50 ,;inul i on tumor cell association with platelet aggregates (a, b) and tumor cell arrest and retention (c, d). CTL, control. Columns, mean; bars,SD.

harvested; control cells were treated with appropriate solvents. Thecells were then washed (twice) with serum-free MEM and tail veininjected into 8- to 9-wk-old male C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME). Animals used for lung colony enumeration weresacrificed by cervical dislocation 21 days postinjection. Their lungswere removed and placed in Bouin's fixative, and visible tumor colonies

were counted as described previously (26). In all experiments, micewere randomly chosen from stock cages for each treatment group. Allgroups received injections within 15 to 25 min post-cell preparation. Aminimum of 12 animals were used per experimental group. Eachexperiment was performed twice.

Histology, Electron Microscopy, and Morphometry

For histological and morphometric evaluation, animals were sacrificed by cervical dislocation at 10 min and 8 h post-tail vein injection;their lungs were removed, processed, and analyzed as described below.Five animals were used per treatment group.

Specimen Preparation. The lungs were immediately removed andplaced in fixative (2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH7.2). Six to ten 1-mm cubes were dissected from the lower lobes of thelung of each animal and stored in the above fixative at 4°C.Six pieces

of lung tissue per mouse (30 per experimental group) were routinelyprocessed and embedded in Medcast resin (Pella, Inc., Redding, CA).Ninety to 120 sections l ¿imthick were cut per group, stained with amixture of 1% méthylèneblue and 1% Azure II, and examined by lightmicroscopy for B16a cell-platelet interaction. In addition, morphome-

try studies as described below were conducted at the light microscopiclevel to corroborate with ultrastructural evidence the extent of B16acell interaction with platelets and endothelial cells. Three to 12 thinsections (60 to 70 nm) per experimental group were cut with a diamondknife from blocks of lung tissue. The thin sections were stained withaqueous uranyl acetate and lead citrate and examined with a Zeiss EM109 electron microscope.

Morphometric Analysis. Thirty randomly selected 1-^m sections per

7688

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

1

1

CTL10'

cx10'

CTLen

CXBh

Fig. 4. Blood TXB2 levels 10 min and 8 h postinjection of control (CTL) orcycloheximide (CX, 50 ;jmol)-treated B16a cells. Columns, mean; bars, SD.

experimental group were used for further analysis. Analysis includedthe percentage of intravascular B16a cells associated with platelets andthe number of B16a cells per unit area (unit area measured as mm2 of

lung section).To examine the association with platelets, sections were scanned at

x400 magnification, and B16a cells with and without associated plateletaggregates were counted. The number of tumor cells observed variedfrom a minimum of 84 to a maximum of 466 per experimental group.Results are expressed as the mean ±SD.

To determine the number of arrested B16a cells/unit area, the areaof each section was determined using a Bloquant image analysis system(R & M Biometrics Corporation, Nashville, TN). For each experimentalgroup a minimum area of 186 mm2 of lung parenchyma was evaluated.The number of B16a cells/mm2 was calculated. Results are expressed

as the mean ±SD.

Radioimmunoassay

The determination of serum TXA2 (measured as the stable hydrolysisproduct, TXB2) was used as another indicator of the extent of plateletactivation in vivo post-B16a cell injection. B16a cells (3.7 x 10"), the

same number as that used in the morphological studies described above,were injected into the tail vein of mice. Six animals were used pergroup. Animals were then anesthetized (ether inhalation), and bloodsamples were removed from the inferior vena cava for TXB2 analysisby radioimmunoassay as previously described (25). The animals werebled at 10 min and 8 h post-tumor cell injection. Blood was drawn intosyringes containing 25 units of heparin, indomethacin (cyclooxygenaseinhibitor; final concentration, 25 ^mol), and prostacyclin (final concentration, 100 nmol) to prevent platelet activation during sampling. TXB:levels were measured by radioimmunoassay. Briefly, following centrif-ugation (150 x g), supernatants were transferred to acetone (4°C)and

extracted by octadecyl (C18) reverse-phase chromatography prior toradioimmunoassay analysis as previously described (25). Prior to assay,samples were dissolved in 1 ml of phosphate buffered saline-bovine y-globulin (Miles Laboratories, Inc.). Anti-TXB2 antiserum was purchased from Seragen (Cambridge, MA) and [3H]TXB2 was purchasedfrom Du Pont-New England Nuclear (Boston, MA). Standard andexperimental analyses were computed with a Hewlett Packard Model41CV computer using a weighted data reduction program (25). Resultsare expressed as ng of TXB2/ml of blood.

Endothelial Cell Culture

C57BL/6J mice were utilized to obtain the pulmonary-derived endo-thelial cells used in these studies. Mice were decapitated, and lungswere aseptically removed and rapidly placed into cold Ca2*- and Mg2*-

free HBSS. Following this initial wash, the pleural lining of the lungwas fixed gently by applying 70% ethanol over the lung surface. Thisprocedure eliminated mesothelial cell contamination in the developingculture preparation. After rinsing in HBSS, small tissue pinches, usingfine forceps, were obtained close (1 to 2 mm) to the lung periphery toavoid obvious large vessels. Tissue expiants were then treated with 0.1 %

collagenase (Type II; Worthington, Malvern, PA) in Ca2+- and Mg2+-free HBSS for 20 min at 37°C.The collagenase-treated tissue was

carefully removed and plated as expiants into 100-mm tissue culturedishes containing Dulbecco's modified Eagle's medium supplemented

with 20% FBS. After 12 to 14 days in culture, the expiants wereremoved, and areas containing endothelial cells, identifiable by theirtypical cobblestone growth pattern, were marked and allowed to growan additional 7 days. Isolated colonies of endothelial cells, devoid ofspindle cells, were selectively tryspinized using cloning penicylinders.Three separate endothelial cell clones were isolated, grown to con-fluency, and recloned. One surviving clone was established, routinelysubcultured at 1:2 split ratios every 7 to 10 days, and designated CD3.Endothelial cell characteristics were verified by growth behavior pattern, the presence of Factor VIII, and prostanoid production accordingto published procedures (27, 28). These cells were used in adhesionassays as described below. Endothelial cells were grown in Dulbecco'smodified Eagle's medium and 10% FBS in the presence of 10% CO2 at37°Cin tissue culture flasks.

Adhesion

Endothelial cells (CD3) were plated on 24-well plates (Falcon; BectonDickinson, Oxnard, CA) at a concentration of 250,000 cells/well andgrown to confluency. Tumor cells were pretreated with the cytoskeletalinhibitors colchicine (50 /¿mol)or cycloheximide (50 j/mol) as describedabove, washed, resuspended in platelet wash containing 2 IHMCaCI2and 2 mM MgCl2, and plated on wells containing CD3 cells. Tumorcells (5 x IO5)were plated on wells containing confluent CD3 cells andincubated for 10 min at 37°C.Nonadherent cells were removed by

washing each well (twice) with platelet wash (phenol red-free MEM,pH 7.2; Gibco) containing 2 mM CaCI2 and 2 mM MgCl2. In addition,tumor cells were treated with cycloheximide as described above, washedwith sterile MEM containing 5% fetal calf serum, allowed to recoverfor 8 h at 37°C,and then tested for their ability to adhere to CD3

endothelial cells.

Immunofluorescence

B16a cells (5 x IO5cells/ml) were grown on coverslips placed in 6-well culture plates (Falcon) for 48 h at 37°C.These cells were incubatedwith cycloheximide (50 ^mol) or ethanol (0.1 %) for 8 h at 37°C,rinsed

in PBS buffer (pH 7.2), and drained to remove excess buffer. To localizethe vimentin-containing intermediate filaments, cells were permeabi-lized with acetone and processed for routine ¡mmunofluorescence asdescribed previously.4

Antivimentin antibody (ICN, Lisle, IL) and fluorescein isothio-cyanate-labeled rabbit anti-goat antibody (ICN) were used to localizethe vimentin-containing intermediate filaments as described previously.4

Statistics

Raw data from tumor colony-forming assays were analyzed forsignificance using nonparametric statistical analysis (Kruskal-Wallis)and presented as the percentage of respective solvent control. Otherdata were analyzed by the analysis of variance test and the Scheffe testusing the STATA VIEW 512 + software (Brain Power, Inc., Calabasas,CA) on a Macintosh Plus computer (Apple Computer, Cupertino, CA).Differences were considered significant when P < 0.05.

RESULTS

Effect of Cytoskeletal Disruption on Lung-colonizing Ability.B16a cells were treated with colchicine, cytochalasin D, cycloheximide, or the appropriate solvents, and they were tested fortheir ability to aggregate homologous platelets and tail veininjected into C57BL6/J mice as described in "Materials andMethods." Treatment of B16a cells with these compounds

disrupted microtubules, microfilaments, and vimentin intermediate filaments.4 Under the conditions used, cycloheximide

treatment did not alter the indirect immunofluorescent staining7689

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TLMOR CELL OTOSKELETON AND MK'IAS'I AMS

Fig. S. Ten min postinjection, a control BI6a cell is arrested in the pulmonar) microvasculaturc with its plasma membrane intiniatcl) juxtaposed loan endothclialcell plasma membrane (arrow). In a. a localized aggregate of activated platelets can be observed in contact with the BI6a plasma membrane (*). In h. as first reportedby Menter el ai. (43), characteristic tumor cell processes interdigitale with the platelet aggregate (*). Similarly, in c. 10 min postinjection. colchicine-treated BI6acells were observed in intimate contact with endothelial cells (arrow) and associated with focal aggregates of activated platelets (*). In d, colchicinc-trcatcd B16a cells

demonstrated evidence of process formation in areas of contact with activated platelets (arrow), a. x 5.500; h. x 6.000; c. x 6.500; d, x 9.193.

7690

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

Fig. 6. Ten min postinjection cycloheximide-treated B16a cells were observedin the microvasculature in contact with endothelial cell plasma membrane (arrow).Although unactivated platelets are visible immediately adjacent to the BI6a cellplasma membrane (*), no platelet aggregates were observed, and the B16a cellplasma membrane is smooth without process formation (a). A higher magnification of the above micrograph shows the area of tumor cell-platelet contact. The

pattern for either microtubules or microfilaments (data notshown). As reported earlier4 colchicine treatment did not alter

the ability of B16a cells to induce platelet aggregation, whereastreatment with cytochalasin D or cycloheximide consistentlyinhibited their ability to aggregate homologous platelets (datanot shown). Enumeration of lung colonies 21 days later revealedsignificantly decreased lung colony formation by cytochalasinD or cycloheximide-treated B16a cells (Fig. 1). In contrast,colchicine-treated cells did not form a decreased number oflung colonies but rather, a slight increase was observed in someexperiments (Fig. 1).

Morphology and Morphometric Analysis. In v/vo, tumor cell-platelet-endothelial cell interactions by solvent control andcolchicine- and cycloheximide-treated cells were evaluated directly by light and electron microscopy at 10 min and 8 h post-

tumor cell injection. Selection of these time intervals was basedon the morphological studies of Crissman et al. (3, 4). Usingelutriated B16a cells, significant tumor cell-platelet interactionwas observed 10 min following tail vein injection and wasmaximal at 8 h post-tumor cell injection (3). Morphometricanalysis and quantitation of tumor cell-platelet interaction wasperformed by light microscopic examination of 1-^m sectionsand confirmed by ultrastructural examination as described in"Materials and Methods." B16a cells arrested in the microvas

culature were identified by their large size and abundant cytoplasm (Fig. 2) as described by Crissman et al. (3, 4). Fig. 2compares an ethanol-treated control cell with a cell treated withcycloheximide. At 10 min postinjection, morphological examination revealed large platelet aggregates associated with intra-vascular tumor cells in all control groups (Fig. 2a) and thecolchicine-treated group 8 h postinjection (data not shown).However, the cycloheximide-treated group showed numerousintravascular tumor cells within the pulmonary vessels, surrounded by occasional erythrocytes, but without evidence ofassociated platelet aggregates (Fig. 1b). Light microscopic examination of lungs in all control and experimental groups at 8h postinjection demonstrated numerous platelet aggregates indirect association with tumor cells. Morphologically, no qualitative or quantitative differences were observed between individual groups (data not shown). Quantitative analysis revealedthat 90 to 95% of control cells were associated with plateletaggregates in both the 10-min and 8-h groups (Fig. 3, a and b).Similarly, about 95% of colchicine-treated cells were associatedwith platelet aggregates at 10 min and 8 h postinjection (Fig.3a). In contrast, we observed a significant (P< 0.0001) decreasein the number of cycloheximide-treated cells associated withplatelet aggregates at 10 min postinjection (Fig. 3¿>).However,at the 8-h interval, there were no significant differences withrespect to the platelet association with cycloheximide-, colchicine- or solvent-treated B16a cells (Fig. 3, a and ¿>),suggestingthat the effect of cycloheximide on the integrity of B16a vimen-tin intermediate filaments and tumor cell-platelet aggregatingability is reversible (see below).

Analysis of B16a cell distribution at 10 min (measured ascells/mm2) revealed no difference among solvent controls andcolchicine- or cycloheximide-treated cells (Fig. 3, c and d),

indicating a similar pattern of initial arrest in the microvasculature. Eight h following the arrest, the number of cells/mm2

in both control groups decreased from the number observed at10 min. However, there was no difference between controlgroups at 8 h (Fig. 3, c and d). In contrast, at 8 h the number

platelets are clearly not activated, as indicated by their round profiles and intactsecretory granules. (*). a, x 9,300; b. x 21,818.

7691

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

;:;

r

Fig. 7. Eight h postinjeclion. control B16a cells (arrow in a) were observed within the microvasculature in contact with platelet aggregates (*). At this time interval,both cyclohcximidc (!>)-and colchicine (arrow in c)-treated B16a cells were observed within the microvasculature associated with aggregates of activated platelets (*).A colchieine-treated BI6a cell. 8 h postinjection, is in direct contact with the subendothelial matrix (arrows) following endothclial cell retraction (tl). Similarly, controland cycloheximide-treated BI6a cells were also observed in contact with the subendothelial matrix 8 h postinjection (not shown), a. ~x8,374; ft. x 5.962; c. 7.040; d,

23.225.

7692

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

Fig. 8. Immunofluorescence studies demonstrate the effect of cycloheximidetreatment on disruption of B16a intermediate filaments. In control cells, a well-defined network of vimentin-containing intermediate filaments is observed (a,arrows). Cycloheximide treatment disrupts the intermediate filaments without achange in fluorescence intensity (ft). This effect of cycloheximide on intermediatefilaments was reversible, as evidenced by a reorganization of the vimentin intermediate filaments, when B16a cells were allowed to recover for 8 h (c, arrows), a,X 1,033; A, x 840; c. x 828.

of cells/mm2 in the cycloheximide-treated group was significantly (P < 0.02) lower than the number of cells/mm2 in thecolchicine-treated group (Fig. 3, c and d).

In addition to direct morphological examination, circulatingTXB2 levels were used as another indicator of the extent of invivo platelet activation. Thromboxane B2 is the hydrolysis product of the short-lived (t,/i = 30 s) thromboxane A2, the majorcyclooxygenase metabolite of arachidonic acid produced byagonist-stimulated platelets (29), including tumor cell agonists(30). Serum thromboxane B2 levels in animals given injectionsof cycloheximide-treated cells were significantly lower thanthose of animals given injections of solvent-treated B16a cellsat 10 min (Fig. 4). However, no difference was observed at 8 h(Fig. 4), consistent with the morphological data presented above(Fig. 3b).

Ultrastructural Observations. Ultrastructural examination oflung tissue removed 10 min postinjection revealed control cells

100

90-

80-

70-

60-

1 40"

I 30-

3 20-

* 10-

0

Ctl

-5 5 10 15 20

time (min)

100-

90-

80-

70-

60-

30-

f& 20-

*io-

0-

cx

B

T~I—I I I0 5

l—I—i—r

10

time (min)

~i—r20

Fig. 9. In /i, treatment of B16a cells with cycloheximide (C.V) significantlydecreased their ability to aggregate homologous platelets when compared withcontrol (ctl). In B, following an 8-h recovery' from CX treatment, BI6a cellspartially regained their ability to aggregate homologous platelets.

arrested in the microvasculature with their plasma membranesintimately juxtaposed to endothelial cell plasma membranes(Fig. 5a). A focal aggregate of activated platelets was in contactwith a limited portion of the B16a cell plasma membrane, thelatter of which formed processes projecting into the plateletaggregate (Fig. 5b). Similarly, colchicine-treated B16a cells

were also juxtaposed to the endothelial cell plasma membraneand demonstrated focal platelet aggregates (Fig. 5c) and cellmembrane process formation (Fig. 5d).

Ultrastructurally, contact between the plasma membrane ofcycloheximide-treated B16a cells and the plasma membrane ofendothelial cells was similar to controls (Fig. 6a). However,platelet activation was absent, even with platelets which appeared to be in contact with the tumor cell plasma membrane(Fig. 6b). Endothelial cell retraction was evident 8 h postinjection of control and colchicine- or cycloheximide-treated B16acells; this time interval also demonstrated B16a cell contactwith the subendothelial cell matrix (Fig. la). Focal aggregatesof activated, degranulated platelets were observed in contactwith cells from all three groups (i.e., control and colchicine orcycloheximide treated) at 8 h postinjection as were tumor cellprocesses (Fig. la).

Reversibility of the Cycloheximide Effect. Ben-Ze'ev et al.

(14) demonstrated that the effect of cycloheximide on disruption of B16-F1 melanoma cell vimentin intermediate filaments

was reversible. In vivo B16a cells regain their ability to activateplatelets 8 h following arrest (this study). We have demon-

7693

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

65—60

—

55—50

—

45—40

—35

—30

^25^20

—15

—10

—5

—TTTTCTL

CX CTL CX

118H

RECOVERY

Fig. 10. H Uvi of disrupting B16a vimentin intermediate filaments (with andwithout recovery) on their adhesion to CD3 endothclial cells at 10 min. CTL,control; CX, cycloheximide.

strated that intact intermediate filaments are necessary fornormal TCIPA (Ref. 16; Footnote 4). Therefore vimentin intermediate filaments of B16a cells were examined followingtreatment with cycloheximide (50 ¿¿mol,8 h, 37°C)and removal

of cycloheximide followed by an 8-h recovery period. The 8 hrecovery period was chosen to coincide with the time intervalused in the in vivo studies. As previously demonstrated,4 B16a

cells exhibit a distinct intermediate filament network as visualized by indirect immunofluorescence (Fig. 8a). Cycloheximidetreatment disrupted B16a intermediate filaments (Fig. 8¿>)which partially reformed 8 h following removal of cycloheximide (Fig. 8c). Immediately following treatment, cyclohexim-ide-treated cells failed to induce platelet aggregation (Fig. 9a).However, following an 8-h recovery from cycloheximide, B16acells aggregated homologous platelets to approximately 70% ofthat observed with untreated B16a cells (Fig. 9b).

Effect of Cycloheximide on Tumor Cell Attachment to Endo-thelial Cells. Morphometric data indicated that 10 min following the intravascular arrest, an equivalent number of tumorcells/mm2 were present in all groups in the lung microvascula-

ture (Fig. 2, c and d). Since lung colony formation is decreasedfollowing disruption of intermediate filaments, but not following disruption of microtubules, this suggested that initial tumorcell arrest and attachment to endothelial cells may be independent of intermediate filaments. To test the hypothesis we usedan in vivo adhesion assay with lung microvessel endothelialcells derived from the C57BL/6J mouse and cycloheximide-treated B16a cells immediately following treatment and following an 8-h recovery period (see "Material and Methods"). The

adhesion assays were conducted for 10 min before nonadherentcells were removed. This time interval was chosen to coincidewith the time used in the in vivo studies to evaluate the initialB16a cell arrest in the microvasculature. After a 10-min adhe

sion period, no difference in the number of adherent cells wasfound between untreated and cycloheximide-treated B16a cells(Fig. 10). However, if the cells were allowed to adhere for 45min before being washed and counted, a significant decrease inthe number of adherent cells was observed in the cycloheximide-treated group, when compared with the untreated controls (datanot shown). Colchicine treatment did not alter tumor celladhesion in either a 10-min or 45-min adhesion assay (data notshown). When B16a cells were allowed to recover (8 h) fromcycloheximide treatment, there was no difference in the numberof adherent cells when compared with controls following a 10-min (Fig. 10) or 45-min adhesion assay (data not shown).

DISCUSSION

The tumor cell cytoskeleton is proposed to play a role inmetastasis by mediating cell migration (11), adhesion (12, 31),invasion (15), and cell shape change (31). Recently, Chopra etal. (Ref. 16; Footnote 4) presented in vitro evidence demonstrating that disruption of microfilaments or intermediate filamentsinhibits macroscopic TCIPA and platelet adhesion to theplasma membrane of W256 cells and B16a cells. This adhesionresponse appears to be mediated by a tumor cell integrin receptor (i.e., IRGpIIb/IIa) (Ref. 6; Footnote 5) that is immunolog-ically related to the platelet glycoprotein III) Ila (i.e., «¡»A).Pretreatment of tumor cells with monoclonal or polyclonalantibodies which react against IRGpIIb/IIIa inhibits macroscopic TCIPA and platelet adhesion to the plasma membrane(Refs. 16 and 32; Footnotes 4 and 5), suggesting that thisreceptor mediates tumor cell-platelet interaction. Light fixationof W256 cells with paraformaldehyde inhibits plasma membrane receptor mobility (including IRGpIIb/IIIa) without altering membrane fluidity (16). This treatment also results in aninhibition of TCIPA and adhesion of platelet aggregates to thetumor cell plasma membrane (16). Similarly, disruption ofmicrofilaments or intermediate filaments also inhibitsIRGpIIb/IIa receptor mobility, concomitant with an inhibitionof TCIPA (16). Taken collectively, the above results suggestthat the loss of platelet-aggregating ability following disruptionof certain cytoskeletal elements is due in part to a loss ofIRGpIIb/IIIa mobility and/or the loss of the ability of thereceptor to bind adhesion protein ligands.

Evidence in the literature suggests that platelets are important during the arrest and attachment phase of the metastaticcascade. The studies by Crissman and coworkers (3, 4, 19)demonstrated that: (a) initial tumor cell arrest involves thejuxtaposition of the tumor cell-endothelial cell plasma membranes; (b) platelet activation and their association with intravascular tumor cells probably occur postarrest; (c) plateletactivation proceeds over several hours postarrest and coincideswith the initiation of endothelial cell retraction; and (d) 24 hpostarrest, many tumor cells remain intravascular and adherentto the subendothelial matrix with an absence of associatedplatelet thrombi. In general, this sequence of events is reportedby other investigators ( 17, 33, 34). These morphological studiesindicate that the first 24 h postarrest constitute the critical timeinterval for platelet involvement in metastasis. Although theexact mechanism(s) for platelet-enhanced metastasis is unknown, platelets have been demonstrated to enhance tumor cell

' K. V. Honn, I. M. Grossi, K. K. Nelson, L. A. Umbarger, L. A. Fitzgerald,J. S. Hatfield, S. E. G. Fligiel, B. W. Steinert. C. A. Diglio. J. D. Taylor, and J.M. Onoda. Correlation among expression of integrin i»nbfi3,tumor cell-inducedplatelet aggregation, tumor cell adhesion, and lung colonization in elutriatedsubpopulations of the B16a amelanotic melanoma and Lewis lung carcinoma,submitted for publication.

7694

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

adhesion to endothelium (25, 35, 36) and induce reversibleendothelial cell retraction (20), thereby temporarily exposingthe subendothelial matrix to tumor cell adhesion and proteol-ysis (20). Therefore it seems reasonable to propose that the lossof platelet-aggregating ability is, in part, responsible for decreased lung colonization following disruption of certain cyto-skeletal elements.

Colchicine disruption of microtubules does not inhibit tumorcell-platelet interaction in vitro4 or in vivo (this study) and does

not reduce lung colony formation (this study). However, disruption of microfilaments or intermediate filaments inhibits lungcolon formation as well as tumor cell-platelet interaction invitro1 and in vivo (this study). These results are in partial

agreement with those of Hart et al. (12) in that we observed adecrease in lung colonization following disruption of microfilaments. However, we could not reproduce the previously reported decrease in lung colonization and adhesion to endothelial cells following microtubule disruption with colchicine (12).As mentioned previously, colchicine treatment can, in some celllines, result in a secondary disruption of intermediate filaments(38, 39), which may account for the discrepancy between ourresults and those of Hart et al. (12). Our results are in agreementwith those of Ben-Ze'ev and Raz (14) who observed that dis

ruption of vimentin intermediate filaments reversibly reducedlung colonization. However, since cycloheximide, a proteinsynthesis inhibitor, was used to disrupt intermediate filaments,we cannot totally exclude other mechanisms.

When [125I]iododeoxyuridine-radiolabeled B16 cells were

treated with colchicine, cytochalasin B, or a combination ofboth and tail vein injected, it was observed that initial pulmonary retention, measured at 2 min postinjection, was unchangedfrom that of controls (12). However, at 1 h postinjection, fewercolchicine- or cytochalasin B-treated cells remained in the lung(12), suggesting that cytoskeletal disruption does not affectinitial arrest in the microvasculature and possibly does notaffect initial adhesion to endothelium. The results of the presentstudy support that hypothesis. First, we did not observe adifference in the number of Bl 6 cells/mm2 of lung parenchyma

among the control and treated groups 10 min following injection, whereas a significant decrease was observed in the cyclo-heximide-treated group as compared with the colchicine-treatedgroup at 8 h postinjection. Second, neither colchicine norcycloheximide affected initial (i.e., 10 min) tumor cell adhesionto pulmonary microvessel-derived endothelial cells in vivo (thisstudy; Footnote 6). However, in a longer (i.e., 45 min) adhesionassay there were fewer cycloheximide-treated cells adherent toendothelial cells, while the colchicine-treated group remainedunchanged from controls.6 The cycloheximide results suggest

that different receptor mechanisms may be responsible forinitial and late stage tumor cell adhesion to endothelium andthat the former is independent of cytoskeletal control. Wereported previously that (a) tumor cell adhesion to endotheliumcan be stimulated by a number of factors, i.e., platelets (35),eicosanoids (37), thrombin (40), etc.; (b) this stimulated adhesion is mediated in part by the integrin receptor IRGpIIb/IIIa(37); and (c) stimulated adhesion is dependent upon receptormobility mediated by intermediate filaments and microfila-ments (Ref. 41; Footnote 7). This study now demonstrates thatthe initial basal adhesion to endothelial cells is independent ofIRGpIIb/IIIa (37) and cytoskeletal involvement.

' Unpublished observations.7 H. Chopra, J. Timar. X. Rong, L. A. Fitzgerald. J. D. Taylor, and K. V.

Honn. 12-(S)-HETE induces an actin and vimentin-dependent increase in integrin,n, .•' surface expression on melanoma cells, submitted for publication.

The role of the tumor cell cytoskeleton in metastasis is notyet completely understood. Previous studies suggest that thetumor cell cytoskeleton may play a role in metastasis by modulating tumor cell motility and interaction with endothelialcells ( 10). While support for those hypotheses exists, the presentstudy suggests that the tumor cell cytoskeleton may modulateyet another event (i.e., TCIPA), which may play an importantrole in tumor cell arrest and adhesion to the vessel wall duringmetastasis. TCIPA is a cytoskeleton-dependent, receptor-mediated phenomenon (Refs. 16 and 42; Footnote 4). However,the exact biochemical mechanisms involved in this process areas yet unknown and remain to be elucidated.

ACKNOWLEDGMENTS

The authors thank M. Banconi, J. Frenchik, S. Kotsonis, S. Weis-man, and J. Salem for assistance with the in vivo experiments.

REFERENCES

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

Weiss, L., Orr, F. W., and Honn, K. V. Interactions of cancer cells with themicrovasculature during metastasis. FASEB J., 2: 12-21, 1988.Weiss, L.. Orr, F. W., and Honn, K. V. Interactions between cancer cells andthe microvasculature. Clin. Exp. Metastasis, 7: 127-167, 1989.Crissman, D., Hatfield, J., Schaldenbrand. M., Sloane. B. F., and Honn, K.V. Arrest and extravasation of B16 amelanotic melanoma in murine lungs:a light and electron microscopic study. Lab. Invest., 53: 470-478, 1985.Crissman, J. D., Hatfield, J. S.. Menter, D. G., Sloane, B. F., and Honn, K.V. Morphological study of the interaction of intravascular tumor cells withendothelial cells and subendothelial matrix. Cancer Res., 48: 4065-4072,1988.Nicolson, G. L. Cancer metastasis: tumor cell and host organ propertiesimportant in metastasis to specific secondary sites. Biochim. Biophys. Acta,948: 175-224. 1988.Grossi, I. M.. Hatfield, J. S., Fitzgerald. L. A., Newcombe, M., Taylor, J.D.. and Honn, K. V. Role of tumor cell glycoproteins immunologicallyrelated to glycoproteins Ib and Hb/IIIa in tumor cell-platelet and tumor cellmatrix interactions. FASEB J., 2: 2385-2395, 1988.Sloane, B. F.. Dunn. J. R., and Honn, K. V. Lysosomal cathepsin B:correlation with metastatic potential. Science (Wash. DC). 212: 1151-1153,1981.Sloane, B. F., Rozhin. J. Lah, T. T., Day. N. A.. Buck, M., Ryan. R. E.,Crissman. J. D.. and Honn. K. V. Tumor cathepsin B and its endogenousinhibitors in metastasis. In: G. Prodi, L. A. Liotta. P. L. Lollini, S. Garbisa,S. Gorini, and K. Hellman (eds.). Cancer Metastasis. Biological and Biochemical Mechanisms and Clinical Aspects, pp. 259-268. New York: PlenumPress, 1988.Sloane. B. F., Ryan. R. E., Rozhin, J.. Lah. T. T.. Crissman, J. D.. andHonn. K. V. Cathepsin B: a proteinase linked to metastasis? In: H. Tschesche(ed.), Proteinases in Inflammation and Tumor Invasion, pp. 93-106. Berlin:Walter de Gruyter & Co., 1986.Raz, A., and Ben-Ze'ev, A. Cell-contract and - architecture of malignant cellsand their relationship to metastasis. Cancer Metastasis Rev., 6: 3-21, 1987.Hagmar, B.. and Ryd, W. Tumor cell locomotion: a factor in metastasisformation? Influence of cytochalasin B on a tumor dissemination pattern.Ing. J. Cancer. 19: 576-580, 1977.Hart. I. R.. Raz. A., and Fidler, J. 1. Effect of cytoskeleton disrupting agentson the metastatic behavior of melanoma cells. J. Nati. Cancer Inst.. 64: 891-899. 1980.Raz, A., and Geiger, B. Altered organization of cell-substrate contacts andmembrane-associated cytoskeleton in tumor cell variants exhibiting differentmetastatic capabilities. Cancer Res., 42: 5183-5190. 1982.Ben-Ze'ev. A., and Raz, A. Relationship between the organization and

synthesis of vimentin and the metastatic capability of BI6 melanoma cells.Cancer Res., 45: 2632-2641, 1985.Mareel. M. M., and De Mets, M. Effect of microtubule inhibitors on invasionand on related activities of tumor cells. Int. Rev. Cytol., 90: 125-168, 1984.Chopra. H., Hatfield, J. S., Chang, S. Y., Grossi, I. M., Fitzgerald, L. A.,O'Gara, C. Y., Marne«, L. J., Diglio, C. A., Taylor, J. D., and Honn. K. V.

Role of tumor cell cytoskeleton and membrane glycoprotein IRGpIIb/IIIa inplatelet adhesion to tumor cell plasma membrane and tumor cell inducedplatelet aggregation. Cancer Res., 48: 3787-3800. 1988.Sindelar, W. F., Tralka, T. S., and Ketcham, A. S. Electron microscopicobservations on formation of pulmonary metastasis. J. Surg. Res., IS: 137-161. 1975.Menter, D. G., Sloane. B. F., Steinen, B. W.. Onoda, J., Craig, R., Harkins,C.. Taylor. J. D.. and Honn, K. V. Platelet enhancement of tumor celladhesion to subendothelial matrix. Role of platelet cytoskeleton and plateletmembrane. J. Nati. Cancer Inst., 79: 1077-1090. 1987.Menter. D. G.. Hatfield. J. S.. Harkins. C. J., Sloane. B. F.. Taylor. J. D.,Crissman. J. D., and Honn, K. V. Tumor cell-platelet interactions in vitro

7695

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TUMOR CELL CYTOSKELETON AND METASTASIS

and their relationship to in rivo arrest of hematogenously circulating tumorcells. Clin. Exp. Metastasis, 5: 65-78, 1987.

20. Honn, K. V., Grossi. I. M., Diglio, C. A., Wojtukiewiez, M., and Taylor. J.D. Enhanced tumor cell adhesion to the subendothelial cell matrix resultingfrom 12(S)-HETE induced endothelial cell retraction. FASEB J.. 3: 2285-2293, 1989.

21. Sloane. B. F., Honn, K. V., Sadler, J. G.. Turner. W. A., Kimpson, J. J.. andTaylor, J. D. Cathepsin B activity in HI CMmelanoma cell: a possible markerfor metastatic potential. Cancer Res., 42: 980-986. 1982.

22. Brown, S. s.. and Spudich, J. A. Cytochalasin inhibits the rate of elongationof actin fragments. J. Cell Biol., «J:657-662, 1979.

23. Margolis. R. L.. and Wilson. L. Addition of colchicine-tubulin complex tomicrotubule ends: the mechanism of substoichiometric colchicinc poisoning.Proc. Nati. Acad. Sci. USA, 74: 3466-3470, 1977.

24. Sharpe, A. H., Chen, L. B., Murphy, J. R., and Fields. B. N. Specificdisruption of vimentin filament organization in monkey kidney CV-1 cellsby diptheria toxin, endotoxin, and cycloheximide. Proc. Nati. Acad. Sci.USA, 77:7267-7271, 1981.

25. Menter, D. G., Sieinert. B. W., Sloane. B. F.. Taylor. J. D., and Honn. K.V. A new in vitro model for investigation of tumor cell-platelet-endothelialcell interaction and concomitant eicosanoid biosynthesis. Cancer Res., 47:2425-2432. 1987.

26. Honn. K. V. Inhibition of tumor cell metastasis by modulation of the vascularprostacyclin/thromboxane A¡system. Clin. Exp. Metastasis. /: 103-114,1983.

27. Diglio, C. A., Grammas. P., Giacomelli. F., and Weiner. J. Primary cultureof rat cerebral microvascular endothelial cells. Lab. Invest., 46: 554-563,1982.

28. Diglio, C. A., Grammas, P., Giacomelli, F., and Weiner, J. Rat-heart derivedendothelial and smooth muscle cultures: isolation, cloning, and characterization. Tissue & Cell, 20: 477-492. 1988.

29. Hamberg. M.. Svensson, J., and Samuelsson, B. Thromboxanes: a new groupof biologically active compounds derived from prostaglandin endoperoxides.Proc. Nati. Acad. Sci. USA, 72: 2994-2988, 1975.

30. Honn. K. V.. Steinert, B. W.. Moin, K.. Onoda. J. M., Taylor, J. D., andSloane. B. F. The role of platelet cyclooxygenase and lipoxygenase pathwaysin tumor cell induced platelet aggregation. Biochem. Biophys. Res. Commun.. 145: 384-389. 1987.

31. Raz. A.. Zoller, M., and Ben-Ze'ev, A. Cell configuration and adhesiveproperties of metastasizing and non-metastasizing BSp73 rat adenocarci-noma cells. Exp. Cell Res., 162: 127-141, 1986.

32. Boukcrchc, H.. Berthier-Vergnes, O., Tabonc, M.. Dorè,J., Leung, L. L. K.,and McGregor, J. L. Platelet-melanoma cell interaction is mediated by theglycoprotein Ilb-Illa complex. Blood, 74:658-663, 1989.

33. Tanaka, N. G.. Tohgo. A., and Ogawa. H. Platelet-aggregating activities ofmetaslasizing tumor cells. V. In situ roles of platelets in hcmatogenousmétastases.Invasion Metastasis, 6: 209-224. 1986.

34. Chew, E. C., Josephson, R. L., and Wallace. A. C. Morphologic aspects ofthe arrest of circulating cancer cells. In: L. Weiss (eds.). Fundamental Aspectsof Metastasis, pp. 121-150. Amsterdam: North Holland. 1975.

35. Honn, K. V.. Onoda, J.. Pampalona, K., Battaglia. M., Neagos, G.. Taylor,J. D., Diglio, C. A., and Sloane, B. F. Inhibition by dihydropyridine classcalcium channel blockers of tumor cell platelet-cndothclial cell interactionsin vitro and metastasis in vivo. Biochem. Pharmacol.. 34: 235-241, 1985.

36. Abecassis, J., Millon-Collard, R., Klein-Soyer, C., Nicora, F.. Fricker, J.,Beret/, A., Eber, M., Muller, D., and Cazenave. J. Adhesion of breast cancercell line MCF-7 to human vascular endothelial cells in culture. Enhancementby activated platelets. Int. J. Cancer, 40: 525-531, 1987.

37. Grossi, I. M., Fitzgerald, L. A., Umbarger, L. A.. Nelson, K. K., Diglio,C.A., Taylor, J. D., and Honn, K. V. Bidirectional control of membraneexpression and/or activation of the tumor cell IRGpllb/IIla receptor andtumor cell adhesion by lipoxygenase products of arachidonic acid and linoleicacid. Cancer Res., 49: 1029-1037, 1989.

38. Henderson, D., and Weber. K. Immunoelectron microscopial identificationof the two types intermediate filaments in established epithelial cells. Exp.Cell Res.. 132: 297-311, 1981.

39. Hynes. R. O., and Désirée,A. T. lOnm filaments in normal and transformedcells. Cell, 13: 151-163, 1978.

40. Wojtukiewiez, M. Z., Ciarelli, J. J., Walz, D. A., and Honn, K. V. Thrombinenhances expression of an integrin receptor and increases adhesion. Proc.Am. Assoc. Cancer Res., 31: 80, 1990.

41. Chopra, H., Marzouq, L., Taylor. J. D., and Honn. K. V. The lipoxygenaseproduct I2(.V)-HETE alters the organization of the tumor cell cytoskeletonand modulates intcgrin receptor function. Proc. Am. Assoc. Cancer Res., 30:88. 1989.

42. Chopra, H., Hatfield. J. S.. Fligiel, S. E. G., Nelson, K. K., Diglio. C. A..Taylor. J. D., and Honn. K. V. The role of the tumor cell cytoskeleton intumor cell interactions with endothelial cells and platelets. Proc. Am. Assoc.Cancer Res., 31:62. 1990.

43. Menter. D. G.. Harkins, C., Onoda, J.. Riordcn. W'.. Sloane. B. F., Taylor.J. D.. and Honn. K. V. Inhibition of tumor cell induced platelet aggregationby prostacyclin and carbacyclin: an ultrastructural study. Invasion & Metastasis. 7: 109-128. 1987.

7696

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1990;50:7686-7696. Cancer Res   Hemi Chopra, Suzanne E. G. Fligiel, James S. Hatfield, et al.   Tumor Cell-Platelet-Endothelial Cell Interactions

Study of the Role of the Tumor Cell Cytoskeleton inin VivoAn

  Updated version

  http://cancerres.aacrjournals.org/content/50/23/7686

Access the most recent version of this article at:

   

   

   

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

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/50/23/7686To request permission to re-use all or part of this article, use this link

on May 2, 2021. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from