American Journal of Pathology, Vol. 144, No. 4, April 1994
`Copyright © American Societyfor Investigative Pathology
`
`Variant Sublines of Early-Stage Human
`MelanomasSelected for Tumorigenicity in Nude
`Mice Express a Multicytokine-Resistant
`Phenotype
`
`Hiroaki Kobayashi, Shan Man,
`John R. MacDougall, Charles H. Graham,
`Chao Lu, and Robert S. Kerbel
`From the Division of Cancer Research, Sunnybrook Health
`Science Centre and Departments ofMedical Biophysics and
`Molecular & Medical Genetics University of Toronto,
`Ontario, Canada
`
`interleukin-6, interleukin-1 and tumor necrosis
`factor-a, and did so in a stable manner. Thus the
`results support the bypothesis that a progressive
`multicytokine resistance accompanies the pro-
`gression of buman melanoma. The availability of
`such related sublines should provide a valuable
`resource to help study the changes associated
`with, andperbaps causative of, disease progres-
`sion in human malignant melanomas.
`(Am J
`Pathol 1994, 144:776—786)
`
`Surgical removal of early-stage radial growth
`phase orvertical growth phase primary cutane-
`There are a numberof ways in which growth factors
`ous human melanomas usually results in cure of
`the disease. Hence there are few examples of
`can contribute directly to the proliferative advantage
`of cancer cells and tumor progression. These include
`genetically-related paired human melanoma cell
`the production of stimulatory autocrine growth fac-
`linesfor study in which one member ofthepairis
`JSrom a curable early-stage lesion and the partner
`tors,
`increased sensitivity to mitogenic paracrine
`growth factors, and acquisition of resistance to ex-
`is a more aggressive malignant variant. A rapid
`method ofobtaining such variants is described. It
`ogenous(paracrine) or endogenousinhibitory growth
`factors.’ With respectto the last possibility, the best
`consists of injecting cellsfrom established early-
`known paradigm is the tendencyof cells from most
`Stage radial growth phase or vertical growth
`phase melanoma cell lines—which are normally
`types of carcinoma or leukemia to expressresistance
`in vitro to the growth inhibitory effects of members of
`non- or poorly tumorigenic in nude mice—into
`such hosts, where the cell inoculum is co-mixed
`the transforming growth factor-B (TGF-B) family, such
`as TGF-B1.?:3 In our laboratory, we have been study-
`with Matrigel, a reconstituted basement mem-
`brane extract. This resulted in the rapidforma-
`ing loss of responseto negative growth factors in hu-
`tion ofprogressively growing solid tumors from
`man melanoma. Two
`important
`findings have
`which permanent cell lines were established.
`emerged from these studies. First, the number of
`Subsequently,
`the sublines were found to be
`growth factors or cytokines that may function as po-
`frankly tumorigenic upon_retransplantation
`tential growth inhibitors of melanocytes or melanomas
`
`into new nude mousebosts in the absence ofMa-
`trigel co-injection. This process was repeated a
`secondtime, resulting in the isolation of second-
`ary sublines, manifesting a stepwise increase in
`tumorigenic properties. The tumorigenic variant
`sublines were examined for their relative sensi-
`tivity to apanel ofdifferent cytokines that are nor-
`mally growth inhibitoryfor melanoma cellsfrom
`early-stage primary lesions. All the sublines were
`found to express an increased resistance to
`the cytokines transforming growth factor-B,
`
`Supported by grants from the National CancerInstitute of Canada,
`the National Institutes of Health (CA-41233) and the Share Founda-
`tion. HK, JRM, and CLare recipients of a fellowship or studentship
`from the Medical Research Council of Canada. CHGis a Fellow of
`the National CancerInstitute of Canada and RSKis a Terry Fox Ca-
`reer Scientist of the National CancerInstitute of Canada.
`
`Accepted for publication November 30, 1993.
`Address reprint requests to Dr. Robert S. Kerbel, Division of
`Cancer Research, Reichmann Research Building, S-218, Sunny-
`brook Health Science Centre, 2075 Bayview Avenue, Toronto, On-
`tario M4N 3M5, Canada.
`
`Genentech 2086
`Hospira v. Genentech
`IPR2017-00737
`
`776
`
`Genentech 2086
`Hospira v. Genentech
`IPR2017-00737
`
`

`

`777
`Multicytokine Resistance in Variants of Early-Stage Melanoma
`AJP April 1994, Vol. 144, No. 4
`
`may be quite large and include not only TGF-B,* as
`expected, but also interleukin-1a@ (IL-1a@), IL-6, tumor
`necrosis factor-a (TNF-a) and oncostatin M.° Second,
`resistance to such factors is frequently manifested in
`a seemingly progressive, stepwise manner during
`disease progression.*-”
`Our studies were made possible by virtue of the
`fact that most human melanomas become more ma-
`
`lignant by progression through a series of clinically
`and pathologically well-defined stages of disease.?-?
`Moreover,cell lines can be established from all these
`stages; such lines can then be compared to normal
`melanocytesor atypical nevus derived melanocytes.
`The first stage of tumorigenic melanoma growth is
`the so-called radial growth phase, (RGP) wherein
`primary melanomas grow mainly in the epidermis in
`a plaquelike (horizontal) manner.'° These lesions are
`always curable and seem to consist of tumor cells
`incompetent for metastasis. The next phase is the
`so-called vertical growth phase (VGP); such primary
`tumors begin to grow downward into the underlying
`mesenchymally-derived dermis.®:° VGP lesions less
`than 0.76 mm in thickness (thin VGP) are usually
`curable (about 90 to 95% of the time) by surgery, and
`like RGP lesions, are believed to be comprised
`mainly of metastatically incompetent melanoma
`cells.2° However,
`the thicker VGP lesions (‘“thick”
`VGP) carry a much worse prognosis and generally
`behave genotypically and phenotypically similar to
`the lesion of distant metastasis.°:° Therefore, RGP
`and thin VGP primary tumors may be considered
`generally as early-stage melanoma, whereas thick
`VGP primary tumors and distant metastases can be
`considered as advanced-stage melanoma.
`The availability of a large number of cell lines es-
`tablished from different stages of melanoma progres-
`sion has proven to be an invaluable tool
`to study
`many aspects of tumor biology as a function of hu-
`man tumor progression.'°-'? This model neverthe-
`less suffers from one inherentlimitation: once an au-
`
`thentic early-stage, primary RGP or early VGP lesion
`has been removed, there is, by definition, rarely a
`follow-up recurrent metastatic tumor that can bere-
`moved from the same patient and studied in a com-
`parative mannerwith the initial, genetically related,
`lesion. To circumvent this problem, one can use a
`large number of independentcell lines from the dif-
`ferent disease stages, but obtained from different
`patients,
`to study meaningfully the sequential bio-
`logical changesthat occur in melanoma cells during
`tumor progression. However, the true relative malig-
`nant status of such cell lines at the time of study is
`frequently impossible to evaluate.
`
`The purpose of the present study was twofold: 1)
`to develop a methodof isolating genetically related
`variants of early-stage RGP or VGP primary mela-
`nomasthat behave in a much more aggressive fash-
`ion in vivo, and 2) to use theserelated cell lines to
`evaluate further the hypothesis that progression of
`melanoma is associated with increasing, stepwise,
`resistance to multiple independent cytokines—a
`phenomenon we have called multicytokine resis-
`tance.’ The strategy we adopted wasto inject early-
`stage melanoma cells into nude mice using the Ma-
`trigel-assistance method. '* This is based on the fact
`that certain types of human cancers do not form
`tumors readily, or at all, when injected into nude
`mice, but will do so when the cells are co-injected
`with Matrigel, a reconstituted basement membrane
`extracellular matrix extract.'* This includes, for ex-
`ample, prostatic carcinoma, breast carcinoma, reti-
`noblastoma, and small cell
`lung cancer.'%15-19 In
`this regard, early-stage RGP or VGP human melano-
`mas are frequently non- or poorly tumorigenic in
`nude mice'®:'' (and Dooley T, personal communica-
`tion), but we reasonedthatinjection of such cells with
`Matrigel might give rise to progressively growing
`solid tumors. Cells from these could then be estab-
`
`lished in culture and subsequently evaluated for tu-
`morigenic capacity (in the absence of Matrigel) and
`for their sensitivity to a number of inhibitory cyto-
`kines.
`
`Materials and Methods
`
`Cell Culture and Establishment ofin
`Vivo-Passaged Lines
`
`All human melanomacell lines usedin this study were
`maintained in RPMI 1640 (Gibco Laboratories, Grand
`Island, NY) supplemented with 5% fetal bovine serum
`(Hyclone Laboratories, Logan, UT). Three human
`melanomacell lines (WM35, WM1341B, WM793)es-
`tablished from early-stage primary lesions by Herlyn
`et al,'° were used in these studies. The clinical and
`pathological characteristics of each line are de-
`scribed elsewhere.'°:'' Under the conditions we
`
`have employed, two of these cell lines (WM35 and
`WM1341B) are insufficiently tumorigenic to be able to
`give rise to progressively growing lethal tumors when
`injected subcutaneously (s.c.) into athymic nude
`mice (see below for details). °
`In vivo selection of tumorigenic sublines were es-
`tablished from these three cell lines as follows (see
`Figure 1 for summary using WM35as an example): in
`the case of the WM35cell line, 2 x 10© cells were
`
`

`

`Kobayashiet al
`778
`AJP April 1994, Vol. 144, No. 4
`
`WM 35
`(2x10° cells)
`
`+MG
`
`-MG
`
`LS
`
`LS
`
`|
`82 days |
`* Ss SL
`
`leeee!)
`
`leees!|
`
`leeee!
`
`35-P1-N1
`
`35-P1-N2
`
`35-PI-N3
`
`(2x10° cells)
`
`+MG
`
`-MG
`
`35-P2-N3
`35-P2-N2
`35-P2-N1
`Figure 1. Flow diagram summarizing the establishment of in vivo-
`passaged sublines derived from WM35, an early-stage human mela-
`nomacell line. MG: Matrigel. Solid circles represent growing tumors
`in nude mice or tumorcells in culture dishes. An identical procedure
`was used to generate variantsfrom the WM1341B and WM793 early-
`stage melanomacell lines. The tumor volumes (mean * SD) andpe-
`riod of time after inoculation, at which tumors were excised to estab-
`lish each in vivo-passaged sublines, are: 35-P1 sublines, 3057 +
`1139 mm?and 82 days; 35-P2 sublines 2075 + 646.7 mm? and 70
`days; 1341-P1 sublines, 458.6 + 148.7 mm? and 95 days; 1341-P2
`sublines, 1725 + 569.7 mm? and 70 days; 793-P1 sublines, 1208 +
`280.6 mm? and 72 days; 793-P2 sublines, 3125 + 221.8 mm? and
`46 days.
`
`suspendedin 0.1 ml of cold serum-free RPMI 1640
`and mixed with an equal volumeof cold Matrigel (10
`mg/ml) (Collaborative Research, Bethesda, MD). The
`resulting suspension (0.2 ml of cell suspension con-
`taining 5.0 mg/m! Matrigel) was immediately injected
`s.c. into the right flank of 5- to 8-week-old female athy-
`
`mic (Swiss) nude mice (Harlen Sprague Dawley). This
`resulted in progressive tumor growth, as summarized
`in the Results section. Eighty-two daysafter inocula-
`tion, first-passage cell lines (designated 35-P1-N1,
`35-P1-N2, and 35-P1-N3 where “P1” stands forfirst
`passage and “N”the individual mouse number) were
`adapted to culture by mechanical and enzymatic dis-
`aggregation from three distinct tumors (3057 + 1139
`mm? mean + SD tumor volume) growing progres-
`sively in independentanimals.A total of 2 x 10® cells
`prepared by equal mixing ofthree first-passagelines
`were co-injected with Matrigel
`into animals in the
`same manner as discussed above. Second-passage
`(P2) lines (designated, for example, as 35-P2-N1, 35-
`P2-N2, and 35-P2-N3) were established from three
`distinct tumors (2705 + 646.7 mm? mean + SD tumor
`volume) 70 daysafter the inoculation.
`Similarly, as described in Figure 1, we established
`the first-passage lines of WM1341B (designated
`1341-P1-N1,
`1341-P1-N2,
`and
`1341-P1-N3) or
`WM793 and the second-passagelines of WM1341B
`(that is, 13841-P2-N1, 1341-P2-N2, and 1341-P2-N3)
`or WM793. The mean tumor volumes and period of
`time after inoculation, at which tumors were excised
`to establish each in vivo-passagedline, are summa-
`rized in the legend of Figure 1. All of these passaged
`variant P1 and P2 sublines were considered as es-
`
`tablished cultured cell lines after being passaged in
`monolayerculture at least 10 times during a period of
`6 weeks and used for experiments.
`
`In Vivo Tumor Growth Assay
`
`To examinethe effect of Matrigel on tumorigenicity of
`early-stage human melanomacells in athymic nude
`mice, various numbers of cultured melanomacells
`were inoculated s.c. with Matrigel, as described
`above. Control mice were given a s.c. injection of the
`same numberof cells in 0.2 ml of serum-free RPMI
`
`in the absence of Matrigel. Moreover,
`1640 media,
`various numbersof cells prepared by equal mixing of
`first-passage sublines (P1-N1, P1-N2, and P1-N3)
`were also inoculated without Matrigel to evaluate the
`effect of in vivo passages on the tumorigenic prop-
`erties ofall lines. Mice were monitored up to 3 months
`after inoculation and, if progressively growing tumors
`were observed, the diameters of tumors were meas-
`ured with calipers to calculate the tumor volumefrom
`the (length x width?/2) as described by Geranetal.?°
`At the conclusion of the measurement period, mice
`were sacrificed and their tumors were excisedforhis-
`
`tological analysis.
`
`

`

`779
`Multicytokine Resistance in Variants of Early-Stage Melanoma
`AJP April 1994, Vol. 144, No. 4
`
`Southern Blot Fingerprinting Analysis
`
`Wewished to confirm the human genetic origin of the
`in vivo-passaged variants, and the absenceof sig-
`nificant contamination by mouse-derived DNA ac-
`quired during the tumor growth in nude mice. To do
`this, genomic DNA of each passaged subline
`adapted to culture from its respective growing tumor
`after inoculation with Matrigel was compared with that
`of the parental cultured cell lines by Southern blot
`analysis using mouse or humanspecific hybridization
`DNA probes. Genomic DNA wasextracted from the
`parental and in vivo-passaged lines by lysis in 1%
`sodium dodecy! sulphate with subsequent protein
`and RNA digestions. As a control, mouse DNA was
`extracted from spleen tissue obtained from athymic
`nude mice. The resulting DNA (5 ug) was digested
`with EcoRI and size-fractionated in 0.8% agarose
`gels. The separated fragments were immobilized
`onto Zeta-probe blotting membrane (Bio-Rad Labo-
`ratories, Richmond, CA) and membrane probing was
`carried out essentially as outlined elsewhere.?! After
`hybridization with mouse DNAprobe, the membrane
`was exposed to x-ray film (Kodak). Hybridization
`using human DNA probe was performed with the
`same membraneafter stripping off the mouse DNA
`probe.
`Probes were prepared using the random priming
`method (Pharmacia, Uppsala, Sweden). The mouse
`probe consisted of random primed radiolabeled
`Balb/c genomic DNA digested with BamHI, EcoRI
`and Hinaill restriction enzymes. The human probe,
`cut from plasmid P17H8, consisted of random primed
`radiolabeled DNAspecific to the a-satellite sequence
`on chromosome 17, as has been previously de-
`scribed.?2
`
`Growth Assay to Assess Inhibitory
`Cytokine Sensitivity
`
`The effects of various cytokines on the growth of each
`cell line were examined by [H]thymidine incorpora-
`tion assay as described previously.°:23 Five thousand
`cells were seeded in each well of 96-well tissue cul-
`ture plates (Nunc, Roskilde, Denmark) with 100 ul of
`ExCell 300 medium (JRH Biosciences, Lenexa, KS)
`supplemented with 1% fetal bovine serum and al-
`lowed to attach overnight. Human recombinant cy-
`tokines were prepared at various concentrations in
`ExCell 300 medium supplemented with 1% fetal bo-
`vine serum and 100 ul of this solution was added to
`each well. Forty-eight hours later, cells were pulsed
`with cell containing 2 uCi of [SH]thymidine in 50ul of
`serum-free ExCell medium for 4 to 6 hours and then
`trypsinized. The DNAwasharvested ontofilter mats
`(LKB Wallac, Turku, Finland) and [SH]thymidine in-
`corporation was measured with a Betaplate liquid
`scintillation counter (LKB Wallac). Cytokines (pur-
`chased from UBI, Lake Placid, NY) and their concen-
`trations used were IL-6 (0.5-25 ng/ml), TGF-61
`(0.25-5 ng/ml),
`IL-1@ (0.5-5 ng/ml), and TNF-a
`(0.5-25 ng/ml). [H]thymidine incorporation in con-
`trols without added factors was considered as 100%.
`
`Results
`
`Effect of Co-Injection with Matrigel on
`Tumor Growth of Early-Stage Melanoma
`Cell Lines
`
`After tumorcell inoculation, all animals were observed
`for 3 months, or until their tumors reached a maximum
`of approximately 2 cm in diameter. Table 1 showsthe
`
`Table 1. Effects ofMatrigel and in vivo Passage on Tumorigenicity in Nude Mice of Early-Stage Human
`Melanoma Cell Lines
`
`Numberof
`inoculated
`WSS
`35-P1
`WM1S416
`w34i-p1
` ___—SWM793
`793.4
`
`
`
`
`
`cells -MG=-MG“MG +MG +MG _MG “MG +MG -MG
`
`
`
`
`
`Tumor-take incidence”
`
`5/5
`5/5
`3/5
`5/5
`5/5
`o/5
`5/5
`5/5
`4/10
`2 x 108
`5/5
`5/5
`2/5
`5/5
`5/5
`O/5
`5/5
`5/5
`0/5
`1 x 108
`ND
`ND
`NDt
`5/5
`5/5
`O/5
`5/5
`5/5
`0/5
`5 x 108
`4/5
`5/5
`1/5
`4/5
`5/5
`O/5
`3/5
`5/5
`0/5
`+x 108
`2/5
`4/5
`o/5
`4/5
`5/5
`ND
`3/5
`4/5
`ND
`1x 104
`
`1x 103 1/5 ND 2/5 1/5 ND 5/5 1/5 O/5 4/5
`
`
`
`
`
`
`
`
`
`* Tumor development and growth was observed until 3 months after s.c. inoculation of various numbers of parental or in vivo-passaged
`(P1) cell
`lines into nude mice in the presence (+MG) or absence (—MG) of Matrigel. The cell lines designated as 35-P1, 1341-P1, and
`793-P1 refer to thefirst-passage sublines derived from primary tumors after inoculation of the parental WM35, WM1341, or WM793, respec-
`tively, with Matrigel, as described in Materials and Methods.
`TND, not determined.
`
`

`

`Kobayashiet al
`780
`AJP April 1994, Vol. 144, No. 4
`
`tumorigenic profiles of several parental early-stage
`melanoma lines or in vivo-passaged sublines (first
`passage), injected with or without Matrigel into nude
`mice. It is obvious by comparing the tumor-take in-
`cidence of parental (nonpassaged) tumorlines, with
`or without co-injection of Matrigel, that Matrigel sig-
`nificantly enhanced the tumor-formingability ofall the
`cell lines. In particular, of the three cell lines, the tu-
`morigenicity of WM1341B line showed the highest en-
`hancement by Matrigel: remarkably, a 100% tumor-
`take incidence was obtained bythe s.c. inoculation of
`no more than 10° tumorcells co-injected with Matri-
`gel, even though no tumors developed after 2 x 10°
`tumor cells were inoculated without Matrigel. With
`both WM35 and WM/793lines, the tumor-take inci-
`denceby the inoculation of 10% cells with Matrigel
`(40% and 80%, respectively) was higher than that by
`the injection of 2 x 10© cells without Matrigel (10%
`and 60%, respectively).
`
`Morphological and Genetic Preservation
`of Tumor Characteristics during in Vivo
`Passages
`
`By hematoxylin and eosin staining, all tumors derived
`from a parental WM35line inoculated s.c. with or with-
`out Matrigel and from first-passage WM35 lines
`
`grownin the absenceof Matrigel showed similar mor-
`phological appearances, typical of human melanoma
`(Figure 2, A to C). Similarly, neither WM793 tumors
`arising in Matrigel-injected animals nor tumors de-
`rived from in vivo-passage (P1) sublines showed a
`significant change in morphological pattern com-
`pared to the parental WM793-derived tumor in the
`absence of Matrigel (Figure 2, D to F).
`Figure 3 shows a DNAfingerprinting assay by
`Southern blot analysis using mouse (Figure 3A) or
`human (Figure 3B) specific DNA probes. The DNA
`sample derived from mouse spleen tissue probed
`with random primed mouse genomic DNAresulted in
`a smear consisting of numerous bands (Figure 3A,
`lane 1). On the other hand, no bands with the mouse
`probe were seen in the lanes loaded with DNA
`samples from the cultures of in vivo-passaged (P1)
`sublines as well as the parental early-stage human
`melanomacell lines (Figure 3A, lanes 3 to 14). When
`probed with a random primed human DNA probe
`specific to the a-satellite sequence on chromosome
`17 (Figure 3B), P1 sublines of WM793 (lanes 4 to 6)
`and WM35(lanes 8 to 10) showed exactly the same
`patterns of bands as their parental WM793(lane 3)
`and WMS5(lane 7) cell lines, respectively.
`/n vivo-
`passaged WM1341B sublines (lanes 12 to 14) dis-
`played similar but not identical patterns of bands
`when compared to the parental WM1341B cell line
`
`
`
`
` ite
`
`eft
`eft
`iio
`Figure 2. Histological appearance of tumortissues derivedfrom parental (nonpassaged) human melanomacell lines and their in vivo-passaged
`(P1) sublines, inoculated with or without Matrigel. Some tumor specimens were obtained after s.c. inoculation of 2 X 10° cells ofparental early-
`stage melanomacell lines with (WM35: B; WM793: E and H; WM1341B: \) or without (WM35: A; WM793: D) Matrigel. Others were obtained by
`injection of 2 X 10° cells offirst-passage (P1) sublines in the absence of Matrigel (35-P1: C and G; 793-P1; F). A to F: hematoxylin and eosin
`staining. Note that these three distinct tumors derived from WM35 (A to C) or WM793 (D to F) cell lines show a similar morphological pattern
`typical of human melanoma.
`
`

`

`781
`Multicytokine Resistance in Variants of Early-Stage Melanoma
`AJP April 1994, Vol. 144, No. 4
`
`A
`
`3
`
`123 45 6 7
`
`8 91011121314
`8 91011121314
`
`123 45 6 7
`
`Figure 3. DNA fingerprinting profiles of early-stage human mela-
`nomacell lines andtheir in vivo-passaged variants. DNA samples ex-
`tractedfrom monolayercultures ofparental (nonpassaged) cell lines
`and in vivo-passaged sublines were probed with random primed
`mouse genomic DNA (A). The same membrane was used for hybrid-
`ization with random primed human DNA specific to the a-satellite se-
`quence on chromosome 17 after stripping off the mouse DNA probe
`(B). Lane 1: mouse DNA extracted from spleen tissue obtained from
`athymic nude mice (control); lane 2: empty; lane 3: parental WM793
`line; lanes 4 to 6: first-passage sublines of WM793 (793-P1-N1, 793-
`P1-N2, and 793-P1-N3,
`respectively); lane 7: parental WM35 line;
`lanes 8 to 10.first-passage sublines of WM35 (35-P1-N1, 35-P1-N2,
`and 35-P1-N3, respectively); lane 11: parental WM1341B line: lanes
`12 to 14: first-passage sublines of WM1341B (1341-P1-N1, 1341-P1-
`N2, and 1341-P1-N3, respectively).
`
`(lane 11): in the sublines severaldifferent sized bands
`were apparent, which suggested some DNA-
`rearrangements had occurred during in vivo pas-
`sages. However, the pattern observed in lane 12
`(1341-P1-N1) indicated that this cell line may be a
`cross-contaminant of WM35 or WM793cells, or one
`of their selected sublines. Because of its equivocal
`origin, this cell line was not includedin further studies.
`Thus, by using mouse or human DNAprobes, we con-
`
`firmed that the in vivo-passaged sublines were in-
`deed humanin origin, and in all likelinood, were de-
`rived from the parentalcell lines used to inoculate the
`mice and were not contaminated by host-derived
`mouse DNA.It is also apparent that this procedure
`may be useful
`in picking up occasional cross-
`contaminating cell lines.
`
`Evidence of Tumor Progression during in
`Vivo Passages: Increased Tumor Growth
`Properties in Passaged Cell Lines
`
`By comparing the tumor-take incidence between pa-
`rental (nonpassaged) early-stage melanoma lines
`and in vivo-passagedsublinesafter s.c. inoculation in
`the absence of Matrigel, it was apparent that in vivo-
`passage resulted in the emergence of sublines hav-
`ing a highly elevated tumor-forming ability and that
`this was independent of the promoting effect pro-
`vided by Matrigel (Table 1). During the 3-month pe-
`riod of observation, at least 2,000-fold and 100-fold
`fewercells of first-passaged lines were required to
`produce tumors with similar tumor-take incidence by
`parental WM35, WM1341B, and WM793cell lines, re-
`spectively (see Table 1).
`Because a 100% tumor-take incidenceof each cell
`line was obtained 3 months after s.c. inoculation of
`10° tumor cells with Matrigel, we compared tumor
`growth curves among parental and in vivo-passaged
`cell lines. As shownin Figure4, all of in vivo-passaged
`variants acquired highly enhanced tumor-forming
`ability. Greater tumor growth wasseenasthe in vivo-
`passage numberincreased, especially in the case of
`aseries of WM1341B cell lines: the tumor volume dou-
`bling time of
`tumors derived from the parental
`WM1341B cell line was 29.3 days; for 1341-P1 sub-
`lines, 18.7 days; for 1341-P2 sublines, 12.8 days; for
`the parental WM35cell line, 17.5 days; for 35-P1 sub-
`lines, 13.6 days; for 35-P2 sublines, 11.9 days; for the
`parental WM793cell line, 17.5 days; for 793-P1 sub-
`lines, 11.2 days; and for 793-P2 sublines, 10.0 days.
`
`Multicytokine Resistance in Human
`Melanoma Cells Is Acquired through in
`Vivo Passages
`
`the growth of
`We have previously reported that
`WM35early-stage melanomacell line is inhibited in
`vitro by IL-6, IL-1a, TNF-a, TGF-81, and oncostatin
`M, whereas the majority of advanced stage human
`melanomacell lines we examined were found to be
`partially or completely resistant to these inhibitory
`effects.4-© Therefore,
`the sensitivity to these cyto-
`
`

`

`Kobayashiet al
`782
`AJP April 1994, Vol. 144, No. 4
`
`10000
`
`WM35
`
`WM1341B
`
`WM793
`
`-_-“-«~
`
`“EE
`
`— 1000
`®
`Ee
`
`100
`
`=O
`
`o
`
`>5
`
`oO
`Ee
`>
`
`
`
`Ee
`
`17.54days
`
`29.28days
`
`17.48dayVr
`
`11.87 days
`
`18.73 days
`
`x
`
`12.84 days
`
`%
`
`11.24 days
`
`10.00 days
`
`Weeksafter Inoculation
`Figure 4. Effects of in vivo passage on the tumor growth properties of early-stage human melanomalines in athymic nude mice. 0.1 ml cell sus-
`pension containing 1 million cells was injected with the same volume of Matrigel (10 mg/ml) into five nude mice per group. All mice developed
`tumors and results represent the mean tumor volume + SEM calculated as described in Material and Methods. O: parental line: ®@, first-passage
`line (P1); AQ: second-passage line (P2). A: WM35 cell lines; B: WM1341B cell lines; C: WM793 cell lines.
`
`kines (with the exception of oncostatin M) was com-
`line (P1 and P2 lines) manifested a significantly lower
`pared between parental WM35 cells and in vivo-
`sensitivity to 5 ng/ml IL-6, comparedto the sensitivity
`passaged sublines (P1 and P2 lines) by using the
`of the parental WM1341B line (Figure 6A). Mostof the
`[H]thymidine incorporation assay. Shown in Figure
`passaged WM1341B sublines, except 1341-P2-N1,
`5 is a representative result for each cytokine from
`were statistically more resistant to 5 ng/ml TGF-B1
`than the original WM1341B line (Figure 6B). At other
`several tests using various concentrations of IL-6
`(0.5-2.5 ng/ml), TGF-B1 (0.5-2.5 ng/ml), IL-1e@ (0.5-
`concentrations of IL-6 (1-25 ng/ml) or TGF-81 (0.5-5
`2.5 ng/ml), and TNF-a (0.5-5 ng/ml). All of the P1 and
`ng/ml), similar resistance to each cytokine was seen
`
`P2 sublines showedastatistically significant in-
`in the in vivo-passaged WM1341B sublines (data not
`crease of [SH]thymidine incorporation after 48 hours
`shown).
`exposure to 2.5 ng/ml of IL-6 (Figure 5A), TGF-B1
`Because IL-6 has no inhibitory effects on the
`growth of the WM793line,° TGF-B1, IL-1a, and TNF-a
`(Figure 5B), and IL-1a (Figure 5C) or 1 ng/ml of TNF-a
`(Figure 5D) as comparedwith that of parental WM35
`were used to check for the acquired cytokine resis-
`cell lines. Data from experiments using other concen-
`tance bythe in vivo-passaged WM793 sublines (P1
`trations of
`the four cytokines showed the dose-
`and P2 lines) (Figure 6). After 48 hours exposure to
`responseinhibitory effects of cytokines to each line,
`0.5 ng/ml TGF-B1,
`four of six in vivo-passaged
`but the differential cytokine sensitivity between pa-
`WM793 sublines showed a significant increase in
`[SH]thymidine incorporation rates, as compared with
`rental and in vivo-passagedcell lines was consistent
`with the data shown here.
`that of the parental WM793line (Figure 6C). Statisti-
`The WM1341B cell line is not sensitive to IL-1a@ and
`cally significant resistance to 1.25 ng/ml IL-1a was
`TNF-a, but is inhibited by exposure to IL-6° and TGF-
`seen in four of six in vivo-passaged WM793sublines
`B1.4 Hence, the changein sensitivity to cytokines dur-
`(Figure 6D). At other concentrations of TGF-B (0.25 to
`ing in vivo passages was examined using IL-6 and
`2.5 ng/ml) or IL-1@ (0.5 to 5 ng/ml), we obtained com-
`TGF-B1 (Figure 6). Similar to the WM35cell line and
`parable results (data not shown). In the case of ex-
`its sublines, every in vivo-passaged WM1341B sub-
`posure to 2.5 to 25 ng/ml of TNF-a, the differential
`
`

`

`783
`Multicytokine Resistance in Variants of Early-Stage Melanoma
`AJP April 1994, Vol. 144, No. 4
`
`IL-1a, 2.5ng/ml
`
`TNF-a 1ng/ml
`
`120
`
`100
`
`80
`
`60
`
`40
`20
`
`0
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`TAF2.5ng/ml
`
`°
`
`80
`
`60
`
`#0
`
`0
`
`o 2 $ 2 z > 2
`S ee
`oak
`me n Nn
`nN
`= oO o aaqdeaa
`© ©) 6 wo
`wo
`wo
`oo 9 oO mo
`
`o Z $ 3 Zz 2 2
`Ss
`24
`a
`7
`7 a a
`= a Oo.
`0.
`al aa
`obo 6 6 wo Yo
`oO HO
`HO 0
`
`sensitivity was not as obvious between in vivo-
`passaged sublines and parental WM793line (data
`not shown).
`
`Discussion
`
`The results of our studies contain two potentially im-
`portant findings. First, it is possible to obtain tumors
`rapidly in nude mice from injections of early-stage
`human melanomacell lines, provided the cells are
`co-injected with Matrigel; moreover, cell lines estab-
`lished from these tumors readily engraft and grow
`progressively wheninjected into new nude mousere-
`cipients in the absence of Matrigel. Second, virtually
`all the tumorigenic variants obtained from different
`early-stage melanomas we examined manifested in-
`creased resistance to the growth-inhibitory effect of a
`panel of cytokines, including TGF-B1, TNF-a, IL-1a,
`and IL-6. Thus, the results support the hypothesis that
`acquisition of multicytokine resistance” accompanies
`human melanomaprogression. Expression of such a
`phenotype maybe aneffective way for tumors to ac-
`
`quire a growth advantage and giverise to tumor sub-
`populations of increasing aggressiveness.’
`With respect to our Matrigel results, injection of hu-
`man tumorcells with this extract seems to be an ef-
`
`fective procedureto obtain tumor growth in nude mice
`using various types of tumor that are normally very
`poorly tumorigenic, or completely unable to grow in
`such hosts. This wasfirst discovered by Fridmanetal
`using human small cell carcinoma cell lines.'%:'° It
`has since been confirmed by others using humantu-
`mors
`such as prostate cancer,'’:'® retinoblas-
`toma,'®'° and breast cancer.'? Whereas malignant
`melanomasare normally among the most tumorigenic
`of all human cancersin nude mice, this does not apply
`to cell lines from the early-stages of this disease. '°
`Thus, Herlyn and his colleagues have reported that
`cell lines established from RGP or early VGP primary
`tumors (removed from the patients who did not ex-
`perience recurrent disease, ie, were Cured by surgi-
`cal excision of the tumors), do not generally grow
`readily when injected into nude mice.'° A similar ob-
`servation has been noted by Dooley using six newly
`
`420
`
`IL-6 2.5ng/ml
`
`100
`
`80
`
`60
`
`=>‘
`Oo
`or
`Cc
`oO
`oO
`Ee
`° 40
`o~
`— 20
`@O
`
`x S
`
`— o
`
`c
`.
`D jo0]
`@O
`& 400
`/Hithymidine incorporation ( growth)
`Figure 5.
`assay for cytokine sensitivity. Cells were plated —
`at 5X 10? cells/well in 96-well plates, allowed
`and
`to attach overnight and incubated with or
`Ee
`without various concentrations of cytokines
`(0.5 to 2.5 ng/mlof IL-6, TGF-B1, and IL-la or >
`0.5 to 5 ng/ml of TNF-a). After 48 bours, [3H]-
`<=
`thymidine incorporation into DNA was meas-
`'
`ured. Representative data are shown at concen-
`—_—
`trations of 2.5 ng/mlof IL-6 (A), TGF-B1 (B),
`and IL-1a (C) or 1 ng/ml of TNF-a (D). Data L
`are expressed as mean + SEM oftriplicate de- QU 20
`terminations compared with the controls un-
`treated of each cell line (which was considered
`as 100%). WM35: parental early-stage human
`melanoma cell line; 35-P1-N1,
`-N2 and -N3:
`first-passage sublines; 35-P2-N1, -N2, and -N3:
`second-passage sublines. Probability was calcu-
`lated by Student's t-test. Parental WM35line vs
`each passaged subline in A, B, C and D, P <
`0.01.
`
`

`

`Kobayashiet al
`784
`AJP April 1994, Vol. 144, No. 4
`
`80
`
`60
`40
`
`20
`
`0
`
`— 120
`o
`100
`=
`Cc
`Oo
`oO
`eee
`Oo
`SS
`—
`@®
`aw
`40}
`—
`e
`~
`®
`Cc
`—
`‘-
`£
`~
`££
`-
`'
`oa
`40
`L
`oO
`— 20
`
`120
`
`4100
`
`80
`
`60
`
`IL-6 Sng/ml
`
`a
`A
`Tar-f) Bipngyrn
`
`120
`
`100
`
`80
`
`60
`40
`
`20
`
`0
`
`120
`
`100
`
`80
`
`60
`
`40
`20
`
`0
`
`TGF-B1 0.5ng/ml
`
`4
`Illa tseengie
`
`/Hithymidine incorporation assay
`Figure 6.
`for cytokinesensitivity. Cells were plated at 5 X
`10° cells/well in 96-well plates, allowed to at-
`tach overnight and incubated with or without
`various concentrations of cytokines (1 to 25
`ng/ml IL-6 and 0.5 to 5 ng/ml TGF-B1 for the
`WM1341B cell lines; 0.25 to 2.5 ng/ml TGF-B 1,
`and 0.5 to 5 ng/ml IL-la for the WM793 cell
`lines). After 48 hours, PH]thymidine incorpora-
`tion into DNA was measured. Representative
`data are shownat concentrations of 5 ng/ml of
`IL-6 (A) and TGF-B1 (B) for WM1341B cell
`lines and 0.5 ng/ml TGF-B1 (C) and 1.25
`ng/ml IL-1 (D) for WM793cell lines. Data are
`expressed as mean + SEMoftriplicate determi-
`nations compared with the controls of each cell
`line (considered as 100%). WM1341B and
`WM793: parental early-stage human mela-
`noma cell lines; 1341-P1-N1,-N2,-N3 and 793-
`P1-N1,-N2,-N3: first-passage sublines; 1341-P2-
`N1,-N2,-N3 and 793-P2-N1,-N2,-N3: sec