[CANCER RESEARCH 47, 1398-1406, March 1, 1987]

An Animal Model for Colon Cancer Metastasis: Establishment andCharacterization of Murine Cell Lines with EnhancedLiver-metastasizing Ability1

Robert S. Bresalier,2 Erkki S. Hujanen, Steven E. Raper, F. Joseph Roll, Steven H. Itzkowitz, George R. Martin, and

Young S. KimGaslroinleslinal Research [R. S. B., S. H. I., Y. S. KJ, Cell Biology [S. E. R], and Liver Center [F. J. R.] Laboratories, Veterans Administration Medical Center;Departments of Medicine ¡R.S. B., F. J. R., S. H. I., Y. S. KJ, Surgery ¡S.E. R.], and Pathology [Y. S. K.J, the University of California, San Francisco, California94121; and Laboratory of Developmental Biology and Anomalies [E. S. H., C. R. MJ, National Institute of Dental Research, NIH, Bethesda, Maryland 20892

ABSTRACT

A detailed understanding of the pathogenesis of colon cancer metastasis has been hindered by the lack of appropriate animal models whichaccurately reflect events in this complex process. An animal model forcolon cancer metastasis is described in which spontaneously metastasiz-

ing colonie tumors are formed after injection of murine colon cancer cellsinto the cecal wall of BALB/c mice. Using this model, tumor cells withdifferent liver-metastasizing potential were selected and shown to possess

several properties known to be associated with other metastatic cell lines.The ability of tumor cells to invade a reconstituted basement membraneand to secrete type IV collagenase was directly proportional to theirmetastatic ability. In addition, liver-metastasizing cells preferentially

migrated toward liver extracts in a Boyden chamber assay, as comparedto extracts of brain or lung, and adhered rapidly to highly purified hepaticsinusoidal endothelial cells versus hepatic parenchyma! cells in vitro.This model may thus be useful for studying many aspects of the pathogenesis of colon cancer metastasis.

INTRODUCTION

Most cancers are potentially lethal because of the ability ofmalignant cells to disseminate from the primary tumor andproliferate at distant sites (1, 2). Tumor cell subpopulationswith different metastatic abilities exist within the same primarytumor (3-5), and métastasesresult from the nonrandom, selective dissemination of those tumor cells possessing the ability toinvade lymphatic and vascular structures, embolize in the bloodstream, survive interactions with blood components and theimmune system, and be transported to distant organs wherethey adhere to target structures, extravasate, and establishsecondary tumor foci (1-5). Methods for identifying those cellsmost likely to metastasize and understanding the biologicalproperties which allow these cells to complete the process is ahigh priority of cancer research. Suitable models for cancermetastasis are necessary for such detailed analysis of this complex multistep process (1). Several experimental models usinga number of metastatic variants and clones have been studiedinvolving a variety of established cell lines and even primarytumors, particularly sarcomas (1,6, 7). Metastatic models forcarcinomas have been more limited and most often utilizemurine carcinomas injected s.c. or by i.v. tail vein injection intosyngeneic animals (1, 8-10). Some human tumors can metastasize in athymic nude mice after s.c., footpad, or i.v. injection(11-13), with métastasesoften limited to lymph nodes or lungs.While these models may be appropriate for studying tumorssuch as melanoma, lymphoma, various sarcomas, and perhaps

Received 6/27/86; revised 11/11/86; accepted 12/1/86.The costs of publication of this article were defrayed in part 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 work was supported by USPHS Grant CA14905 from the NationalCancer Institute, by Grant PDT-293 from the American Cancer Society, and bythe Research Service of the Veterans Administration. Presented in part at theannual meeting of the American Gastroenterological Association in New YorkCity, New York, May 1985.

*To whom requests for reprints should be addressed, at G. I. Research Lab(151M2), VA Medical Center, 4150 Clement St., San Francisco, CA 94121.

even mammary carcinoma, in the case of colon cancer thesesystems may bypass many of the complex processes of themetastatic cascade.

While hepatic métastasesare a frequent event in patientswith colon cancer, they rarely occur in existing animal modelsof colonie neoplasia. Melanoma cell lines have been developedwhich preferentially metastasize to the liver after intraportal ori.v. injection (14), and hepatic and pulmonary nodules havebeen reported in syngeneic C57BL/6 mice or in preconditionednude mice given injections intrasplenically of murine (15) orhuman (16, 17) cell lines, respectively. It has been suggestedthat the establishment of experimental primary tumors in thecolon after colonie wall injection may provide an accurate modelfor studying colon cancer mestastasis after both syngeneic (18-20) and heterotopic (20, 21) implantation. In this paper wedescribe our experience with an animal model in which murinecolon cancer cells reproducibly metastasize to the liver of syngeneic animals via normal routes in a defined period of time.This model was used to establish colon cancer cell lines withhigh liver-metastasizing abilities, and these cells have beenpartially characterized with regard to biological propertiesknown to be associated with the invasiveness of other metastaticcell lines.

MATERIALS AND METHODS

Cell Lines

Parental cell line 51B was established in this laboratory from trans-phiniable murine colon carcinoma 51. The original transplantabletumor, established by Dr. T. H. Corbe«(22) by repeated s.c. injectionwith 1,2 dimethylhydrazine, forms grade 11-111carcinomas when passaged in syngeneic animals. Unlike the original transplantable tumorwhich was reponed to metastasize to regional lymph nodes and lungsfrom the s.c. site in greater than 80% of animals tested, 51B has lowermetastatic ability when injected s.c. into syngeneic BALB/c mice (seebelow). Subsequent metastatic derivatives of cell line 51B were established as described below.

Preparation of Cell Lines for InjectionAll cell lines were grown and maintained in DM1M ' supplemented

with 10% fetal bovine serum, penicillin (100 units/ml), streptomycin(100 Mg/ml), and a 7% COi environment. Early passage cell lines(passagedfewerthan 15times) wereused for all studies unless otherwisenoted. Confluent cultures were harvested by brief trypsinization (0.05%trypsin-0.02% EDTA in Hanks' balanced salt solution without calciumand magnesium), washed several times with CMF, and resuspended ata final concentration of 5 x IO7cells/ml in serum-free medium. Thepresence of single cell suspensions was continued by phase contrastmicroscopy,and cell viabilitywas determined by trypan blue exclusion.

Cecal Injection

BALB/c mice (15- to 18-g females obtained from Simonsen Laboratories, Gilroy, CA) were anaesthetized with methoxyfluorane by

3The abbreviations used are: DMEM, Dulbecco's modified Eagle's minimalessential medium; CMF, calcium-magnesium-free phosphate-buffered saline;BSA, bovine serum albumin.

1398

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

inhalation, and the abdomen was prepared in a sterile fashion. A smallabdominal incision was made, and the cecum was identified and isolatedbetween sterile gauze; 5x10' viable tumor cells in 0.1 ml serum-free

medium were injected into the cecal wall from the serosal side using asterile tuberculin syringe and slightly curved 27-gauge needle. Frequenttrituration of cells was performed in a sterile syringe in order tomaintain uniform cell suspensions and to prevent clumping. Cells wereinjected so as to visibly infiltrate between the submucosal and subserosaltissues. The serosal surface at the injection site was dabbed gently withsterile gauze dampened with 95% ethanol in an attempt to kill freetumor cells and the organs replaced in situ. The abdomen was closedwith continuous nylon sutures, and the animal was returned to its cage.After 4 weeks, animals were sacrificed by cervical dislocation and theabdominal organs and thorax examined for the presence of macroscopic"primary" cecal tumors and métastases.Pilot studies demonstrated that

at this time cecal tumors of approximately 0.5 cm were present withoutperitoneal spread of tumor. Organs including cecum, liver, mesentericlymph nodes, and lungs were divided, and paired specimens wereprocessed for histológica!examination and tissue culture.

Histológica!Examination and Scoring of Métastases

Specimens for histológica!examination were fixed in 10% formalinfor 24 h. The entire liver of each animal was sectioned at 0.2-cmintervals, the cut surfaces examined, and the specimens were embeddedin paraffin. Representative sections of cecum, lymph nodes, and lungswere also cut and embedded in paraffin. Five urn sections were then cutand stained with hematoxylin and eosin and periodic acid-Schiff toconfirm the presence of tumor and examined for microscopic métastases by light microscopy.

Macroscopic métastaseswere defined as those visible to the nakedeye. All macroscopic métastaseswere verified by histological examination. Microscopic métastaseswere those visible on histological examination by light microscopy of serial hematoxylin-eosin-stained tissuesections.

Electron Microscopy

Tissue sections of liver were immersed for 2 h in a fixature consistingof 2.5% glutaraldehyde-0.8% paraformaldehyde in 0.2 M sodium bicarbonate buffer. Tissues were osmicated with 1% osmium tetroxidecontaining 1.5% potassium cyanide for 90 min, washed in 0.2 Msodiumbicarbonate buffer for 10 min, and then were dehydrated in ethanol andembedded in Epon. Sections were stained with lead citrate and uranylacetate and examined in a Philips EM 300 microscope (Philips Electronic Instruments, Inc., Mahwah, NJ).

Culture of Metastatic Foci and Establishment of Cell Lines with Enhanced Metastatic Activity

Specimens of macroscopic hepatic métastasesand liver potentiallycontaining métastaseswere washed several times in tissue culturemedium containing 100 units/ml penicillin G, 100 ng/ml streptomycin,and 2.5 ¿tg/mlfungizone. The tissue was minced finely with a sterilescalpel blade and resuspended in DMEM supplemented as above andwith 10% fetal calf serum. Various amounts of tissue were plated into25-cm2 plastic tissue culture flasks which were gassed to contain 7%CO2 and incubated as a closed system at 37°C.Visible tissue fragments

were discarded after 72 h and the medium changed. Cultures wereobserved daily for cell growth with frequent changes of medium. Fungizone was deleted after five medium changes. Colonies derived frommétastaseswere grown in culture and combined, populations wereexpanded, and new cell lines were established. Care was taken to freezelarge numbers of early passage cells for future use to minimize thepotential problem of genetic instability of passaged cells.

Development of Cell Lines with Enhanced Metastatic Ability

Cell lines with enhanced metastatic ability were selected in a manneranalogous to that originally described by Fidler (23). Cell lines derivedfrom métastasesproduced as above were reinjected into the ceca ofadditional mice, and the procedure was repeated serially in an attemptto select for tumor cells with high metastatic ability.

Invasion Assay

The invasive potential of tumor cells was determined with an in vitroinvasion assay (24). Briefly, cells were tested for their ability to penetratethrough a reconstituted basement membrane matrix which was composed of purified bovine tendon type I collagen [0.2 nig/matrix (25)],type IV collagen [0.2 mg/matrix (26)], and laminin [0.2 mg/matrix(27)] compressed to form a 200- to 600-//m thick mat (Helitrex Corp.,Princeton, NJ). The barrier was placed into a modified Boyden chamberdirectly onto a type IV collagen-coated Nuclepore filter (5 ¿igprotein/filter; pore size, 1.0 ¿im).

A suspension of tumor cells (7 x 10s) in DMEM containing 2%acid-treated NuSera (28) and 0.5% BSA was introduced into the uppercompartment and DMEM containing 0.5% BSA was placed into thelower compartment of the chamber. For chemoinvasion assays (29),the medium in the lower compartment also contained partially purifiedorgan-derived attractants (prepared as in the chemotaxis assay below).Similar extracts have been used in previous studies of tumor cellchemoinvasion and chemotaxis (24, 29). After 24-36 h of incubationat 37°Cin 95% air and 5% CO2, the barrier was separated from the

underlying Nuclepore filter. The cells that had penetrated the matrixand attached to the upper surface of the Nuclepore filter were detachedwith trypsin/EDTA (0.05/0.1%) solution and counted using an electronic counter. The data are expressed as the percentage of cells whichpenetrated the matrix ±SD.All assays were performed at least intriplicate.

Chemotaxis Assay

To assess the tendency of tumor cells to be directionally attracted toorgan-specific factors, they were studied using a modified Boydenchamber assay (30). Brain, liver, and lung tissues were removed from8- to 10-week-old C57/BL female mice, and partially purified extractswere prepared as previously described (29), with slight modification.

Briefly, tissues were washed three times in DMEM containing 50Mg/mlgentamicin. After homogenization at 4°Cin a buffer ( 1 ml buffer/1 g tissue) containing 0.05 M Tris-HCl, pH 7.4, 2 M guanidine HC1,0.5 MNaCl, 5 mMW-ethylmaleimide, 5 mM EDTA, 1 mM phenylmeth-ylsulfonylfluoride, and 50 ng/m\ gentamicin, homogenates were extracted by shaking with buffer for 24 h at 4°C.Extracts were then

centrifuged at 100,000 x g for 4 h and the supernatant fractions dialyzedagainst phosphate-buffered saline. After an ammonium sulfate precipitation (0-60%), supernatants (dialyzed versus DMEM) were tested fortheir chemotactic activity. The protein level in the extracts was determined by the method of Lowry et al. (31).

For the chemotaxis assay, polyvinylpyrrolidone-free Nucleporepolycarbonate filters (pore diameter, 8 ftm; Neuroprobe, Inc.) werecoated with type IV collagen (5 ¿¿g/fi'ter)and placed into a modifiedBoyden chamber to separate the upper and lower compartments. Tissueextracts (10% solutions) in DMEM containing 0.5% BSA were introduced into the lower compartment and tumor cells (5 x 10s, detached

with 0.1% EDTA) in DMEM containing 0.5% BSA were placed intothe upper compartment. After a 5-h incubation at 37°C,the filters were

removed and stained in Hemacolor solutions 1, 2, and 3 (Harelco,Gibbstown, NJ). Filters were then placed top side up on glass slidesand cells that had not migrated were removed with cotton swabs.Migration of cells was quantitated by counting eight randomly selectedhigh power (x 400) fields/filter. Results are expressed as the averagenumber of migrated cells per high power field. All assays were done intriplicate.

Type IV Collagenase Assay

The type IV collagen-degrading metalloprotease activity secreted byselected tumor cells was measured as previously described (32, 33).Briefly, nearly confluent tumor cell monolayers were washed threetimes with DMEM and subsequently incubated in serum-free medium(DMEM) at 37°Cwith 95% air/5% CO2. After a 24-h incubation,

medium was collected and cell debris was removed by centrifugation at2000 rpm for 5 min. Medium was concentrated 100-fold with ammonium sulfate precipitation (0-60%) followed by dialysis against 0.05 MTris-HCl, pH 7.6-0.2 M NaCl, at 4°C(34). Type IV collagen degradation was assayed using soluble [3H]proline-labeled type IV procollagen

as a substrate (35). The enzyme activity was expressed as the amountof type IV collagen degradation (cpm released/IO7 cells).

1399

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

Cell-Cell Attachment

Labeling of Tumor Cells. Tumor cells were labeled according to themethod of Fidler and Nicolson (36). Cells were grown as describedabove in DMEM supplemented with penicillin, streptomycin, and 10%fetal calf serum. Actively growing subconfluent cultures were incubatedfor 24 h with medium containing 0.3 n<Õ[I25l]iododeoxyuridine/ml

medium (specific activity, 2000 Ci/mmol; ICN Radiochemicals, Irvine,CA) which labeled nearly all cells as determined by autoradiography.

Preparation of Target Cells. Purified hepatocytes and hepatic sinusoidal endothelial cells were prepared by a modification of that previously described (37, 38) from the livers of 200- to 250-g male Sprague-Dawley rats. Cells were purified by collagenase perfusion, Stractandensity gradient centrifugation, and centrifugal elutriation. Equal numbers of the fractionated liver cells were plated into 35-mm culture dishes(Lux Scientific, New York, NY), and grown to confluency at 37°Cin a

humidified incubator under 2% CO2 in air. The purity of endothelialcells prepared by this method has been verified by electron microscopy,specific uptake of modified lipoproteins (acetoacetylated low-densitylipoprotein), and the presence of cytochemical and morphologicalmarkers of vascular endothelia (37).

Attachment Assay. Tumor cells labeled as above were washed fivetimes with CMF, harvested by very brief trypsinization and gentlemechanical dissociation, suspended in "cold" tissue culture medium,

and the number of cells determined using an electronic cell counter(Coulter Electronics, Hialeah, FL). Cell viability was greater than 95%as tested by trypan blue dye exclusion and single-cell suspensionsdocumented by phase microscopy. Target cells were grown to confluentmonolayers in 35-mm culture dishes as described above, washed withCMF, and overlayed with single-cell suspensions of 5 x 10s labeled

tumor cells in 2 ml DMEM/dish. Duplicate dishes not containingtarget cells were used as controls ("plastic"). Dishes were incubated for10 or 30 min at 37°Cin a humidified incubator under 7% CO2 in air.

At specified time points cultures were observed by phase microscopyfor attachment of tumor to target cells, and the overlying media andunattached cells were aspirated and collected. Cultures were washed x3 with CMF to remove unbound cells, and washes were added to thepreviously aspirated media. Tumor cells remaining bound to target cellswere collected separately by trypsinization and mechanical scraping.Collected material was then subjected to gamma counting (BeckmanInstruments, San Jose, CA). The percentage of adhesion was definedas

Counts bound (cells)Counts in medium + washes + counts bound (total counts) x 100

Statistical Analysis

The data were analyzed by an independent statistician. To take intoaccount the multiple comparisons for chemoinvasion and invasionassays run for 24 h (between the three cell lines and between the twomethods), pairwise comparisons between groups were done using theTukey's Studentized range test. Because of the inequality of variance

for the two cell lines compared at 36 h, / tests for unequal variancewere done for comparison between cell lines. For chemoinvasion versusinvasion at 36 h, a t test for equal variance was done. We usedBonferroni-adjusted P values for these tests to allow for multiplecomparisons. For the chemotaxis assay a comparison of means by oneway analysis of variance was shown to be significant at the 0.0001 level.Pairwise comparisons between groups were therefore performed usingthe Tukey's Studentized range test.

RESULTS

Liver-metastasizing Potential of Tumor Cells. Murine coloncancer cell line 51B formed poorly to moderately differentiatedcarcinomas without liver-metastasizing ability when injecteds.c. into BALB/c mice (0 of 20 animals). Similarly, this cellline did not form hepatic tumors when injected into the tailveins of additional mice (0 of 10 animals). On the other hand,cells derived from line 51B when implanted into the cecal wall

developed into colonie tumors (Fig. 1) which spontaneouslymetastasized to the liver in a minority (4 of 21) of the animals.Métastaseswere demonstrated by light microscopy to invadethe liver from portal tributaries, and to be present in subcapsularareas, or as intraparenchymal microfoci (Fig. 2). Electron microscopy revealed tumor cells in the hepatic sinusoids in apposition to sinusoidal endothelium and intercalated between sinusoidal endothelium and hepatic parenchymal cells (Fig. 3).

In order to determine whether tumor cell lines with progressively increased metastatic activity could be developed in ourmodel, métastasesfrom the original tumor line were culturedand used as a source of cells for successive cycling. Cells withprogressively increased metastatic potential were isolated insuccessive cycles (Table 1). Thus parental cell line 51B formedvisually apparent hepatic métastasesin 2 of 21 and microscopicmétastasesin 4 of 21 animals tested, while all animals giveninjections of the cell lines derived from the fourth and fifthmetastatic cycles developed macroscopic métastases.In addition the number of hepatic tumor nodules per animal increasedwith serial selection. This was most pronounced with cells fromthe fifth metastatic cycle. Mesenteric lymph node involvementby tumor could also be demonstrated (Fig. 4) in those animalsdeveloping liver métastases,but pulmonary métastasesfromthe cecal site were only observed in one animal.

Although 51B appeared more heterogeneous in vitro (stellatecells with multiple processes, occasional round cells with abundant cytoplasm, triangular cells) than its most metastatic derivatives (predominantly triangular cells), s.c. or cecal homograftsand métastasesformed after injection of the selected lines didnot differ histologically from those formed after injection ofthe parental cell line. Tumors were moderately to poorly differentiated with only sparse production of periodic acid-Schiff-positive mucin, and appeared similar histologically to thosereported by Corbett for transplantable tumor 51 (22). The invitro doubling times of all cell lines were similar, and s.c. andcecal homografts from the various cell lines grew with similarlatency and growth rates, suggesting that doubling time variation did not explain the difference in metastatic ability. Rather,various studies as discussed below indicate that the métastasescontain more invasive cells which show some specificity in theirresponse to liver factors.

Métastasesfrom Sites Other Than the Cecum. In a separateset of experiments additional paired groups of animals weregiven injections of cell line 51B or 51B LiM-5 either s.c. or i.v.via the tail vein. For s.c. injections, animals were given injections in the right posterior or anterior flank with 5 x IO6tumor

cells, sacrificed at 5 weeks, and organs examined for macroscopic and microscopic métastasesas described above. Of the10 animals given injections s.c. of parental cell line 5IB, 2developed spontaneous pulmonary parenchymal métastases(one macroscopic, one microscopic only), 2 additional animalshad mediastinal lymph node métastases,and none was notedto have developed liver métastases.In contrast 0 of 7 animalsgiven injections s.c. of cell line 51B LiM-5 developed pulmonary métastases,2 of 7 developed mediastinal lymph nodemétastasesand 4 of 7 had macroscopic liver métastases.Of theanimals given injections i.v. of 51B (2.5 x IO6 cells) 2 of 12

had single microscopic foci of tumor present in the liver. Incontrast, 7 of 10 animals given injections of 51B LiM-5 hadmultiple areas of macroscopic tumor present in this organ.Pulmonary métastaseswere present in both groups of animalsgiven injections i.v.

Invasion. The ability of parental cell line 51B and its metastatic derivatives 51B LiM-4 and 51B LiM-5 to penetrate areconstituted basement membrane was tested in the absence(invasion) and presence (chemoinvasion: chemotaxis plus in-

1400

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

kFig. 1. Experimental cecal "primary" tumor. BALB/c mice were sacrificed 4 weeks after injection of 5 x 10* tumor cells into the cecal wall. Tumor cells (71 are

seen in relation to normal mouse colonie mucosa. Hematoxylin-eosin, x 100.

vasion) of partially purified liver extract in the lower compartment of the Boyden chamber (Fig. 5). Experiments were performed for 24 h with the three cell lines and in addition for 36h with 51B and 51B LiM-4. More cells crossed the basementmembrane barrier at the longer time and the more metastaticlines were also more invasive. This was particularly evidentwhen a liver extract was added to the lower compartment. Theliver extract increased the invasion of all tumor cell lines. Cellline 51B LiM-4 was 3-fold and 51B LiM-5 was 5.6-fold moreinvasive and chemoinvasive at 24 h than cell line 51B studiedunder the same conditions (P < 0.001; see Fig. 5). At 36 h,both invasion and chemoinvasion by cell line 51B LiM-4 wereapproximately 6-fold greater than that of 51B under the sameconditions (P < 0.005 for invasion or chemoinvasion).

Chemotaxis. The possibility that these metastatic tumor cellswere preferentially attracted to a liver factor(s) was tested inthe Boyden chamber assay using filters which had been coatedwith type IV collagen and extracts from brain, lung, and liver(Fig. 6). Liver extract stimulated the movement of 51B LiM-5tumor cells 12-fold compared 0.5% bovine serum albuminalone, and 2.2- and 2.8-fold compared to extracts of lung orbrain, respectively (P < 0.001 for migration to the presence ofliver extract versus BSA, brain, or lung extract).

In order to further define the chemotactic movement of cellsfrom line 51B LiM-5 in response to liver factors, a Zigmond-Hirsch checkerboard analysis was performed (39) (Fig. 7).Different concentrations of liver extract were placed on bothsides of the Boyden chamber. Cells in the upper chambermigrated most actively and in a dose-dependent manner whenthe concentration of liver extract was greatest in the lowerchamber (Fig. 7, data below the diagonal). This extract alsostimulated the chemokinesis or random migration of 5IB LiM-

5 cells (Fig. 7, data on the diagonal), especially at high concentrations of extract. Little increase in cell migration was observedwhen a negative gradient of extract was established (Fig. 7, dataabove the diagonal).

Collagenase Type IV. The type IV collagen-degrading metal-loprotease activity secreted by parental cell line 51B and metastatic cell lines 51B LiM-4 and 51B LiM-5 was compared usingsoluble [3H]proline-labeled type IV procollagen as the substrate(Table 2). Cell line 51B LiM-4 produced 2.5-fold (P < 0.05)and 51B LiM-5 9-fold (P < 0.005) greater collagen degradingactivity than did parental cell line 51B.

Tumor Cell Adhesion. The ability of tumor cells from metastatic line 51B LiM-3 to bind selectively to purified hepatocytesand hepatic sinusoidal endothelial cells was tested in vitro(Table 3). Tumor cells were found to bind rapidly and preferentially to hepatic sinusoidal endothelial cells when comparedto hepatocytes or a plastic control. At 10 min tumor celladhesion to endothelial cells was more than 5-fold greater thanto hepatocytes (P < 0.001). Binding to hepatocytes differedlittle from that of control.

DISCUSSION

The study of cancer metastasis can be greatly facilitated bymodels which accurately reflect events in the metastatic process.We describe an animal model for colon cancer metastasis inwhich métastasesarise from colonie tumors which are formedin a relatively short period after injection of tumor cells intothe cecal wall of syngeneic BALB/c mice. By injecting a cellline derived from a transplantable murine colon carcinoma intothe cecal wall, we found that tumor cells can disseminate fromthe resulting "primary" tumor in a minority of animals and

1401

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

f. .\v:vcrvo>*-^.«s/ '.";: •••oy®<»*o.,- »t. •»,;

,¿-ix , .•„•'••y A -1... :áMrW>«L«to t.«

•-Ì-

Fig. 2. Liver métastasesfrom experimental cecal tiinvading the liver parenchyma from a portal tributar)400.

.: " .i A *u ofrom experimental cecal tumors. I, macroscopic liver métastases(arrowheads). X 2; fi, tumor cells similar to those of the primary turnma from a portal tributary. PV. portal vein. Hematoxylin-eosin, x 120; C intraparenchymal focus of tumor cells. Hematoxylin-eosin,

tumorx

form both micro- and macrometastases in the liver. Variousobservations discussed below suggest that this may represent apartially organ-specific metastasis.

Lines of more highly metastatic cells can be established byvarious selection procedures, including the sequential injectionof cells derived from métastasesin the classic method popularized by Fidler (23). This approach was attempted here bycultivating tumor cells which had spread to the liver, injectingthem into additional animals, and reculturing subsequent métastases. This resulted in new lines of colon cancer cells whichproduced métastasesfrom cecal sites in all animals given injec

tions. Pulmonary lesions on the other hand were very rare,occurring in only one animal with extensive hepatic métastases.

At least some metastatic cells demonstrate a common phe-notype in their interactions with basement membranes, bindingto them, secreting proteases, and invading the extracellularmatrix (33, 40-51). Some of these properties were examinedusing the colon cancer cells developed here to determinewhether for these epithelial cancer cells invasiveness and metastatic potential were correlated. Indeed, when tested against areconstituted basement membrane composed of collagen IVand laminin, their ability to penetrate it paralleled their meta-

1402

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

I

r

.

rbc

jcl H '••

I

v -H-.: -;* ' ' '&Fig. 3. Electron micrograph demonstrating tumor cells (T) in relation to hepatic sinusoidal endothelium (£)and hepatic parenchyma! cells ill}- S, sinusoid; rbc,

red blood cell; c, collagen. Arrow, apparent contact between tumor cell and endothelium. x 9600.

1403

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

Table 1 Metastatic ability of tumor cell linesThe liver-metastasizing ability of cancer cell lines was evaluated after injection

and formation of experimental cecal primary tumors. Animals were sacrificed 4weeks after cecal injections of 5 x 10* tumor cells into the cecal wall. Populationdoubling times were determined from the logarithmic portion of the growth curve.Each time point of the growth curve represented the mean of three separatecultures. Macroscopic métastaseswere those visible to the naked eye. Microscopicmétastaseswere determined by examination of serial tissue sections as describedin the text. The number of hepatic tumor foci per animal was determined byexamining 5 jim microscopic sections after the entire liver was sectioned at 0.2-cm intervals. Animals given injections of parental cell line SIB were studiedinitially and as simultaneous controls with cycles 4 and 5. The low incidence ofliver metastasis for the parental cell line noted initially was reproduced insubsequent experiments (0 of 10 and 2 of 11 animals given injections).

No. of hepaticmetastasis-bear

ingmiceCell

lineSIB51BLÕM-1

51BLÕM-251BLÕM-351BLÕM-4SIBLiM-SMetastatic

cycleParent12

34

5Doubling

time(h)10.610.1

10.611.512.310.6Macro

scopic2/21

0/74/66/88/8

10/10Micro

scopic4/21

1/74/66/88/8

10/10No.

of hepatictumor foci/

animal(range)1-34

3-65-105-15

50-100

static potential. In addition, the cells were found to producecollagen IV-degrading enzyme levels proportional to their met-astatic activity. These studies suggest that the increased meta-static activity observed in the selected cells could be related, inpart, to their ability to pass the basement membrane barrierswhich limit the spread of cells from the cecum and from thehepatic sinusoids to the liver parenchyma.

Clinical experience and experimental studies (36, 52, 53)strongly suggest that many human and animal tumors metas-tasize to preferential sites. Although this might be in part based

on their site of entry into the circulation (52), other factorsmust be involved (1, 52). Such factors may include tissue-specific chemoattractants (29,54-59) and preferential adhesionof metastatic tumor cells to target organ parenchymal cells orendothelium. In our study liver extract stimulated the movement of highly liver-metastasizing tumor cells to a significantlygreater degree than did bovine serum albumin or extracts oflung or brain (Fig. 6). This same liver extract enhanced theinvasion of tumor cells through the reconstituted basementmembrane. These results are similar to those reported by Hu-janen and Terranova (29) who demonstrated preferential migration of melanoma, sarcoma, and breast carcinoma cells toextracts from the organs for which they show metastatic predilection. The nature of such attractants remains to be determined, but these workers found the factors in brain and liver tobe of different molecular weights. According to these results,the metastasizing tumor cells can disseminate among differenttissues based on the presence of local factors stimulating theirmigration.

Preferential organ attachment of metastatic tumor cells hasbeen studied in vitro by a number of investigators (60-66).These studies have examined adhesion of tumor cells to frozentissue sections (61), organ culture slices (60), or cultured cellmonolayers (63-66). Using highly purified monolayer culturesof target cells, we have demonstrated specific in vitro adherenceof radiolabeled metastatic colon cancer cells to hepatic sinusoidal endothelial cells but not hepatocytes (Table 3). Rat targetcells were used since highly purified murine cells cannot beeasily prepared by our method in sufficient numbers from themouse to allow such studies. Our results are not surprising,since blood-borne metastatic cells must first adhere to endothelial walls and invade through basement membranes before en-

* . • *¿ - - .Fig. 4. Regional lymph node metastasis showing tumor cells (7") similar to those of the primary tu

1404

«in,x 250.

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

Sit SIBLiM-4 S1BUH-S 51B 51BUM1

Fig. 5. Invasion of parental and metastatic tumor cells through a reconstitutedbasement membrane matrix composed of purified bovine type I collagen (0.2 mg/matrix), type IV collagen (0.2 mg/matrix), and laminili (0.2 mg/matrix). Tumorcell suspensions (7 x 10* cells) in DMEM containing 0.5% BSA and 2% acid-treated NuSera were placed into the upper compartment of a modified Boydenchamber and allowed to penetrate the matrix onto a type IV collagen-coatedNuclepore filter (pore diameter, 1.0 ^m). Assays were done in the presence(chemoinvasion) or absence (invasion) of partially purified liver extract (10%) inthe lower compartment of the modified Boyden chamber. Data are expressed asa percentage of cells which penetrated the matrix after 24 (left) and 36 (right) h(±SD (bars)). Data were obtained for 5IB LiM-5 at 24 h only. «,P < 0.005; ««,P < 0.001 compared to parental cell line 51B studied under the same conditions:•,P< 0.05; +, P< 0.001 chemoinvasion versus invasion.

[J.1M'xÌ50-40-30-20-10-*

ñmmMM*9.5%

BSA Brun LungInnExtract

Fig. 6. Directed migration (chemotaxis) of highly liver-metastasizing tumorcell line 51B LiM-5 to target organ extracts in a modified Boyden chamber.Tumor cells 5 x 10* in DMEM containing 0.5% BSA were added to the uppercompartment and liver, lung, or brain extracts (10%) in DMEM containing 0.5%BSA were placed in the lower compartment of the chamber. After a 5-h incubationat 37"C, type IV collagen-coated Nuclepore barrier filters (pore diameter, 8 Mm)

were removed and migration to the lower surface of the filter quantitated. Thedata are average number (±SD (bars)) of migrated cells per high power field.Cells were counted in eight high power fields, and assays were performed intriplicate. «,P < 0.001 for liver versus BSA, brain, or lung extracts.

% Liver Extract Upper Compartment

ExtractLowerCompartment5

3se\\01S1008

±3

\17±237

±648

±71a±216

±2\\.21

±427

±5511±411±3X

25±3\20

±61011

±412±117±4s36

±5\

Fig. 7. Checkerboard analysis of the response of 51B LiM-5 cells to liverextract. Different concentrations of liver extract in DMEM containing 0.5% BSAwere added to the upper and lower compartments of a chemotaxis chamber.Tumor cells in DMEM containing 0.5% BSA were introduced into the uppercompartment. Studies were carried out for 4 h at 37"C. Numbers within innerboxes, number of migrated cells per high-power field (x 400). Responses of cellsto a positive gradient are shown below the diagonal, to a negative gradient, abovethe diagonal, and in the absence of a gradient, on the diagonal. Results areaverages ±SD of triplicate experiments.

countering organ parenchymal cells. Nicolson et al. (60) foundthat B16 melanoma cells often bound to exposed blood vessels,and therefore attached to endothelial rather than parenchymalcells in organ culture. Others, however, have shown that hepa-tocyte surface molecules may be involved in the adhesion ofsome types of carcinoma cells to rat hepatocyte cultures (64)and that the mechanism of adhesion of different tumor cells to

Table 2 Type IV collagen-degrading activity of tumor cell linesType IV collagen-degrading metalloprotease activity secreted by selected tumor

cell lines. Enzyme activity is expressed as amount of type IV collagen degradation(cpm released/101 cells from [3H]proline-labeled type IV procollagen). Results

represent the mean ±SD of three separate experiments each with triplicatedeterminations.

Cellline51B

51BLÃŒM-451B LiM-5Type

IV collagen-degrading activity (cpm

released/ IO7cells)191

±8.5475 ±106a

1704 ±23*

" P < 0.005 compared to 51B.

Table 3 Percentage of tumor cell adhesion to liver-derived cellsSelective binding of metastatic tumor cells from line 51B LiM-3 to purified

hepatocytes, hepatic sinusoidal endothelial cells, or plastic control in vitro wasevaluated. Cultured tumor cells were labeled for 24 h at 37°Cin mediumcontaining 0.3 >iCi/ml [I25l]iododeoxyuridine. Tumor cells (5 x 10*) were platedonto confluent monolayers of liver-derived cells or 35-mm plastic control dishesand incubated for 10 or 30 min at 37*C. Overlying medium was collected, cultures

washed x 3, and bound cells harvested and counted. The percentage of adhesionis defined as

Counts boundMedium + washes + counts bound x 100

Incubationtime(min)10

30Endothelial

cells47.7±4.3°'* (8)c

64.7 ±9.5 (5)°"Hepatocytes8.0±3.9 (8)17.1 ±11.6(5)Plastic11.2

±1.5(8)29.8 ±6.2 (5)

•Mean ±SD.bp<o.ooi.c Numbers in parentheses, number of experiments.d P < 0.005 compared to either hepatocytes or plastic.

hepatic organ target cells may be different (66). We have notyet tested whether the adhesion of our colon cancer cells tohepatic sinusoidal endothelial cells is specific for this endothe-lium in particular or endothelial cells in general. Experimentsby Alby and Auerbach (65) have demonstrated that capillaryendothelial cells derived from different organs are not alike andthat differences expressed at the cell surface of these cells canbe distinguished by tumor cells.

The model which we have described should be useful forstudying many factors which might affect the invasive activityand metastasis of colon cancer cells. Similarly, the highly liver-metastasizing cells selected may be used in this model to assesstherapeutic regimens that could lessen the spread of this disease.

ACKNOWLEDGMENTS

We thank Trish Harrington and Janice Woo for preparation of themanuscript and Mary Barker for preparation of the electron micrographs.

REFERENCES

1. Schirrmacher, V. Cancer metastasis: experimental approaches, theoreticalconcepts, and impacts for treatment strategies. Ad. Cancer Res., 43: 1-73,1985.

2. Fidler, I. J., Gersten, D. M., and Hart, I. R. The biology of cancer invasionand metastasis. Adv. Cancer Res., 28:149-250, 1978.

3. Hart, I. R., and Fidler, I. J. The implications of tumor heterogeneity forstudies on the biology and therapy of cancer metastasis. Biochim. Biophys.Acta, 651: 37-50, 1981.

4. Heppner, G. Tumor heterogeneity. Cancer Res., 44: 2259-2265, 1984.5. Fidler, I. J., and Hart, I. R. Biologic diversity in metastatic neoplasms: origins

and implications. Science (Wash. DC), 217:998-1003, 1982.6. Fidler, I. J. General considerations for the study of experimental cancer

metastasis. In: H. Bush (ed.), Methods of Cancer Research, pp. 339-439.New York: Academic Press, Inc., 1978.

7. Nicolson, G. Cancer metastasis—organ colonization and the cell-surfaceproperties of malignant cells. Biochim. Biophys. Acta, 695: 113-176, 1982.

8. Yamori, T., Tsuruo, T., Naganuma, K., Shigeru, T., and Yoshio, S. Isolationand characterization of highly and rarely metastatic clones from murine

1405

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ANIMAL MODEL FOR COLON CANCER METASTASIS

colon adenocarcinoma 26. Invasion Metastasis, 4: 84-97, 1984.9. VanLamsweerde, A-L., Henry, N., and Vaes, G. Metastatic heterogeneity of

cells from Lewis lung carcinoma. Cancer Res., 43:5314-5320, 1983.10. Heppner, G. H., Dexter, D. L., DeNucci, T., Miller, F. R., and Calabresi, P.

Heterogeneity in drug sensitivity among cell subpopulations of a singlemammary tumor. Cancer Res., 38: 3758-3763, 1978.

11. Giovanella, B. ( '.. and Fogh, J. The nude mouse in cancer research. Adv.

Cancer Res., 44:69-120, 1985.12. Sordat. B. C. M., Ueyama, Y., and Fogh, J. Metastasis of tumor xenografts

in nude mice. //;: J. Fogh and B. C. Giovanella (eds.). The Nude Mouse inExperimental and Clinical Research, pp. 95-147. New York: Academic Press,Inc., 1982.

13. Kyriazis, A. P., Kyriazis, A. A., McCombs, W. B. Ill, and Kereiakes, J. S.Biological behavior of human malignant tumors grown in the nude mouse.Cancer Res., 41: 3995-4000, 1981.

14. Tao, T-W., Matter, A., Vogel, K., and Burger, M. M. Liver-colonizingmelanoma cells selected from B-16 melanoma. Int. J. Cancer, 23: 854-857,1979.

15. Lafreniere, R., and Rosenberg, S. A. A novel approach to the generation andidentification of experimental hepatic métastasesin a murine model. J. Nati.Cancer Inst., 76: 309-321, 1986.

16. Kozlowski, J. M.. Fidler, I. J., Campbell, D., Xu, Z-L., Kaigh, M. E., andHart, I. R. Metastatic behavior of human tumor cell lines grown in the nudemouse. Cancer Res., 44:3522-3529, 1984.

17. Giavazzi. R., Campbell, D. E., Jessup, J. M., Cleary, K., and Fidler, I. J.Metastatic behavior of tumor cells isolated from primary and metastatichuman coloréela!carcinomas implanted into different sites in nude mice.Cancer Res.. 46: 1928-1933, 1986.

18. Tan, M. H., Holyoke, E. D., and Goldrosen, M. H. Murine colon adenocarcinoma: syngeneic orthotopic transplantation and subsequent hepatic metastasis. J. Nati. Cancer Inst., 59: 1537-1544, 1977.

19. Goldrosen, M. Murine colon adenocarcinoma. Immunobiology of metastasis.Cancer (Phila.). 45:1223-1228, 1980.

20. Bresalier, R. S., Râper,S. E., and Kim, Y. S. A new animal model for humancolon cancer metastasis. Gastroenterology, 88: 1334, 1985.

21. Wang, W. R., Sordat, B., Piquet, P., and Sordat, M. Human colon tumorsin nude mice: implantation site and expression of invasive phenotype. In: B.S. Sordat (ed.), Immune-Deficient Animals, pp. 239-245. Basel: S. Karger,1984.

22. Corbe«,T. H., Griswold, D. P., Roberts, B. J., Peckham, J. C., and Schabel,F. M. Tumor induction relationships in development of transplantable cancers of the colon in mice for chemotherapy assays, with a note on carcinogenstructure. Cancer Res., 35: 2434-2439, 1975.

23. Fidler, I. J. Selection of successive tumor cell lines for metastasis. Nat. NewBiol.. 242: 148-149, 1973.

24. Terranova, V. P., Hujanen, E. S., Loeb, D. M., and Martin, G. R., Thornburg,L., and Glushko, V. Use of a reconstituted basement membrane to measurecell invasiveness and selection for highly invasive tumor cells. Proc. Nati.Acad. Sci. USA, 83: 465-469. 1986.

25. Gallop, P. M., and Seifter, S. Preparation and properties of soluble collagens.Methods Enzymol., 6:635-641, 1963.

26. Kleinman, H. K., McGarvey, M. L., Liotta, L. A., Gehron-Robey, P.,Tryggvason, K., and Martin. G. R. Isolation and characterization of type IVprocollagen, laminin, and heparan sulfate proteoglycan from EHS sarcoma.Biochemistry, 21:6188-6193, 1982.

27. Timpl, R., Rohde, H.. Robey, P. G., Rennard, S. I., Foidart, J. M., andMartin, G. R. Laminin—a glycoprotein from basement membranes. J. Biol.Chem., 254: 9933-9937, 1979.

28. Loskutoff, D. J. Effects of acidified fetal bovine serum on the fibrinolyticactivity and growth of cells in culture. J. Cell Physiol., 96: 361-370, 1978.

29. Hujanen, E. S., and Terranova, V. P. Migration of tumor cells to organ-derived chemoattractants. Cancer Res., 45: 3517-3521, 1985.

30. Postlethwaite, A. E., Snyderman, R., and Kang, A. H. The chemotacticattraction of human fibroblasts to a lymphocyte-derived factor. J. Exp. Mod..144: 1188-1203, 1976.

31. Lowry, O. H., Rosebrough, N. J., Fair, A. L., and Randall, R. J. Proteinmeasurement with the folin-phenol reagent. J. Biol. Chem., 193: 265-275,1951.

32. Liotta. L. A., Tryggvason, K., Garbisa, S., Gehron-Robey, P., and Abe, S.Pania! purification and characterization of a neutral protease which cleavestype IV collagen. Biochemistry, 200: 100-104, 1981.

33. Turpeenniemi-Hujanen, T., Thorgeirrson, V. P., Hart, I. R., Grant, S. S.,and Liotta, L. Expression of collagenase IV (basement membrane collagen-ase) activity in murine tumor cell hybrids which differ in metastatic potential.J. Nati. Cancer Inst., 75: 99-103, 1985.

34. Salo, T., Liotta, L. A., Keski-Oja, J., Turpeenniemi-Hujanen, T., andTryggvason, K. Secretion of basement membrane collagen-degrading enzymeand plasminogen activator by transformed cells—role in metastasis. Int. J.Cancer, JO: 668-673, 1982.

35. Tryggvason, K., Gehron-Robey, P., and Martin, G. R. Biosynthesis of typeIV collagen. Biochemistry, 19: 1284-1289, 1980.

36. Fidler, I. J., and Nicolson, G. L. Organ selectivity for implantation survivaland growth of BI6 melanoma variant tumor lines. J. Nati. Cancer Inst., 57:1199-1202, 1976.

37. Irving, M. G., Roll, F. J., Huang, S.. and Bissell. D. M. Characterizationand culture of sinusoidal endothelium from normal rat liver: lipoproteinuptake and collagen phenotype. Gastroenterology, 87:1233-1247, 1984.

38. Friedman, S. L., Roll, F. J., and Bissell, D. M. Isolation and culture oflipocytes from normal rat liver (Abstract). Hepatology. 4: 1080, 1985.

39. Zigmond, S. H., and Hirsch, J. G. Leukocyte locomotion and chemotaxis:new method for evaluation and demonstration of cell-derived chemotacticfactors. J. Exp. Med., 137: 387-410, 1973.

40. Liotta, L. A., Garbisa, S., and Tryggvason, K. Biochemical mechanismsinvolved in tumor cell penetration of the basement membrane. In: L. A.Liotta and I. R. Hart (eds.), Tumor Invasion and Metastasis, pp. 319-334.The Hague: Nijhoff, 1982.

41. Henry, N., VanLamsweerde, A-L., and Vaes, G. Collagen degradation bymetastatic variants of Lewis lung carcinoma: cooperation between tumorcells and macrophages. Cancer Res., 43: 5321-5327, 1983.

42. Liotta, L., Abe, S., Gehron-Robey, P., and Martin, G. R. Preferentialdigestion of basement membrane collagen by an enzyme derived from ametastatic murine tumor. Proc. Nati. Acad. Sci. USA, 76: 2268-2272,1979.

43. Kramer, R. H., and Vogel, K. G. Selective degradation of basement membranemolecules by metastatic melanoma cells. J. Nati. Cancer Inst., 72:889-899,1984.

44. Liotta, L. A., Tryggvason, K., Garbisa, S., Hart, L, Foltz, C. M., and Shapie,S. Metastatic potential correlates with enzymatic degradation of basementmembrane collagen. Nature (Lond.), 284:67-68, 1980.

45. Salo, T., Liotta, L. A., and Tryggvason, K. Purification and characterizationof a murine basement membrane collagen-degrading enzyme secreted bymetastatic tumor cells. J. Biol. Chem., 25«:3058-3063, 1983.

46. Liotta, L. A., Thorgeirsson, V. P., and Garbisa, S. Role of collagenases intumor cell invasion. Cancer Metastasis Rev., /: 277-288, 1982.

47. Eisenbach, L., Segal, S., and Feldman, M. Proteolytic enzymes in tumormetastasis. II. Collagenase type IV activity in subcellular fractions of clonedtumor cell populations. J. Nati. Cancer Inst., 74:87-93, 1985.

48. Ogilvie, D. J., Hailey, J. K., Juacaba, S. F., Lee, E. C. G., and Tarin, D.Collagenase secretion by human breast neoplasms: a clinicopathologic investigation. J. Nati. Cancer Inst., 74: 19-27, 1985.

49. Jones, P. A., and DeClerk, Y. A. Destruction of extracellular matricescontaining glycoproteins, elastin, and collagen by metastatic human tumorcells. Cancer Res., 40: 3222-3227, 1980.

50. Markus, G., ( 'annoio. S. M., Kohga, S., Madega, J. M., and Mittleman, A.

Plasminogen activator secretion of human tumors in short-term organ culture, including a comparison of primary and metastatic colon tumors. CancerRes., «.-5517-5525, 1983.

51. Carlsen, S. A., Ramshaw, I. A., and Warrington, R. C. Involvement ofplasminogen activator production with tumor metastasis in a rat model.Cancer Res.. 44: 3012-3016, 1984.

52. Sugarbaker, E. B. Patterns of metastasis in human malignancies. In: i. J.Marchalonis, M. G. Hanna, and I. J. Fidler (eds.), Cancer Biology Reviews2, pp. 235-278. New York: Marcel Dekker, Inc., 1981.

53. Shearman, P. I., and Longenecker, B. M. Selection for virulence and organ-specific metastasis of herpes virus-transformed lymphoma cells. Int. J. Cancer, 25: 363-369, 1980.

54. Mensing, H., Albini, A., Krieg, B. F., Pontz, B. F., and Miller, P. K. Enhancedchemotaxis of tumor derived and virus-transformed cells to fibronectin andfibroblast conditioned medium. Int. J. Cancer, 33:43-48, 1984.

55. Mundy, G. R., DeMartino, S., and Rowe, D. W. Collagen and collagen-derived fragments are chemotactic for tumor cells. J. Clin. Invest., 68: 1102-1105, 1980.

56. Orr, F. W., Gondek, M. p.. Ward, P. A., and Mundy, G. R. Partialcharacterization of bone derived chemotactic factor for tumor cells. Am. J.Pathol., 99:43-52, 1980.

57. Romualdez, A. G., and Ward, P. A. A unique complement derived chemotactic factor for tumor cells. Proc. Nati. Acad. Sci. USA, Y72: 4128-4132,1975.

58. Lacovara, J., Cramer, E. B., and Quigley, J. P. Fibronectin enhancement ofdirected migration of B16 melanoma cells. Cancer Res., 44: 1657-1663,1984.

59. McCarthy, J. B., Palm, S. L., and Furcht, L. T. Laminin and fibronectinpromote the haptotactic migration of B16 mouse melanoma cells in vitro. J.Cell Biol., 98: 1474-1480, 1984.

60. Nicolson, G. L., Dulski, K., Basson, C., and Welch, D. R. Preferential organattachment and invasion in vitro by B16 melanoma cells selected for differingmetastatic colonization and invasive properties. Invasion Metastasis, 5:144-158, 1985.

61. Netland, P. A., and Zetler, B. R. Metastatic potential of B16 melanoma cellsafter in vitro selection for organ-specific adherence. J. Cell Biol., 101: 720-724, 1985.

62. Varani, J., Love«,E. J., Elgebaly, S., Lundy, J., and Ward, P. A. In vitroand in vivoadherence of tumor cell variants correlated with tumor formation.Am. J. Pathol., 101: 345-352, 1980.

63. Roos, E., Tulp, A., Middelkoop, O. P., and VanDePavert, I. V. Interactionbetween liver-metastasizing lymphoid tumor cells and hepatic sinusoidalendothelial cells. In: D. L. Knook and E. Wisse (eds.), Sinusoidal Liver Cells,pp. 147-154. Amsterdam: Elsevier/North-Holland Biomedicai Press, 1982.

64. Middelkoop. O. P., VanBavel, P., Calafat, J., and Roos, E. Hepatocytesurface molecule involved in the adhesion of TA3 mammary carcinoma cellsto rat hepatocyte cultures. Cancer Res., 45: 3825-3835, 1985.

65. Alby, L., and Auerbach, R. Differential adhesion of tumor cells to capillaryendothelial cells in vitro. Proc. Nati. Acad. Sci. USA, 81: 5739-5743.

66. Roos, E., Middlekoop, O. P., and VanDePavert, I. V. Adhesion of tumorcells to hepatocytes: different mechanisms for mammary carcinoma compared with lymphosarcoma cells. J. Nail. Cancer Inst., 73: 963-969, 1984.

1406

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1987;47:1398-1406. Cancer Res   Robert S. Bresalier, Erkki S. Hujanen, Steven E. Raper, et al.   Liver-metastasizing Abilityand Characterization of Murine Cell Lines with Enhanced An Animal Model for Colon Cancer Metastasis: Establishment

  Updated version

  http://cancerres.aacrjournals.org/content/47/5/1398

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/47/5/1398To request permission to re-use all or part of this article, use this link

on June 7, 2020. © 1987 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from