[CANCER RESEARCH 48, 731-740, February I, 1988]
Biological and Karyotypic Characterization of a New Cell Line Derived from
Human Gliosarcoma1
Manfred Westphal,2 Marianne Haensel, Dieter Mueller, Rudolf Laas, Regina Kunzmann, Eva Rohde,
Albrecht Koenig, Fritz Hoelzel, and Hans-Dietrich Herrmann
Laboratory for Brain Tumor Biology, Department of Neurosurgery {M. W., M. H., D, M,, E, R., A. K,, H-D. H.], Department of Neuropathology [R. L.J, and Department
of Physiological Chemistry [R, K., F. H.], University Hospital Eppendorf, Martinistrasse 52, 2000 Hamburg 20, Federal Republic of Germany
ABSTRACT
A cell line (NCE-G28) was established from the biopsy material of a
human gliosarcoma of low histológica!differentiation. The initial cultures
showed a mixed population of cells which in later stages became more
uniform due to loss of slower growing constituents. The cells have been
growing steadily for 20 months. A suspension of NCE-G28 cells injected
s.c. as well as i.p. into nude mice produced solid tumors in all cases.
Histologically these tumors closely resembled the original tumor. The
original tumor, the nude mouse tumor, and NCE-G28 cells were immunochemically positive for glial fibrillar) acidic protein as well as for
neural plasma membrane antigen A2BS expression. Two cell strains,
9B2C and 9B2E, were obtained by cloning of the initial cultures and
another strain, NCE-G28T, was derived after explantation of a mouse
heterotransplant. The two subclones were negative for glial fibrillar)
acidic protein expression but stained for cell surface fibronectin. NCEG28T cells initially were positive for glial fibrillar) acidic protein but
lost this property within 8 months of cultivation. Karyotype analysis of
NCE-G28 and the three strains revealed hyperdiploidy and six structur
ally altered marker chromosomes five of which were shared by nearly all
cells. Receptors for epidermal growth factor were detected in all cell lines
with the highest levels (about 300,000 receptors/cell) in the parental cell
line. The epidermal growth factor receptors had an affinity of 2.5 n\i
(Aj) and by affinity cross-linking analysis a molecular weight of 170,000
was found. Initially, NCE-G28 cells responded to epidermal growth factor
as well as fibroblast growth factor with increased rates of proliferation,
while platelet derived growth factor had no effect. In higher passages the
growth factor sensitivity was reduced. Using antibodies directed against
synthetic protooncogene peptides the production of c-.v/.vimmunoreactive
material was detected. NCE-G28 cells produce an autocrine factor which
stimulated proliferation. This factor is present in conditioned medium
and is active on cultured meningiomas and other glioma cell lines. NCEG28 cells can be maintained in serum-free defined medium on plastic
coated with fibronectin or an extracellular matrix from bovine corneal
endothelial cells. The NCE-G28 cell line with its strains provide an in
vitro model system in which the complexity of gliosarcoma cell popula
tions and the interaction of the cloned cellular constituents can be studied.
INTRODUCTION
Gliosarcomas are rare intracranial lesions and constitute only
1-2% of the tumors of glial origin (1). According to the results
of pathological studies of single cases (2-4), small series (5-7),
and more comprehensive reviews ( 1, 8), the existence of glio
sarcoma as an entity has been acknowledged by its incorpora
tion into the WHO classification (9), conceding though that it
is a variant of glioblastoma containing malignant glial and
mesenchymal cellular components. Few gliosarcomas have been
studied in vitro although they provide an interesting model to
investigate the biology of presumptive glial precursor cells in a
sarcoma as well as the mechanisms which may lead to the
induction of sarcoma in a glial tumor (1,5, 7).
In this investigation we describe the initiation and character
ization of a cell line and the derivation of three strains derived
from a histologically confirmed gliosarcoma. This cell system
comprises the continuously cultured uncloned parental line
(NCE-G28), two subclones, and a strain derived from a tumor
produced by injection of parental cells into nude mice. Of
particular interest in this study are the biological effects of
growth factors, the staining patterns of the cell strains for
immunocytochemical markers, and the cytogenetic character
ization.
MATERIALS AND METHODS
Case History. A 67-year-old male patient was admitted to the De
partment of Neurosurgery of the University Hospital Eppendorf with
fatigue of 2 months' duration and a slowly progressive tetraparesis. On
computed tomography scans a large left parietal mass close to the
midline and surrounded by edema was seen which appeared to have no
relation to the falx. The patient underwent craniotomy for a suspected
glioma although tneningiosarcoma was considered as differential diag
nosis. During the operation a tumor was removed which macroscopically appeared to be a glioma but in frozen sections was indeed suspi
ciously reminiscent of meningiosarcoma. Definitive histology showed
that the tumor was a gliosarcoma. The patient recovered postoperatively
and was discharged 3 weeks after surgery. In comparison with the
postoperative control computed tomography scan a recurrence was
noted only 2 months later. Because the condition of the patient rapidly
deteriorated and the decision was made not to intervene again in
consideration of the histológica!diagnosis, the patient expired 4 months
postoperatively.
Histology. The tissues from the primary tumor as well as from the
tumors obtained in nude mice after injection of a suspension of IO6
dispersed cultured cells were fixed in 4% formalin and then embedded
in paraffin. Tissue slices were stained with hematoxylin-eosin, Gomori
silver stain, and also with the use of van Gieson and Giemsa techniques.
Additionally, immunostaining with antisera against GFAP3 and vimentin were performed.
Cell Culture. At surgery two pea-sized specimens which macroscopically were clearly identified as tumor tissue were placed into sterile
CMF. The material was freed from blood by rinsing with CMF, minced
with scissors, and transferred into a mixture of 0.03% collagenase,
0.05% Pronase, and 0.01% DNase (all from Sigma, St. Louis, MO) in
CMF. The suspension was incubated at 37°Cfor about 20 min with
occasional mixing. Thereafter, the cellular aggregates were dispersed
by trituration. The undispersed fragments were allowed to settle, and
the turbid supernatant was removed and centrifuged at 100 x g for 5
min. The supernatant containing the enzymes was aspirated, the cell
pellet dispersed in culture medium, and the suspension distributed into
three T25 tissue culture flasks (Falcon, Oxnard, CA). One of the culture
flasks had been covered previously with extracellular matrix produced
by bovine corneal endothelial cells (10). The resulting cell cultures were
maintained in stock culture medium consisting of DMEM with the
Received 5/21/87; revised 9/10/87; accepted 10/22/87.
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.
' Supported by the Deutsche Forschungsgemeinschaft, Bonn-Bad Godesberg,
by the Johannes Bauer Stiftung, Hamburg, and by the H. and 1 . Schilling Stiftung,
Essen, Federal Republic of Germany.
2 To whom requests for reprints should be addressed.
3The abbreviations used are: GFAP, glial fibrillary acidic protein; CMF,
calcium- and magnesium-free Hanks' balanced salt solution; DMEM, Dulbecco's
modified Eagle's medium; ECM, extracellular matrix; EGF, epidermal growth
factor; FGF, basic fibroblast growth factor; FN, fibronectin; PBS, phosphatebuffered saline; PDGF, platelet-derived growth factor; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; STV, phosphate-buffered saline/
trypsin/versene; TGF-a, transforming growth factor a.
731
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
R é^
"*fcps
: > S?,
. «'AI ' n'é^V'
^ ^».*
Fig. I. Tumor histology and morphology of cultured NCE-G28 cells. The cellular and fibrillary composition of the original tumor and of the heterotransplanted
tumor is depicted in H & E stain (A and C, respectively, x 400) and in Gomori silver stain (B and D, respectively, x 80). E-H, phase contrast photomicrographs of
the cell system established from the corresponding tumors (x 100). /. cells of the parental strain NCE-G28 at passage 8; /•',
culture of the cloned 9B2C strain
containing mostly bipolar cells with extended cellular processes; G, culture of 9B2E strain consisting of more voluminous, bipolar cells with shortened cellular
extensions; //. G28T culture derived from the heterotransplanted tumor.
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
v-. *
*»»-. Jfc^
>
.
v
•
m
Fig. 2. Phase contrast photomicrographs of NCE-G28 cells and immunofluorescence staining of the same cells with an antiserum against GFAP (A and B) G28T
cells also in phase contrast and immunofluorescence for GFAP (C and D). Clone 9B2C in phase contrast micrographs and stained with an antiserum against FN i/
and /•').
The same arrangement for clone 9B2E is shown in '/ and //
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION
OF A GLIOSARCOMA CELL LINE
clusters were grown to confluence. The culture dishes were then put on
ice, the medium removed, and the cell layer washed once with cold
binding buffer (0.002% CaCl2-0.002% MgSO4-0.05% bovine serum
albumin in PBS, pH 7.4). For competition experiments under equilib
rium binding conditions EGF was then added in the dose range of
0.075-200 nM. Thereafter, I25l-labeled EGF was added (250,000 cpm/
1Xultureoi/>OJosfc-med3525155J_I'y^.i
iimed
mlG
28
10
0
0.9
0.1
well) and the binding reaction allowed to proceed for 2 h. For saturation
analysis, first a maximally displacing concentration of unlabeled EGF
(200 nM) was added to those wells intended for estimation of nonspe
cific binding. '"I-labeled EGF (700 Ci/mmol) was then added in
0.8
0.2
0.6
0.4
0.4
0.6 ml
Fig. 3. Autostimulation of NCE-G28 cells by self-conditioned medium (c-med
G 28) added to quadruplicate cultures in the amounts indicated (HI). The total
incubation volume was made up to 1 ml with fresh serum-free medium (sfmed).
Cell-counts were performed at day 6 of the incubation. Cells only incubated only
in serum-free medium served as controls (•).Bars, SD.
addition of 10% fetal calf serum, 2 HIM sodium pyruvate, 1 HIM Lglutamine, 25 /ig gentamycin/ml, and 2.5 ¿tg
fungizone/ml. The cultures
were kept at 8% CO2 in humidified atmosphere at 37°C.The established
cell cultures were perpetuated on normal T75 tissue culture flasks
(Costar) and passaged with the help of 0.05% trypsin/0.04% EDTA in
PBS (STV). All media and supplements were obtained from Biochrom,
Berlin, Federal Republic of Germany. The clones designated 9B2C and
9B2E were generated from the parental NCE-G28 cells by limited
dilution in 96-well cluster dishes. Three subsequent cloning steps were
performed.
Growth Experiments. The cells were detached from stock cultures
with STV and seeded in 24-well tissue culture cluster dishes (Falcon)
at a density of 20,000 cells/well. After allowing the cells to attach,
stock culture medium was replaced after 24 h by defined serum-free
medium consisting of a 1:1 mixture of DMEM and Ham's F12 fortified
with the regular supplements mentioned above. Additionally 10 /ug/ml
insulin, 10 ^g/ml transferrin, 10 nM sodium selenite, and 1 mg/ml of
bovine serum albumin complexed with linoleic acid (1% w/w) were
added. The compounds were obtained from Sigma. Test substances or
previously conditioned media were then added every second day. The
experiments were terminated at the indicated time periods by detaching
the cells with STV and counting them in a Coulter Counter. Each
experimental condition was evaluated in quadruplicate. The growth
factors were added for the first time together with the change to serumfree conditions and every second day thereafter. EGF had been purified
according to the method of Cohen and Savage (12). FGF was purified
from bovine brain according to the method of Gospodarowicz et al.
(11). PDGF was obtained as a more than 30% pure preparation from
human blood platelets from J. Hoppe of the Gesellschaft fürBiolo
gische Forschung, Braunschweig, Federal Republic of Germany. The
preparation was maximally active at a concentration of 35 ng PDGF/
ml. The concentrations used in the experiments were found to be most
effective in dose-response experiments (not shown).
EGF Receptor Characterization. Cells seeded in 24-well tissue culture
decreasing concentrations to duplicate wells in the presence or absence
of cold EGF. After 2 h incubation on ice, the buffer was removed and
the cell layer was washed once more with cold buffer and then lysed in
0.5 M NaOH. The lysates were transferred into disposable test tubes,
combined with a wash of the respective wells, and then counted in a
gamma counter. EGF was iodinated by the chloramine-T method (13).
For the affinity cross-linking experiments, the binding was carried out
as described above except that 35-mm culture wells were used. At the
end of the incubation, 20 mM disuccinimidyl suberate (Pierce Chemi
cals, Rockford, IL) dissolved in dimethyl sulfoxide were added. After
15 min, the buffer was aspirated and the cell layer was washed once in
PBS and then lysed in 2% SDS-50 mM Tris-HCl, pH 6.8. Aliquots of
the lysate were counted, while other aliquots were prepared for SDSPAGE in an 8% gel according to standard protocols (14). A mixture of
proteins in the range of M, 14,000-200,000 was used as molecular
weight markers (Bio-Rad, Richmond, CA). The dried gels were then
autoradiographed using Cronex Lightning Plus (DuPont, Wilmington,
DE) intensifier screens.
Immunocytochemistry. Coverslides bearing the cultured cells were
removed from the bottom of 24-well cluster dishes and processed with
routine methods for staining with primary antibodies against FN,
GFAP (both polyclonal antisera raised in rabbits; Dakopats, Glostrup,
Denmark), galactocerebroside (monoclonal mouse hybridoma super
natant), and A2B5 (monoclonal mouse ascites; both kindly provided by
Dr. Mark Noble, Institute of Neurology, Queen Square, London,
United Kingdom). After incubation with fluorescein isothiocyanate or
rhodamine-coupled second antibodies (Dakopats) the stained cultures
were embedded in Cityfluor antifading reagent (The City University of
London) and evaluated with a Leitz fluorescence microscope.
Chromosome Analysis. Mitotic cells from subconfluent cultures in
160-enr tissue culture flasks (Nunc, Roskilde, Denmark) were detached
with STV, pelleted at 80 x g, and then resuspended in 0.6% KC1
solution. After swelling sufficiently, the cells were centrifuged and
subsequently fixed in acetic acid:methanol (3:1). The metaphases were
spread on microscope slides and air-dried. Staining of the chromosomes
was performed with 4% orcein or 0.4% Giemsa after treatment with
0.125% trypsin in 0.9% NaCl solution (15, 16).
Analysis of Protooncogene Products. One confluent T75 culture flask
was used for each strain. The cells were detached with EDTA, pelleted
at 1000 x g, and extracted with 2% SDS or 0.5% Triton X-100-0.5%
deoxycholate in PBS. The extracts were analyzed by SDS-PAGE in 8
Fig. 4. NCE-G28 cells under various cul
ture conditions. A, cells under regular culture
conditions in 10% fetal calf serum and after 6
(B) and 10 (C) days of serum-free culture. Cells
after 10 days of serum-free culture, detached
with STV and plated again under serum-free
conditions on culture dishes without coating
(D) and coated with FN (E) or bovine ECM
(F).
734
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION
NCE-G28p8
OF A GLIOSARCOMA CELL LINE
a narrow network of collagenous tissue which partially was
recognizable as belonging to the considerably proliferated lep
tomeninges. Immunohistochemically GFAP positivity was de
tected in a minority of tumor cells (not shown).
Most of the cells isolated from the original tumor readily
attached to ECM within 4 h. Approximately the same degree
of cell attachment to uncoated plastic dishes was reached on
the next day. The major cell types present initially were compact
and partially polygonal or more elongated, spindle-shaped (Fig.
IE). In addition, very few intermediate forms of extensively
branched, nondividing cells were observed in the primary cul
tures and disappeared within the first two passages of cultiva
tion (not shown). Thereafter, the NCE-G28 cultures presented
as an evenly mixed monolayer of the above-mentioned two
cellular components. Cloning of the cells eventually led to the
isolation of the clones 9B2C and 9B2E (Fig. l, Fand G) with
more homogeneous although different morphology.
Under optimal conditions NCE-G28 cells have a doubling
time of 26 h. When injected s.c. or i.p. into nude mice, tumors
were found in all animals within 3 months. The tumors removed
from the nude mice were indistinguishable from the original
tumor by conventional histology (Fig. I, A and C) or by special
techniques such as Gomori silver staining for fibrillary struc
tures (Fig. l, B and D) and immunohistochemistry for GFAP.
Serial transplantations of the tumor tissues from one animal to
another were possible and the tumors were maintained in
succession for 1.5 years. By explantation from the first nude
mouse tumor the NCE-G28T strain (Fig. \H) was established.
Immunostaining for GFAP revealed marked differences in
the expression patterns among NCE-G28, 9B2C, and 9B2E as
seen in the synopsis of the phase contrast micrographs and the
corresponding immunostainings. The cells of the parental cul
tures stained to about 30-50% with antiserum to GFAP (Fig.
2, A and B); the subclones 9B2C and 9B2E showed no staining
at all. Both clones expressed surface FN on all cells with varying
intensity (Fig. 2, E-H). FN staining in the NCE-G28 cultures
was positive in those cells not expressing GFAP (not shown).
In addition, about 2% of the cells of the parental line stained
positive for A2B5 (not shown) which is a marker associated
with glial progenitor cells. Initially the cells of the NCE-G28T
90
560-oU)Ã
40"ONUMBER
OFCD1£J'
20-
I
Fig. 5. Response of NCE-G28 cells to the addition of growth factors at passage
8 (p8) (left) and passage 90 (p90) (right). Cell growth is shown in the presence
of 10 nM EGF (D), 80 pg FGF/ml (UH)and 35 ng PDGF/ml (EB)in serum-free
cultures; •.control. Symbols, mean ±SE(ears) of quadruplicate cultures. The
cell counts were performed after 8 days of growth factor treatment.
and 12% gels. After electrophoresis the separated proteins were electrophoretically transferred to nitrocellulose sheets and then probed with
monoclonal antisera raised against synthetic peptides deduced from csis and \-erb B gene sequences. Visualization was achieved by incubation
with I25l-labeled Protein A and subsequent autoradiography. The meth
odology is described in detail elsewhere (17, 18).
Animal Experiments. NCE-G2S cells at passages 8, 15, and 87 of
cultivation were detached from a confluent 75-cm2 culture flask and
suspended in serum-free DMEM. About IO6cells were injected s.c. or
i.p. into each of four animals/group. Six-week-old female athymic
NMRI mice were used for the experiments.
RESULTS
The histological analysis of the original tumor (Fig. \A)
revealed a moderately to highly cellular tissue infiltrating brain
and leptomeninges. The intraparenchymatous cells had eosinophilic and granular cytoplasm. The highly pleomorphic and
mostly vesiculated nuclei showed extremes of chromatin den
sity. Several giant cells were scattered throughout the tissue.
Necroses, pseudopalisading, and moderate endothelial prolif
eration were found representing the hallmarks of glioblastoma.
Additionally, the tumor had signs of astrocytic differentiation
such as perivascular pseudorosettes, a background fibrillarity,
and scarce microcystic degeneration. In part, the tumor cells
were elongated and arranged in long parallel interwoven bun
dles reminiscent of fibrosarcoma. The tumor was intersected by
cpm
I
ri*
- 200 k
1
o
10000
G
UJ
Fig. 6. Characterization of EGF receptors
on NCEG28 cells. A. autoradiograph of an
SDS-PAGE gel after affinity cross-linking of
12!I-labeled EGF to cultured cells with disuccinimidyl suberate. lì.saturation analysis of
EGF binding to NCE-G28 cells. Increasing
amounts of I2!l-labeled EGF were added to
duplicate cultures in the presence or absence
of an excess amount of unlabeled EGF. Points,
specific binding per confluent culture; inset,
Scale-hard analysis. C, displacement curve
from an equilibrium binding assay in which
increasing concentrations of unlabeled EGF
competed with a constant amount of I2'l-labeled EGF. Points, total binding for the given
concentration which was assayed in quadrupli
cate cultures. A, thousands.
-116 k
Si 5000
-92k
jjj
-66k
5
10
15
[125-I-E6Fl,nM
C. cpmMfr)12
-45k
EXCESS COLD EGF
(200nM)
o
UJ
-3
-2-10
log doMlnM)
1
735
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION OF A GL1OSARCOMA CELL LINE
The biological response to optimal concentrations of growth
factors was evaluated in cell cultures which were seeded at low
density and were maintained in serum-free defined medium
(Fig. 5). The addition of 80 pg FGF/ml or 10 nM EGF every
second day resulted in increased proliferation rates in passage
eight of cultivation (Fig. 5, left). In this experiment, FGF proved
to be the most effective mitogen. PDGF at a concentration of
35 ng/ml was not effective. The same experiment when repeated
in higher passages showed a reduced responsiveness to the
added growth factors (Fig. 5, right).
EGF receptors on NCE-G28 cells were studied using 125I-
v
Fig. 7. Autoradiographs of SDS-PAGE/Western blots on which NCE-G28
proteins reacting with antisera against \-erb B and c-s/s were visualized with I!!Ilabeled Protein A. Left, anti-v-erA B (23-39); right, anti-sis (67-84). kD, kilodaltons (expressed as M, within the paper).
strain were largely positive for GFAP but after extended culti
vation became mostly negative (Fig. 2, C and D).
For evaluating the possibility of autocrine growth factor
production by NCE-G28 cell cultures, conditioned medium
from cultures maintained for at least 10 days either in medium
supplemented with 1% fetal calf serum or in serum-free condi
tions, was added back to cells at low density (5000 cells/culture)
in serum-free culture conditions. The conditioned medium had
a profound stimulatory effect. The effect was demonstrable
even when only 10% of the fresh medium was replaced (Fig. 3).
Cell death and lysis occurred when the cultures were maintained
in medium which contained 80% or more of the conditioned
medium. In separate experiments conditioned medium was
collected, concentrated by lyophylization, and passed through
a Sephadex G-75 gel column. The fractions were lyophylized,
reconstituted, and tested on cultures of NCE-G28 cells as well
as on primary cultures of meningioma or other glioma cells
from our laboratory. It was consistently shown that the biolog
ical activity eluted in a position corresponding to a rough
molecular weight estimate of 10,000-15,000 if ideal Chromat
ographie behavior is assumed.
In order to collect more of the conditioned medium from
cells grown continuously in serum-free culture, the culture
system was optimized by using ECM. This procedure became
necessary because it was observed that NCE-G28 cells main
tained in serum-free culture for periods longer than 6 days tend
to aggregate and will eventually detach in clumps (Fig. 4, AC). To evaluate the hypothesis that this behavior may be due
to the fact that NCE-G28 cells only insufficiently produce FN
and therefore require serum for attachment, the aggregated
cells were fully detached with STY and plated in culture flasks
which were untreated or coated either with FN (one rinse with
1 ml of a solution of 1 mg FN/ml) or with ECM from bovine
corneal endothelia (Fig. 4, D-F). It was found that the NCEG28 cell aggregates only loosely attached to untreated plastic,
whereas on FN there was active cell migration and monolayer
formation which was even more pronounced on ECM.
labeled EGF. The K¿of the binding reaction of EGF to its
receptors was estimated to be about 2.5 nM in an equilibrium
binding assay with whole cells attached to culture dishes (Fig.
6C). Saturation analysis showed the binding sites to be saturable
at an EGF concentration of 10 nM with a Ka of 2.2 HM (Fig.
6fi). From a Scatchard analysis the receptor density was esti
mated to be 300,000 receptors/cell. This number is in agree
ment with another estimate obtained by direct radioimmunoassay for the solubili/ed EGF receptor as determined at 0.11
Mgof EGF receptor protein/IO6 cells. Further molecular char
acterization of EGF receptors was achieved by affinity crosslinking of I25l-labeled EGF to whole cells using the heterobifunctional agent disuccinimidyl suberate. Subsequent lysis of
the cells and SDS-PAGE followed by autoradiography of the
gels revealed a major band with the molecular weight of 170,000
which was completely displaced with an excess of unlabeled
EGF (Fig. 6A). A semiquantitative comparison of maximal
binding to confluent cultures of NCE-G28, G28T, 9B2C, and
9B2E cells showed that the I25l-labeled EGF binding in the
parental NCE-G28 strain was at least twice as high as in the
other strains derived thereof with unchanged affinity.
Analysis of cell extracts by SDS-PAGE, Western blot, and
immunostaining for c-erb B and c-sis expression using antisera
against synthetic sequences of the respective protooncogene
products (Fig. 7) revealed a whole set of s/s-related immunoreactive bands between M, 35,000 and 86,000. In contrast,
there was no detection of c-erb B product but only crossreactivity with the native EGF receptor at M, 170,000 and a MT
42,000 proteolytic subunit of the EGF receptor.
For further characterization, karyotype analyses were per
formed with parental NCE-G28 cells, with the subclones 9B2C
and 9B2E and with G28T cells derived from the nude mouse
tumor demonstrating aneuploidy in nearly all metaphases ex
amined (Fig. SA). An exception were NCE-G28 cells in passage
four of the cultivation which contained 4% metaphases with 46
chromosomes, possibly due to contamination of the culture
with cells from normal tissue, which apparently were eliminated
during further in vitro cultivation. Parental NCE-G28 cells
largely maintained the distribution pattern of the chromosome
numbers per cell whereas a considerable shift to lower chro
mosome numbers was found in cells of the sublcones; this is
also reflected in the altered morphological properties of the
subclones. There were no G28T cells with less than 46 chro
mosomes indicating the absence of contamination with mouse
cells. As additionally revealed by G banding, analysis of all
chromosomes of NCE-G28T cells showed banding patterns
characteristic of human origin. Although the histograms
showed that most cells of the clonal strains 9B2C and 9B2E
contained fewer chromosomes than the parental cell line, the
distribution profile of the structurally unaltered chromosomes
was very similar in all strains examined, demonstrating a close
cytogenetic relationship (Fig. SB). Exceptions were chromo
somes 11, 12, and 15 which occurred more than disomic in
736
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION
OF A GLIOSARCOMA CELL LINE
number o' metophoses
i4
" o:
ï3
Fig. 8. A, numerical aberrations and distri
bution of chromosomes in NCE-G28 cell
strains; histograms of the number of chromo
somes per metaphase. Parental cells in passage
4 (a) and 28 (*) of the cultivation, cells of the
subclones 9B2E in passage 2 (c) and 9B2C in
passage 8 (d) after subcultivation, and cells of
the strain NCE-G28T in passage 8 (e) after
explantation from the nude mouse tumor. //,
distribution of the structurally unaltered chro
mosomes; /, profiles of parental NCE-G28
cells and G28T cells derived from the heterotransplanted tumor. II, comparison between
9B2C and 9B2E cells.
NCE-G28 CLONES
9B2E
.II uni
16
W
AVERAGE
M
TV
'O
io
90
. «o
NUMBER Of CHROMOSOMESPEB METAPHASE
1 2
34
5 6 7 a 9 10 11 12 13 14 IS 16 17 18 19 20 21 22 X Y
a
»1l
Uff tfÃ-r\\l
6
13
9
8
14
9B2C
U
4
if
10
11
-*f ni
titf
15
16
17
U
19
hM1
20
M2
21
M4
12
22
18
n.
XY
99
M
M5
M6
U
Fig. 9. Karyotype (72,XY) representative of NCE-G28 cells in passage 28 of the cultivation. Four identified marker chromosomes are shown; one marker (U) of
unidentified origin remained.
737
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION
"Hi
OF A GLIOSARCOMA CELL LINE
1
del(l)(pl2)
del(3)(qllq21)
del(6)(q21)
t(l;2)(p33;pll,2)
del(7)(q22)
Fig. 10. Derivation of the marker chromosomes occurring in NCE-G28 cells. The origin of the chromosomal fragments is shown in comparison with the
idiogrammatic representations of the corresponding normal chromosomes.
most NCE-G28 and G28T cells but were not increased in cells
of the cloned strains. Most prominently, chromosome 7 was
overrepresented, in addition to chromosomes 2, 5, 8, 20, and
21. In the position of chromosomes 1, 3, and 6 of the distribu
tion profiles distinct nadirs were observed which found their
equivalents in the rearranged fragments of the marker chro
mosomes Ml, M2, M4, and MS detected in nearly all cells
analyzed.
A karyogram representing NCE-G28 chromosomes in pas
sage 28 of the cultivation (Fig. 9) demonstrates that nearly all
of the chromosomes were identifiable. Six marker chromo
somes occurred in the metaphases analyzed in various passages
of the cultivation. In passage 28 the parental cells contained all
markers but M3; this marker consisting of a translocation
between chromosomes 1 and 2 (Fig. 10) was found exclusively
in cells of the clone 9B2C. Accordingly the occurrence of
structurally unaltered chromosome 1 was reduced in 9B2C cells
(Fig. 85). Marker M6 representing a terminally deleted chro
mosome 7 appeared infrequently although it appeared in more
than 50% of the cells of the parental line and the subline 9B2C.
M6 was completely eliminated during passage of NCE-G28
cells through the nude mouse and subsequent in vitro cultiva
tion.
DISCUSSION
Gliosarcomas are defined as primary malignant brain tumors
(9) and are composed of two dissimilar malignant tissues (2).
It is assumed that in most cases the sarcoma part develops
secondarily from hyperplasia of endothelial (1,7, 19) or nonendothelial (3,20) constituents of the vessel wall within anaplastic
astrocytomas. The reverse situation, that the glioma part is
induced by a sarcoma (8) is less frequent. It is conceivable that
the induction of the sarcoma or glioma part of gliosarcoma
may be due to factors produced by one of the components. The
production of TGF-a as well as PGDF by human sarcomas and
brain gliomas was shown (21, 22). In addition, a tumor angiogenesis factor was isolated from human tumors (23). The fact
that sarcomas may arise even from brain métastasesof primary
carcinomas of other organs (24, 25) may indicate a particular
susceptibility of the brain to develop secondary sarcoma.
In this study we present a cell line derived from a tumor with
738
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION
OF A GLIOSARCOMA CELL LINE
mixed glial and mesenchymal components. The parental cell is important in context with the observation that cultured
line NCE-G28 gave rise to three further strains with different
glioma cells very often express exclusively FN and lose their
biological properties but close karyotypic resemblance to the ability to stain for GFAP (32-35). At present we have not been
able to establish subclones with 100% GFAP-positive cells. It
parental line. It is one of the few cell lines established from
may be speculated that the GFAP-positive NCE-G28 cells,
gliosarcoma and caution has to be exerted to extrapolate con
despite their neoplastic nature as proven by the formation of
clusions from this case to all other tumors of this type. NCEG28
GFAP-positive tumors in the nude mouse and the presence of
cells are still strongly positive for GFAP at passage 90, thereby
leaving no doubt about the glial origin. The fact that the chromosomal aberrations in all metaphases, depend on factors
parental NCE-G28 cell strain contains 2% cells which stain
secreted by cells such as 9B2C or 9B2E which by themselves
are GFAP negative. Our findings are of interest in comparison
positive with the A2B5 antibody (26) even raises the possibility
with two other reports on cell cultures derived from gliosarcoma
that glial progenitor cells (27) are present.
The malignant nature of the glial component of the NCE(36, 37). The cell line SF-539BT (36) is exclusively positive for
FN and it was speculated that only the sarcomatous component
G28 cell line is supported by the fact that repeatedly upon
injection of cell supensions at passages 8, 17, and 87 GFAPsurvived the initial subculturing procedures. The other alter
positive tumors were generated in athymic mice. The two native as suggested by our results is that the switch from GFAP
subclones which were isolated in early passages of cultivation
to FN expression in different cell lineages is an event occurring
subsequent to transformation, thus allowing for differences in
presumably are tumor cell strains because they show unlimited
phenotype coexisting with close similarities of karyotype. In
in vitro growth, are unresponsive to the addition of external
growth factors (results not shown), and their karyotypes are terestingly none of the five marker chromosomes consistently
found in that cell line is related to the markers found in NCEclosely related to the parental cells. Neither of them, however,
G28 cells or the derived strains. In the other case (37), there
produced tumors in nude mice.
It is speculated that induction of one of the two tissue
was also no demonstration of GFAP in cell cultures from a
components is a mechanism underlying the development of GFAP-positive tumor, and the possibility of sarcomatous over
gliosarcoma. This process may be mediated by the secretion of growth was considered.
Karyotype analyses of human neoplasias have provided so
soluble factors which induce or sustain growth of different
cellular components within one tumor. Autocrine secretion of far only limited clues to distinct alterations of chromosomal
growth factors by NCE-G28 cells can be assumed from the structure which may be related to specific diseases (38-43).
demonstration of c-.v/.vimmunoreactivity and the autostimulaEspecially in the chromosomal analysis of gliomas only very
tion experiments. The growth-stimulating activity produced by general indications have been obtained (44-46). Our study
NCE-G28 cells acts on a broad spectrum of brain tumor cells confirms that chromosome 7 bearing the gene for the EGF
and therefore is either a complex mixture of different factors
receptor (47) is overrepresented in accordance with the dem
or a nonspecific factor such as TGF-o. Characterization of the onstration of EGF receptors on NCE-G28 cells despite reduced
autocrine activity and its regulation in the NCE-G28 cell system
cellular responsiveness to EGF. This demonstrates that overrepresentation of chromosome 7 is not necessarily correlated
described will be the subject of an independent study.
The number of EGF receptors on NCE-G28 cells is less than
with increased cellular responsiveness to EGF. Polysomy 7 has
10% of that found on cells of the epidermoid-carcinoma line also been found by others (48, 49) and was reported to be
A431. Addition of EGF to NCE-G28 cells resulted in a prolifcombined with overexpression of the erb B oncogene in six
erative response in early passages of cell cultivation but later cases (48). Overexpression of c-erb B was not detectable in
the cellular responsiveness disappeared. The EGF receptors on NCE-G28 cell and therefore is not always correlated with the
NCE-G28 cells appear to be of regular molecular characteris
overrepresentation of chromosome 7. The location of the gene
tics. They were not altered as it was demonstrated in the case for TGF-a is on the short arm of chromosome 2 (50). This
of a rapidly growing glioblastoma cell line which had a high chromosomal region is involved in the translocation identified
number of receptors per cell with high affinity but altered
in the marker M3 of this study. However, M3 was exclusively
electrophoretic behavior (28). NCE-G28 cells possess roughly
found in cells of the subclone 9B2C which was unresponsive to
three times the amount of EGF receptors in comparison with any of the added growth factors. The results from the chromo
the EGF-unresponsive strains and during long-term cultivation
somal analysis of our study and the comparison with findings
became less responsive to the addition of EGF. This leads to in other gliosarcomas (36, 44) revealed that by the analysis of
G-banding data, at present no consistent aberrations can be
the speculation that the EGF receptors may serve the purpose
to mediate autocrine action of TGF-a which was shown to act implied in gliosarcoma.
The here-described NCE-G28 cell system will prove to be a
via the EGF receptor (29), and which is produced by a broad
useful
model for in vitro-in vivo studies because the tumors
spectrum of human neoplastic cells (22). By analogy it cannot
be excluded that sustained in vitro growth of the cell lines is derived from NCE-G28 cells injected into nude mice resemble
due to maximal self-stimulation by other factors such as PDGF,
the original tumor exactly and are serially transplantable. The
brain-derived growth factor (30), or a growth-promoting factor
autocrine production of growth factor(s) and the presence of
functionally intact EGF and FGF receptors will make this
recently isolated from astrocytoma cells (31).
Referring back to the possible mechanisms causing gliosar
system a suitable candidate for investigating protooncogene
coma a particularly interesting aspect is provided by the result
amplifications and expression in human brain tumor cells.
that none of the subclones of the NCE-G28 cells that stained
Considering that the individual components may depend on
for GFAP but were positive for cell surface FN. The occurrence
each other for differentiation and proliferation, experiments in
which different cell strains derived from one tumor are recomof a number of marker chromosomes shared by the parental
cells and the subclones demonstrates for the first time that
bined will be of interest.
GFAP- and FN-positive cells from one central nervous systemThe present study indicates that gliosarcomas can be culti
vated and maintained in their cellular complexity. Cloning and
derived tumor are karyotypically related and consequently
animal passage allow for characterization of the cellular coinmight be actually derived from one common ancestral cell. This
739
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
CHARACTERIZATION OF A GL1OSARCOMA CELL LINE
ponents and the complexity of their interactions. Our results
suggest that there might be a common ancestral cell from which
cells positive for either GFAP or FN are derived. This finding
would be in contradiction to the assumption that two dissimilar
tissues simultaneously are transformed resulting in gliosar
coma. Present knowledge of these tumors necessitates the use
of more subtle methods of gene analysis which go beyond
conventional G banding of chromosomes in which only gross
genetic rearrangements are detectable. Eventually one might
expect that results from in vitro studies and correlations with
cellular differentiation parameters, growth factor, and receptor
production in combination with expression patterns of protooncogenes might enable us to understand the regulation of
composite tumors such as the gliosarcoma.
22.
23.
24.
25.
26.
27.
28.
ACKNOWLEDGMENTS
29.
We thank H. Nausch and C. Müllerfor excellent technical assistence.
We appreciate the cooperation of the departmental operating room
staff. It is a pleasure to acknowledge the photographic work of S. Freist.
30.
31.
REFERENCES
1. Morantz, R. A., Feigin, I., and Ransohoff, J. Clinical and pathological study
of 24 cases of gliosarcoma. J. Neurosurg., 45:398-408, 1976.
2. Feigin, L, and Gross, S. W. Sarcoma arising in glioblastoma of the brain.
Am. J. Pathol., 31:633-653, 1955.
3. Kishikawa, M., Tsuda, N., Fujii, H., Yokoyama, H., and Kihara, M. Glio
blastoma with sarcomatous component associated with myxoid change. Acta
Neuropathol., 70:44-52, 1986.
4. Pena, C. E., and l-eller. R. Ultrastructure of a composite gliomasarcoma of
the brain. Acta Neuropathol., 23: 90-94, 1973.
5. Lalitha, V. S., and Rubinstein, L. J. Reactive glioma in intracranial sarcoma:
a form of mixed sarcoma and glioma "sarcoglioma." Report of eight cases.
Cancer (Phila.), 43:246-257, 1979.
6. Kepes, J. J., Fulling, K. H., and Garcia, J. H. The clinical significance of
adenoid formations of neoplastic astrocytes, imitating metastatic carcinoma,
in gliosarcomas: a review of five cases. Clin. Neuropathol., 1: \ 39-150,1982.
7. Slowik, F., Jellinger, K., Gaszo, L., and Fischer, J. Gliosarcomas: hislological, immunological, ultrastructural and tissue culture studies. Acta Neuro
pathol., 67: 201-210, 1985.
8. Rubinstein, L. J. The development of continuous sarcomatous and gliomatous tissue in intracranial tumors. J. Pathol. Bacteriol., 71:441-459, 1956.
9. Zülch,K. J. Histological typing of tumors of the central nervous system.
International Histological Classification of Tumours, No. 21. Geneva: WHO,
1979.
10. Weiner, R. I., Bethea, C. L., Jaquel, P., Ramsdell, J. S., and Gospodarowicz,
D. J. Culture of dispersed anterior pituitary cells on extracellular matrix.
Methods Enzymol., 103: 287-294, 1983.
11. Gospodarowicz, D., Cheng, J., Lui, G. M., Baird, A., and Bohlen, P. Isolation
of brain fibroblast growth factor by heparin sepharose affinity chromatography: identity with pituitary fibroblast growth factor. Proc. Nati. Acad. Sci.,
81:6963-6967, 1984.
12. Cohen, S., and Savage, C. R. Preparai ion of epidermal growlh faclor.
Melhods Enzymol., 37:170-176, 1975.
13. Hunier, W. M., and Greenwood, P. C. Preparalion of iodine-131 labelled
human growth hormone of high specific activity. Nature (Lond.), 194: 195196, 1962.
14. Laemmli, U. K. Cleavage of structural proteins during the assembly of the
head of bacteriophage T4. Nature (Lond.), 227: 680-685, 1970.
15. Seabright, M. A rapid banding technique for human chromosomes. Lancet,
2:971-972, 1971.
16. Kunzmann, R., and Holzel, F. Karyotypic alterations in human ovarian
carcinoma cells during long-term cullivation and nude mouse passage. Cancer
Genet. Cytogenet., 2S.-201-212 1987.
17. Niman, H. L. Antisera to a synthetic peptide of the .vivviral oncogene product
recognize human platelet-derived growth factor. Nature (Lond.), .107: 180183, 1984.
18. Niman, H. L., Thompson, A. M. H., Yu, A., Markman, M., Willems, J. J.,
Herwig, K. R., Habib, N. A., Wood, C. B., Houghten, R. A., and Lerner, R.
A. Anti-peptide antibodies detect oncogene-related proteins in urine. Proc.
Nati. Acad. Sci. USA, Õ2:7924-7928, 1985.
19. Greene, H. S. N., and Harvey, E. K. The development of sarcomas from
transplants of the hyperplastic stromal endolhelium of glioblasloma multiforme. Am. J. Palhol., 53:483-499, 1968.
20. McComb, R. D., Jones, T. R., Pizzo, S. V., and Bigner, D. D. Inuminoli!*tochemical detection of Factor VHI/von Willebrand factor in hyperplastic
endothelial cells in glioblasloma multiforme and mixed glioma-sarcoma. J.
Neuropathol. Exp. Neurol., 41:479-489, 1982.
21. Betshollz, C., Heldin, C-H., Nister, M., Ek, B., Wasteson, A., and Wester-
33.
34.
35.
36.
37.
38.
39.
40.
41.
43.
45.
46
47
49,
50.
mark, B. Synthesis of a POGF-like growth factor in human glioma and
sarcoma cells suggests the expression of the cellular homologue to the
transforming protein of simian sarcoma virus. Biochem. Biophys. Res. Com
mun., 117:176-182, 1983.
Nickell, K. A., Halper, J., and Moses, H. L. Transforming growth factors in
solid human malignant neoplasms. Cancer Res., 43:1966-1971, 1983.
Klagsbrun, M., Knighton, D., and Folkman J. Tumor angiogenesis activity
in cells grown in tissue culture. Cancer Res., 36:110-114, 1976.
Schwartz, I., and Feigin, I. H. Sarcoma arising in metastatic carcinoma of
the brain. Acta Neuropathol., 3:74-78, 1963.
Feigin, I., and Budzilovich, G. N. Sarcoma arising in metastatic carcinoma
in the brain. Cancer (Phila.), 54: 2047-2050, 1984.
Eisenbarth, G. S., Walsh, F. S., and Nirenberg, M. Monoclonal antibody to
a plasma membrane antigen of neurons. Proc. Nati. Acad. Sci. USA, 76:
4913-4917, 1979.
Raff, M. C., Miller, R. H., and Noble, M. A glial progenitor cell that develops
in vitro into an astrocyte or an oligodendrocyte depending on culture medium.
Nature (Lond.), 303:390-396, 1983.
Westphal, M., Harsh, G. R., Rosenblum, M. L., and Hammonds, R. G.
Epidermal growth factor receptors in the human glioblastoma cell line SF
268 differ from those in epidermoid carcinoma cell line A-431. Biochem.
Biophys. Res. Commun., 132: 284-289, 1985.
Massague, J. Epidermal growth factor-like transforming growth factor. II.
Interaction with epidermal growth factor receptors in human placenta mem
branes and A431 cells. J. Biol. Chem., 25«:13614-13620, 1983.
Huang, S. H., Kim. MD..
and Huang, J.S. Transforming growth
factor activity of bovine brain-derived growth factor. Biochem. Biophys. Res.
Commun., 139: 619-625, 1986.
Ho, J.-I., Kato, T., Wakabayashi, S., Hara, F., Tanaka, R., and Kalo, K.
Autocrine regulation of glial proliferation and differentiation: the induction
of cytodifferentiation of postmitotic normal glioblasts by growth-promoting
factor from astrocytoma cell. Brain Res., 374: 335-341, 1986.
Baldwin, J. R., McKeever, P. E., and Booker, T. R. Products of cultured
neuroglial cells: II. The production of fibronectin by C6 glioma cells. Neurochem. Res., 10:601-610, 1985.
Chronwall, B. M., McKeever, P. E., and Kornblith, P. L. Glial and nonglial
neoplasms evaluated on frozen section by double immunoflourescence for
fibronectin and glial fibrillary acidic protein. Acta Neuropathol. 59: 283287, 1983.
Jones, T. R., Ruoshlati, E., Schuld, S. C., and Bigner, D. D. Fibronectin and
glial fibrillary acidic protein expression in normal human brain and anaplastic
human gliomas. Cancer Res., 42: 168-177, 1982.
McKeever, P. E., Thomas, C., Laverson, S., and Chronwall, B. M. Proteins
of human gliomas change from in situ to in vitro. J. Neuropathol. Exp.
Neurol., 42: 307, 1983.
Rutka, J. T., Giblin, J. R., Hoifodt, H. K., Dougherty, D. V., Bell, C. W.,
McCulloch, J. R., Davis, R. L., Wilson, C. B., and Rosenblum, M. L.
Establishment and characterization of a cell line from a human gliosarcoma.
Cancer Res., 46: 5893-5902, 1986.
McKeever, P. E., Wichmann, A., Chronwall, B., Thomas, C., and Howard,
R. Sarcoma arising from a gliosarcoma. Southern Med. J., 77: 1027-1032,
1984.
Yunis, J. J. The chromosomal basis of human neoplasia. Science (Wash.),
227: 227-236, 1983.
Atkin, N. B. Chromosomal aberrations in cancer. Cancer Genet. Cytogenet.,
21: 279-285, 1986.
Croce, C. M. Chromosome translocations and human cancer. Cancer Res.,
«.•6019-6023,1986.
Gilbert, F., Feder, M., and Balaban, G. Human neuroblastomas and abnor
malities of chromosomes 1 and 17. Cancer Res., 44: 5444-5449, 1984.
Mitelman, F. Restricted number of chromosomal regions implicated in
aetiology of human cancer and leukemia. Nature (Lond.), 310: 325-327,
1984.
Solieres, J. M. J. C., Hustinx, T. W. J., and Trent, J. M. Possible involvement
of unstable sites on chromosomes 7 and 14 in human cancer. Cancer Genet.
Cytogenet., 19: 151-158, 1986.
Bigner, S. H., Mark, J., Bullard, D. E., Mahaley, M. S., and Bigner, D. D.
Chromosomal evolution in malignant human gliomas starts with specific and
usually numerical deviations. Cancer Genet. Cytogenet., 22: 121-135, 1986.
Bigner, S. H., Mark, J., Burger, P. C., Bullard, D. E., Schuld, S. C., Mahaley,
M. S., and Bigner, D. D. Chromosomes and chromosomal progression of
malignant human gliomas: theoretical and practical implications. In: M. D.
Walker and D. G. T. Thomas (eds.), Biology of Brain Tumors, pp. 43-49.
Boston: Martinus Nijhoff Publishers, 1986.
Shapiro, J. R. Biology of gliomas: chromosomes, growth factors and oncogenes. Semin. Oncol., /J.-4-15, 1986.
Robson, E. B., and Meera Khan, P. Report of the committee on the genetic
constitution of chromosomes 7, 8 and 9. Cytogenet. Cell Genet., 32: 144152, 1983.
Henn, W., Blin, N., and Zang, K. D. Polysomy of chromsome 7 is correlated
with overexpression of the erb B oncogene in human glioblastoma cell lines.
Hum. Genet., 74: 104-106, 1986.
Libermann, T. A., Nusbaum, H. R., Razón,N., Kris, R., Lax, I., Soreq, H.,
Whittle, N., Waterfield, M. D., Ullrich, A., and Schlessinger, J. Amplifcation,
enhanced expression and possible rearrangement of EGF receptor gene in
primary human brain tumors of glial origin. Nature (Lond.), 313: 144-147,
1985.
Tricoli, J. V., Nakai, H., Byers, M. G., Rail, L. B., Bell, G. I., and Shows, T.
B. The gene for human transforming growth factor a is on the short arm of
chromosome 2. Cytogenet. Cell Genet., 42:94-98, 1986.
740
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.
Biological and Karyotypic Characterization of a New Cell Line
Derived from Human Gliosarcoma
Manfred Westphal, Marianne Haensel, Dieter Mueller, et al.
Cancer Res 1988;48:731-740.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/48/3/731
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1988 American Association for Cancer Research.