[CANCER RESEARCH 56. 3075—3086, July 1. 1996)
`
`Multiple Features of Advanced Melanoma Recapitulated in Tumorigenic Variants
`
`of Early Stage (Radial Growth Phase) Human Melanoma Cell Lines: Evidence
`
`for a Dominant Phenotype1
`
`Maria Rosa Bani, Janusz Rak, Dena Adachi, Rodney Wiltshire, Jeffrey M. Trent, Robert S. Kerbel, and
`Yaacov Ben-David2
`
`Division of Cancer Biology Research. Reichmann Research Building, 5-218. Sunnybrook Health Science Centre. 2075 Bayview Avenue, Toronto, Ontario 4N 3M5, Canada
`[M. R. B., J. R., D. A. R. S. K., Y. 8-0.]; Department ofMedicaI Biophysics. University of Toronto, 215 Huron Street, Toronto. Ontario M55 [.4], Canada /R. S. K., Y, 8-0.]; and
`National Center for Human Genome Research, Bethesda, Maryland 20892—4470 IR. W., I. M. 7.]
`
`ABSTRACT
`
`The vast majority of primary human cutaneous melanomas undergo a
`slow and gradual progression from a clinically indolent, curable radial
`growth phase (RGP) to a malignant vertical growth phase. We sought to
`develop a way of isolating genetically related malignant variants from a
`benign RGP human melanoma, called WMSS. The parent and variants
`were then used as a model system to examine to what extent the expression
`of clinically and biologically relevant pbenotypic features characteristic of
`advanced melanomas are associated with (and thus perhaps causative of)
`such a malignant conversion. Such a model system could also be used as
`a means of eventually identifying genetic alterations and cellular changes
`involved in the malignant switch in melanoma progression. To develop
`such a model, we subjected WM35 cells to retroviral insertional mutagen-
`eels, which was followed by selection for progressive growth of solid
`tumors in nude mice. Highly aggressive and phenotypically stable tumor-
`igenic variants were derived which contained at least four integrated
`proviruses. In contrast to the parental WM35 cells, these cell lines ex-
`prmd several phenotypic features characteristic of naturally derived,
`advanced-stage malignant melanoma cells. Thus, in addition to tumor
`forming ability in nude mice, the variants were growth factor and an-
`chorage independent, overexpressed the MUC18 adhesion molecule, and
`lost responsiveness to the growth-inhibitory effect of several cytokines,
`including interleukin 6, transforming growth factor B, interleukin 1B, and
`tumor necrosis factor-a. Tumorigenicity and “multicytokine resistance”
`were dominant traits since in somatic cell hybrids between the parental
`cells and a tumorigenic subline no suppressive effect of the former cell
`population was observed. These findings suggest that one or more domi-
`nantly acting genetic alterations might be involved in this progression of
`RGP melanoma cells. The identity of such alterations remains to be
`determined.
`
`INTRODUCTION
`
`Human cutaneous malignant melanoma is a potentially highly
`aggressive, metastatic form of skin cancer, the incidence of which is
`increasing in many parts of the world at a greater rate than any other
`type of cancer, and for which there is no known effective treatment in
`its more advanced stages (1—4). There are a number of well-defined
`clinical and pathological stages of disease progression associated with
`this type of cancer. For example, normal melanocytes can give rise to
`putative precancerous lesions known as benign atypical melanocytic
`
`Received 10/24/95; accepted 5/1/96.
`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 in accordance with
`18 U.S.C. Section 1734 solely to indicate this fact.
`‘ This work was supported by Grants CA 54908 (Y. 8-D.) and CA 41233 (R. S. K.)
`from the NIH and a grant from the Medical Research Council of Canada (Y. B~D.).
`R. S. K. is a Terry Fox Cancer Research Scientist of the National Cancer Institute of
`Canada. and Y. B, D. is a Research Scientist of the National Cancer institute of Canada
`supported by funds from the Canadian Cancer Society through the Teny Fox Run.
`M. R.B. was a recipient of a fellowship from the Italian Regione Abruzzo and the
`European Economic Community (Specialisti in Rioerca Biomedica) and is a recipient of
`a scholarship from the Sunnybrook Trust for Medical Research.
`2To whom requests for reprints should be addressed. Phone: (416) 480-5711; Fax:
`(416) 4806703.
`
`nevi, the vast majority of which eventually regress and disappear (5).
`Occasionally, however, a nevus can progress to become a so—called
`RGP3 primary melanoma (5). These lesions are rather indolent and
`confined primarily to the epidemiis (“melanoma in situ”), as they tend
`to expand in a horizontal, plaque-like manner (5, 6). RGP melanomas
`are almost always cured by surgical excision and presumably consist
`of tumor cells that are not competent for metastasis (7—10). Slowly
`growing RGP melanomas can evolve over long periods of time and
`eventually change to become more rapidly growing VGP primary
`melanomas (5). Such lesions penetrate the basement membrane sep-
`arating the epidermis from the underlying dermal mesenchyme, and
`contain progressively greater proportions of metastatically competent
`melanoma cells, which can readily proliferate in the foreign environ-
`ment (5). The final phase of human cutaneous melanoma progression
`is the formation of distant organ metastases, particularly from primary
`lesions greater than 0.76 mm in thickness (5). Specific genetic alter-
`ations associated with this sequence of events remain largely un-
`known.
`
`One of the more interesting biological features of cutaneous human
`melanomas is the tendency of cell lines established from RGP or very
`early VGP lesions to recapitulate their clinically indolent growth
`behavior when inoculated into nude mice (9, ll, 12). Even if injected
`orthotopically, solid tumors fail to form or do so slowly, growing only
`to very small sizes and then remaining as such, or regressing alto-
`gether (11). in contrast, cell lines established from advanced mela-
`noma lesions usually form rapidly growing tumors in nude mice (9,
`13). Thus, cell lines from RGP or early-stage VGP primary lesions
`can essentially behave as slowly growing “dormant” but visible tu-
`mors in the skin of nude mice, much as they do in humans.
`The purpose of our studies presented here was to use retroviral
`insertion mutagenesis as an experimental strategy to recapitulate ma—
`lignant progression of human melanoma. This methodology has been
`used as a powerful tool to identify genes involved in the induction and
`progression of murine and avian malignancies generated by tumori-
`genic, slow—transforming retroviruses (14, 15). Such studies have
`resulted in the discovery of genes relevant to human cancer progres-
`sion, including Fli-l, p53, erb—B, and myc (14—16). More recently, the
`in vitra application of this approach, using retroviruses, has made
`possible the identification of a “tumor invasion” gene, known as
`Tiam-l (17), as well as loci responsible for conferring resistance to
`antiestrogens in human breast cancer cells (18) and resistance to
`cisplatin in tumor cells (19).
`It was anticipated that if an RGP-derived human melanoma cell line
`that is nontumorigenic in nude mice is mutagenizcd by a replication-
`defcctive retrovirus, a large pool of mutant cells carrying single or
`multiple viral integration sites would be generated. This pooled pop-
`
`3 The abbreviations used are: RGP, radial growth phase; VGP, vertical growth phase;
`FBS. fetal bovine serum; MSCV. murine stem cell virus; IL. interleukin; TGF. uansform-
`ing growth factor; TNF, tumor necrosis factor; FACS, fluorescence-activated cell sorting;
`Ab, antibody; ICAM-l, intercellular cell adhesion molecule 1; TG, o-thioguaninc; HAT,
`hypoxanthine/aminopterin/thymidine.
`3075
`
`Genentech 2085
`
`Hospira v. Genentech
`nmuhlnmami in”... ,__....,.._,M- an”; ...... 1 aaaaaa nAAAMrAM r: nn47 m 4mm: ANN-mm. Amwmznun». I.” ANMAHDR2017_00737
`
`Genentech 2085
`Hospira v. Genentech
`IPR2017-00737
`
`1
`
`

`

`DOMINANT MALIGNANT CONVERSION IN MELANOMA
`
`tissue culture plates (Nunc, Kamstrup/Roskilde, Denmark) at a density of
`1 X 104 cells/well. At different times after plating, cells were harvested by
`brief exposure to trypsin-EDTA and counted by a hematocytometer or Coulter
`counter.
`
`ulation could then be injected into nude mice such that mutations
`rendering any cells with tumorigenic properties could be efficiently
`selected for. The resulting tumorigenic variants, if successfully recov-
`ered from such tumors, would be amenable to molecular analysis as
`well as identification of biological properties that co-segregate with
`tumorigenicity.
`Our results indicate that indeed such tumorigenic variants of RGP-
`derived melanoma cells can be generated. Moreover,
`the variants
`express a cluster of biological properties that are strikingly similar to
`properties expressed by cell lines established from “authentic” late—
`stage melanoma lesions. Some of these properties, including “multi-
`cytokine resistance" and in viva tumorigenic growth. behave in a
`dominant fashion, as shown by analysis of somatic cell hybrids
`generated between clonally related pairs of tumorigenic and nontu-
`morigenic melanoma cell lines. This latter property suggests that an as
`yet unknown dominantly acting oncogene(s) may be involved in
`human melanoma progression.
`
`MATERIALS AND METHODS
`
`Cells and Culture Conditions. The WM35 human melanoma cell line was
`
`Cytokine-mediated Growth Inhibition Assays. The cytokine sensitivity
`of parental or variant melanoma cell lines was determined by [3H1thymidine
`incorporation assay performed in RPMI 1640 supplemented with 1% FBS and
`appropriate concentrations of cytokines (23, 24). The cytokines used included
`1L—6, 1L~lB, TGF-B, and TNF—a and were purchased from Upstate Biotech—
`nology (Lake Placid, NY). Tumor cells were plated in fiat-bottomed, 96-well
`tissue culture plates (Nunc) at a density of 5 X 103 cells/well and allowed to
`attach overnight. Varying concentrations of human recombinant cytokines in
`100 p.l of medium containing 0.1% FBS were then added. Forty—eight h later
`(96 h for TNF-a), the cultures were pulsed with 2 trCi/well of [3H]thymidine
`(Amersharn, Oakville, Ontario, Canada) in 50 p.l of serum-free medium for
`4—6 h. The incorporated [3H]thymidine was measured in a LKB betaplate
`liquid scintillation counter (Wallac, Turku, Finland).
`Detection of Melanoma-associated Antigens by FACS Analysis. Expo-
`nentially growing cells were harvested from tissue culture plates either by brief
`exposure to 0.25% trypsin-0.02% EDTA or by mechanical scraping with a
`rubber policeman. After washing,
`1 X 10" cells were incubated with the
`appropriate primary antibody for 30 min at 4°C. Cells were washed twice and
`incubated for 30 min at 4°C with 100 [Li of a 1:50 dilution of appropriate
`kindly provided by Dr. Meenhard Herlyn (Wistar Institute. Philadelphia, PA).
`FITC-conjugated secondary antibody (Jackson lmmunoresearch, West Grove,
`This cell
`line was originally isolated from a patient diagnosed with RGP
`PA). The cells were washed again, then fixed with 1% paraformaldehyde in
`primary melanoma on the basis of histological analysis and clinical outcome
`PBS, and analyzed using a FACScan flow cytometer (Becton Dickinson, San
`i.e., cured by surgery (9). The cells were maintained in culture in RPMI 1640
`Francisco, CA) using standard excitation conditions (488-nm argon laster
`(Life Technologies, Inc, Grand Island, NY) supplemented with 5% heat-
`beam). Abs used in this study were directed against known melanoma-asso-
`inactivated FBS (Hyclone, Logan, UT). All cultures were periodically verified
`ciated surface molecules including: the avB3 integrin (rabbit polyclonal anti—
`as Mycoplasma free.
`serum; Telios, San Francisco, CA); ICAM-1 and MUC18 adhesion molecule
`Retroviral Infection. Replication-defective rim—containing MSCV with
`(monoclonal Ab BA4 mouse lgGl; kindly provided by Dr. J. P. Johnson,
`amphotropic host range was produced by infecting GP+envAM12 helper-free
`University of Munich, Munich. Germany; Refs. 25 and 26); CD44 adhesion
`packaging cells (20) with the ecotropic version of the virus (a gift from Dr.
`molecule (monoclonal Ab 50134 mouse IgG kindly provided by Dr. M. Letarte.
`R.G. Hawley, Toronto, Ontario, Canada: Ref. 21). Helper-free recombinant
`Toronto, Ontario, Canada; Ref. 27).
`virus with a titer in the range of 0.5—] X 10° G418«resistant, colony-forming
`Generation of Somatic Cell Hybrids. Hypoxanthine-guanine phosphori-
`units/m1 was used to infect WM35 human melanoma cells using a standard
`botransferase—deficient spontaneous mutants of T35~3.l
`tumorigenic variant
`procedure described previously (19).
`In brief,
`the cells were plated at a
`cells were selected by seeding the cells at clonal density (1 X 105 cells/
`concentration of 2 X 10’ cells per 60-mm dish, allowed to grow overnight, and
`lOO-mm dish) in culture medium containing 60 [.LM TG (2-amino-6-mercap-
`then infected with approximately 1.2 ml of undiluted, freshly thawed, viral
`topurine; Sigma). Somatic cell hybrids between such tumorigenic T35-3.1 TG
`supernatant followed by the addition of fresh medium. After 48 h, the medium
`cells (G418 and TG resistant, HAT sensitive) and nontumorigenic, wild-type
`was removed, and selective medium containing 800 ug/ml of G418 (Life
`parental WM3S cells (G418 sensitive, HAT resistant) were prepared by seed-
`Technologies, Inc.) was added. The resulting G418-resistant WM35 cells from
`ing the mixture of the two partner cell lines (5 X 105 cells each) in 60-mm
`each of the infected cultures were pooled, expanded, and injected into nude
`dishes 24 h prior to initiation of the cell fusion. At the time of fusion, cells were
`mice, as described below.
`washed once with PBS and then treated with 50% polyethylene glycol as a cell
`In Viva Selection of WM35 Tumor'igenic Variants (Tumorigenicity
`fusion agent (Mr 4000'. PEG 4000; Life Technologies, Inc.) for l min (28).
`Assays). Exponentially growing cells were injected (1 X 10") subdennally
`After removal of PEG 4000 by several washes with PBS, the cells were refed
`(22) into the dorsolateral flank of 5—7—week-old BALB/c nu/nu mice (Harlan
`with fresh medium and allowed to recover overnight. The cells were then
`Sprague—Dawley) anesthetized with 60 mg/kg pentobarbital (Nembutal
`so—
`harvested and replated at clonal density (1 X 105 cells/IOO—mm dish) in a
`dium; Abbot, Montreal, Quebec, Canada). The frequency of overt progressive
`double-selective medium containing 800 ug/ml of G418 and 1X HAT sup—
`tumor formation was scored periodically, and perpendicular dimensions of the
`plement containing 100 mu sodium hypoxanthine, 0.4 pM aminopterin, and 16
`subdennal outgrowths were measured using a Vernier caliper. Tumor volume
`p.M thymidine (Life Technologies, Inc.) in RPMI 1640. The resulting doubly
`was calculated as: width2 X length/2 (volume cm3 = mass g).
`resistant colonies were pooled (multiple colonies from nine independent dish—
`Establishment of Tumor-derived Tumor-igenic Variant Sublines. Each
`es/pool). analyzed for DNA content, and used for subsequent experiments.
`of the progressively growing tumors in nude mice was aseptically removed and
`DNA Content Analysis by Flow Cytometry. Ploidy and cell cycle distri—
`either immediately snap frozen and/or formalin fixed for future studies, or
`bution of parental, tumorigenic. and hybrid melanoma cell lines were assessed
`adapted to tissue culture. In brief,
`the tumor mass was grossly cleaned of
`by FACS analysis. Cells were harvested, washed, and resuspended in PBS, and
`necrotic areas, connective tissue, and skin. Tumor cells were then dispersed
`1 X 10° cells were fixed in ice~cold ethanol (final concentration, 50%) for at
`mechanically and by enzymatic disaggregation for 30 min at 37°C in 5 ml of
`least 30 min; cells were then washed and incubated with 1000 units/ml of
`serum-free medium containing 270 units/ml of collagenase type 3 (Worthing-
`RNase (Phannacia Biotech. Inc, Ste. Anne de Bellevue, Quebec, Canada) for
`ton, Freehold, NJ) and 300 units/ml of hyaluronidase (Sigma Chemical Co.. St.
`30 min at 37 °C. Finally, the cells were stained with 50 ug/ml of propidium
`Louis, MO). The suspension was then seeded in RPMI 1640 supplemented
`iodide (Sigma). Analyses were performed on at
`least 10" cells using a
`with 5% FBS and 400—800 pg/ml of G418. The resulting tumor—derived cell
`FACScan flow cytometer.
`lines were designated as “tumorigenic variants." Subsequent molecular char-
`Southern Blot Analysis. High molecular weight DNA was isolated by a
`acterization experiments, for the most pan, were performed using a tumori»
`modification of the proteinase Klphenol/chloroform method, as described
`genic variant cell line called TBS-3.1. This cell line was cloned, after four in
`previously (29). DNA was digested with restriction enzymes and separated by
`virro passages, by limiting dilution, and two independent representative clones,
`electrophoresis through a 0.8% agarose gel. The DNA was then transferred
`called D1 and D2, were subjected to further analyses (Fig. 1).
`onto nylon hybridization membrane using 10X SSC and probed with radiola-
`In Vim: Growth Assays. All of the growth assays were performed in
`beled DNA as described elsewhere (30).
`RPMI 1640 supplemented with 0.5 or 5% FBS. Cells were seeded in 24-well
`3076
`
`HAHWJAAAAA ‘LAM nnnnnvan "roar; ...... l aaaaaa “AAAML‘AV (3
`
`"3/147 63 4nf‘f.‘ AMAvinr-uv‘ AAAAA:A‘:AV\ (Av f" nnnnn
`
`2
`
`

`

`DOMINANT MALIGNANT CONVFJlSlON 1N MELANOMA
`
`
`
`
`
`
`
`PlatedelO‘ooHI/flask
`
`tam/flask
`Infection: 12 ml viral
`(titer- 0.5—1 x 10' 6418‘ CPU/ml)
`
`G418 selection (0.8 Wm!)
`
`
`
`
`
`
`
`
`
`
`Injection: 1 x 10'eolle/mome;
`5 mice/Independent infection
`
`
`
`
`Fromthemquantodsolldtumoreetabllsh
`
`'Tumorigenic variant"
`In vitmeell linen
`
`Isolated clones by Ilmtttng dllutlon procedure
`(cell Ines. each orlglnated from a elnglo coll)
`
`
`
`Fig. 1. Flow diagram illustrating the procedure
`for isolation of tumorigenic human melanoma cell
`variants. A total of seven independent retrovirus
`infection experiments was carried out. each time
`using five mice as recipients for the retrovirus-
`infected WM35 cells isolated in vitro. Tumors
`arose in experiments involving infection #3 (in
`which one mouse was positive) and infection #4 (in
`which two mice were positive). The tumor which
`arose in one of the mice injected with infection #3
`was removed and established in culture.
`It was
`designated T35—3.l, from which two clones were
`subsequently derived. called D1 and D2. Likewise,
`the tumors which arose in two of five mice that
`were injected with infection #4 were established in
`culture and were designated T3541 and T35-4.2.
`respectively.
`
`reached less than 4 mm in diameter. These biologically “benign"
`lesions occasionally regressed or else remained in the subderrnal
`tissue for extended periods of time (observed up to 8 months) but
`never give rise to overt aggressively growing tumors. This indolent
`behavior is reminiscent of the clinical behavior of RGP primary
`melanomas, which essentially behave as dormant, benign tumors (2,
`5, 6). In contrast, in 3 of 31 cases, virally infected WM35 cells grew
`rapidly, forming large and highly vascularized solid tumor masses
`within 4—8 weeks (Figs. 1 and 2). The vigorous host stromal response
`may have been an important factor to facilitate the fairly rapid
`outgrowth of the retrovims-mutagenized tumorigenic clones.
`From the solid tumors, which arose in nude mice, permanent cell
`lines were established in vitro (Fig. 1). These cell variants were found
`
`0 WM35
`+lnttt4,tum.1
`—-¥——lnt04,turn.2
`—I— lntl3.tum.1
`
`
`
`Genomic DNA Phage Library Construction and Cloning of Provlrus
`Flanking Genomic DNA Sequences. The genomic DNA library was con-
`structed by using DNA isolated from clone D2 of the tumorigenic variant
`T35-3.1 cell
`line. After partial EeoRl digestion. genomic DNA was size
`selected on a continuous sucrose gradient (10 to 40%, w/v) by ultracentrifu-
`gation. The 15—23-kb genomic fragments were ligated to EcoRI—digested
`EMBL4 bacteriophage arms (Stratagene, La Jolla, CA), followed by packaging
`of the resulting phage DNA constructs (Gigapack II packaging extract; Strat-
`agene). The library (2—3.5 >< 10° recombinant bacteriophages) was propagated
`in Escherichia cali. host strain LE 392, and screened with a provims specific,
`pgk-neo probe to isolate integrated proviruses and their flanking sequences.
`After four rounds of plaque purification, eight retrovirus-positive phages were
`identified containing human genomic DNA sequences.
`Knryotype Analysk. The WM35 and WM35-3.1 cell lines were cultured
`in RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine,
`100 units/ml penicillin, and 100 jig/ml streptomycin for 7—10 days. To obtain
`metaphase chromosomes. the cells were treated with 0.1 nyml Colcemid prior
`to harvesting, fixed with Carnoy's fixative, and finally GTG banded (31). The
`OTC-banded metaphase chromosomes were photographed. and karyotypes
`were generated and analyzed for each cell line.
`
`RESULTS
`
`
`
`Tumorlieu(g)
`
`0.1 0.01
`
`0
`
`20
`
`40
`
`-
`
`60
`
`80
`
`100
`
`200
`
`Retroviral Infection and Selection of Tumorigenic WM35 Hu-
`man Melanoma Variants. The nontumorigenic human melanoma
`cells WM35 were subjected to retroviral
`insertional mutagenesis,
`followed by subdemtal injection of the pooled population into athy—
`mic nude mice as outlined in the scheme shown in Fig. l. A replica-
`tion—defective MSCV carrying the neomycin phosphotransferase gene
`(neo; Refs. 19 and 21) was used for infection. This produced 60 to
`70% of G4l8-resistant cells, thereby indicating that they harbored at
`least one integrated copy of the provirus. Parental WM35 or G418—
`resistant cells from seven independent virus-infected populations were
`Fig. 2. Conferral of a “tumorigenic phenotype“ to WM35 cells. One million cells from
`the parental WM35 cell
`line and several
`independent. virus-infected populations were
`injected subderrnally into nude mice. No progressively growing tu-
`injected subdermally into the dorsolateral flank of 5-7-weelt—old female athymic nude
`mors arose following injection of the parental WM35 cells (40 mice).
`mice. The mice were monitored for up to 8 months after injection. The dimensions of the
`Only in 60% of mice did WM35 cells give rise to an outgrowth in the
`growing tumor were measured using a caliper. and the mass of each tumor was calculated
`as: tumor mass (g) = volume (cm’) = width2 X length/2. Ears. SD,
`form of a slow growing, flat hyperplastic lesion that in 4 months
`3077
`
`Daya otter lnjoctlon
`
`MANHAIAAAAA ‘VAM .-,—.-..-.—»»An nAAy: AAAAAA l AAAAAA NAAAML‘AV (2 GIN/1'7 KA 4nnr: AMAlen» AAAAA2’5‘;AW IA» 1" AAAAA
`
`3
`
`

`

`DOMINANT MALlGNANT CONVERSION IN MEANOMA
`
`
` TumorMon(9)
`
`Tumormu(g) 8
`
`0
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`404550
`
`‘5
`
`50
`
`20
`
`25
`
`30
`
`35
`
`40
`
`Deneflulnbctlon
`
`nap-rum
`
`Fig. 3. Tumorigenicity of WM35 variants is a remarkably stable phenotype. "Dre tumorigenicity of the WM35-3.1 variant was tested after being maintained in culture continuously
`for different periods of time (5—25 passages) after explantation (left panel) or when clonal cell lines. obtained by limiting dilution. were derived (right panel). One million cells were
`injected subdermally into the dorsolaterai flank of S—7-week—old female athytnic nude mice. Each data point represents the mean calculated for five mice; bars. SD.
`
`to be G4l8 resistant, human in origin. and genetically related to
`WM35. The evidence for this is the following. The same pattern of
`bands was observed when the genomic DNA was digested with
`different restriction enzymes and then was hybridized with a probe for
`human DNA-specific a-satellite sequences (data not shown). Like-
`wise, both WM35 and tumorigenic variant
`(T35—3.1) displayed a
`human karyotype with significant similarity between the two cell lines
`(see below for description). Furthermore. both WM35 and tumori-
`genic variants express the antigens HLA A-B-C (data not shown). The
`absence of contaminating mouse stromal cells was confirmed by
`Southern analysis using total mouse genomic DNA as a probe and by
`FACS analysis. which failed to show cells expressing the BALB/c
`mouse—specific histocompatibility haplotype H-ZDd (data not shown).
`Thus, the tumorigenic variant cell
`lines appear to represent a true
`clonal expansion and progression of WM35 parental cells.
`The tumorigenic sublines of WM35 expressed a remarkable phe-
`
`notype stability. Re-injection of the cells into nude mice reproducibly
`led to 100% tumor take. The tumor-forming ability as measured by
`tumor growth rate,
`incidence, or latency period in the case of the
`T35-3.l cell
`line (Fig. 3,
`left panel) remained unchanged over 8
`months of continuous culture in vitro. Furthermore, two clonal pop-
`ulations T35-3.1D1, T35-3.1D2, and the T35-3.l cell line, from which
`the clones were derived, all expressed virtually identical tumorigenic
`properties (Fig. 3. right panel). Similar results were also obtained for
`the tumorigenic variant TBS—4.2 (data not shown). In all of these
`cases, there was no evidence of significant mouse host cell infiltration
`of the solid tumors which arose upon re-injection of these cell sub-
`lines, unlike that observed in the “primary" transplants (discussed
`above).
`
`We have also compared one aspect of the genetic stability of the
`parental and variant cell lines directly, i.e., by testing the frequency
`of spontaneous mutations at the hypoxanthine-guanine phosphori-
`
`Fig. 4. Expression of melanoma—associated an—
`tigens:
`immunofluorescence analysis. Cells were
`stained with the proper antibody (see “Materials
`and Methods"). followed by FlTC-conjugated ap—
`propriate secondary antibody. Samples were ana-
`lyzed by flow cytometry. and the data were dis-
`played as two dimensional plots with the abscissa
`showing FlTC-conjugated antibody staining. The
`lower left quadrant represents unstained cells.
`while the lower right quadrant represents positive-
`stained cells. Numbers represent the percentage of
`positive cells.
`
`
`
`4
`
`

`

`DOMINANT MAUGNANT CONVERSION IN MELANOMA
`
`E
`
`2Numberofcells/well> 1m
`
`1
`
`3
`
`I
`
`7
`
`O
`
`11
`
`Days after plating
`
`§ 1w
`‘6
`
`é
`3
`2
`
`1600
`
`—o—- m 5
`«4354.1
`+ T3542
`+ T343
`
`1
`
`3
`
`5
`
`7
`
`9
`
`11
`
`
`
`Days after plating
`thr 5.1:: vim: growth rate of WMBS and it: mmorigenie variants. Twenty thousand rxlls were plated in etch well of I 24—well plate in the pretence of 5% FBS (A) o: 0.5% PBS
`(8'). Etch point represent: the averue cell number counted in triplicate by Cwlter cell cwnler (A) or the number of uypan blue-excluding cells counted in quadruplicate by
`helmtocytometer (8). SD for each point was let: then 10%. In C (T35-3.l) and D (WMSS). cells are pictured in the presence of 05% of FBS for 7 days
`
`botransferase locus. Such mutations result in acquisition of TG
`resistance and loss of ability of the cells to grow in HAT—supple-
`mented medium. When the cells were preselected in HAT and then
`passaged in culture for over 20 population doublings, none of the
`
`cell lines expressed a "mutator" phenotype in that WM35, T35-3.l.
`and TBS-4.2 cells gave rise to TG-resistant colonies at a similar
`steady,
`low rate of about 10—6—10_5 cells/generation (data not
`shown).
`
`In
`
`a
`
`In
`
`N
`
`'1
`
`W
`I
`m
`TNF Cl (W)
`
`m
`
`an I
`
`at
`
`to
`1
`".4 5 (right!)
`
`no
`
`m
`
`..
`
`..
`
`..
`0.
`
`0.
`
`l
`
`1
`
`K
`
`A “I
`
`E g
`
`o i
`
`E .
`
`
`
`OJ
`
`I
`
`II
`
`IN
`
`3
`E O
`
`I.- a (mint)
`
`TBF 13 (new)
`
`Fig. 6. Loss of myonsivenas to the Winhibiwey effects ofcytokinu. CeHs were plated at a density of 5 X 103 cellafwell in flat—bottomed. 96<Wdl plates, allowed to attach
`overniflt, and incubated with or without cytokine at the indium! concentrations. After 43 h of incubatim (96 h for TNF-a), cell: were pulsed for 5 h with 2 “Ci [’Hkhymidine/well
`and faucet-ed byW of [sfllmymidine inoorportion into DNA hmrpontion in the absence ofcytoltines (comm!) was oomidered IOO‘E, and hictxponu‘on in the presence
`ofcytah’ne w normalized to this mull much data point is expuened as 3 am oftripicue deaemtinatiom; bars. SD, Data an representative of It least one independent expennueru.
`3079
`
`nnlllmlAAAAAl IVA-an J-lu‘d-J-wh—AA AAHAA..VV‘AI» Ava A». nAAAML‘AI‘ C nn47 IA 4nnr: AMAV:AHW AAAAAIAHAH (Aw I" nnnnn
`
`5
`
`

`

`DOMINANT MALIGNANT CONVERSION IN MELANOMA
`
`
`
`(’9
`o‘o‘
`SKN‘Pfifi’
`§§
`éé’éwoww
`§§§v§5o§5r§5r§5e~~eew¢$
`é
`sska‘bQ‘bQJQQQ
`
`B
`
`9.4 Kbp
`
`6.6 Kbp
`
`4.4 Kbp
`
`2.3 Kbp
`
`2.0 Kbp
`
`Fig. 7. Genomic analyses of retrovirally in-
`duced tumorigenic variants. Genomic DNA (15
`pg) isolated from cell lines or explanted solid
`tumornnsus wastfigeued wnh an msukdon
`enzyme Beck] and separated by electrophoresis
`through a 0.8% agarose gel. A. DNA was trans-
`ferred onto nylon membrane and hybridized
`with a provirus-specific probe (600 bp of neo
`sequence). Arrows,
`the position of common
`buxk.W%135.hunun mehnomacdlhnede-
`rived from an early stage lesion; WM 35 infl3.
`WM35 cells infected with MSCV prior to injec-
`tion; T35-3.l, tumorigenic variant cell line de-
`rived from explanted tumor mass induced fol-
`lowing injection of WM35-mm; T35-3.Ia and
`T35-3.lb,
`independent cell
`lines
`re-isolated
`from explanled tumor masses grown upon T35-
`3.l
`injection; T35-3J I292 and T35-3.l #294.
`tumor masses grown in mouse #292 and #294.
`respectively. by injection of TBS-3.1 after hav-
`ing been in culture for 6 months subsequent to
`awn KohUon;Aa BL Dl,Bi Bi um DZ
`independent clonal cell lines. isolated by limit-
`ing dilution prowdure from T3531; DI tumor
`and D2 tumor, rumor masses generated by in-
`jection of Dl and D2, respectively. B. integra-
`tion #4, detected by 3' flanking genomic se-
`quence specific probe (probe A, Fig. 7C).
`Arrow,
`the position of the band with reduced
`mobility conesponding to the allelic rearrange-
`ment due to the integrated provirus.
`
`melanoma. Thus. we observed that a large proponion of cells (77—
`Tumorlgenic Variants Express a High Level of the MUC18
`84%) in the variant cell lines express high levels of MUC18 glyco-
`Adhesion Molecule. Melanoma progression is known to be associ-
`protein, whereas its expression in the WM35 parental cells was very
`ated with defined alterations in expression of cell surface glycopro-
`low (16% positive cells). Interestingly, this expression profile closely
`teins (32) including, for example, the CD44 (32) and ICAM-1 adhe-
`resembles the results reported previously of immunohistochemical
`sion molecules (25).
`the vitronectin (aVB3) receptor (33), and the
`analyses of clinical melanoma specimens and cell
`lines in which
`MUC 18 adhesion molecule (26, 34—36). Fig. 4 shows the flow
`MUC18 expression was found to be a prognostic indicator of subse-
`cytometry profile of the expression of some of these antigens in
`quent disease progression (32. 34—36).
`parental WM35 cells and the in viva isolated tumorigenic variants. A
`Tumorigenic Variants Express a Growth Factor-independent
`high proportion of WM35 cells in vitro express the vitronectin recep-
`Phenotype in Vitro. Unlike cell
`lines derived from “early-stage”
`tor, integrin 6V3} (98% of cells), CD44 (94% of cells), and ICAM-1
`melanoma,
`the cells isolated from advanced melanomas can fre-
`(94% of cells). A similar level of expression is also detected in three
`independent
`tumorigenic variant cell
`lines T35-3.l, TBS—4.2, and
`quently be maintained in culture without exogenously added growth
`factors or serum (12, 13, 37). We compared, therefore, the growth
`T35—4.l, which is in agreement with observations published previ-
`properties of parental WM35 cells and their tumorigenic derivatives
`ously (33). The expression profile of MUC18 antigen, however, was
`under nonrestrictive (5% serum) or restrictive conditions (0.5% se—
`dramatically altered in tumorigenic cell lines derived from WM35
`3080
`
`[NAHUAIAAAIAA I—Am na—Il—In—wvnr‘ ”Any: ...... I nnnnnn HAAAMkA» I3 404—} R 4nnr: AMAAAAV‘ AARAA:A‘:AV~ ‘AV i" AAAAA
`
`6
`
`

`

`DOMINANT MAUGNANT CONVERSION 1N MELANOMA
`
`rum). As illustrated in Fig. 5.4, no differences in growth rates were
`observed between the parental WM35 cells and the three independent
`tumor-derived variant cell lines T35—4.l, TBS-4.2, and TBS-3.1 when
`they were cultured in the presence of 5% fetal bovine serum. How-
`ever, in the presence of 0.5% serum, very significant differences were
`detected. All three tumorigenic variants were able to grow under these
`conditions, albeit at a very slow rate. In contrast, the parental WM35
`cells not only failed to grow but in fact rapidly decreased in number,
`apparently losing their viability (Fig. SB). Unlike the variant cell lines
`(Fig. 50, the parental WM35 cells also gradually lost their spindle-
`shaped morphology and began to round up and detach from the dish
`(Fig. 5D).
`The WM35 Tumorigenic Variants Acquire a Multicytokine-
`resistant Phenotype. It has only been shown recently that some of
`the aggressive growth properties of melanoma cells in viva may be
`linked to the acquisition of resistance to multiple gr