Download Isolation and Characterization of Human

(CANCER RESEARCH 51. 6660-6667. December 15. I99I|
Isolation and Characterization of Human Melanoma Cell Variants Expressing High
and Low Levels of CD44
Mary Birch, Stephen Mitchell, and Ian R. Hart1
Imperial Cancer Research Fund Laboratories, Lincoln 's Inn Fields, London H C2A 3PX, United Kingdom
ABSTRACT
Variants of the human melanoma cell line LT5.1 were selected for
high and low expression of the \1, 90,000 CD44 glycoprotein by using
the Hermes-3 monoclonal antibody combined with fluorescence-activated
cell sorting. Cells were single cell cloned and clones of CD44 highexpressing and CD44 low-expressing phenotype were isolated. The var
iants, which exhibited up to a 7-fold difference between high and low
expression, have maintained a stable phenotype over a period of 3 months
in tissue culture. Northern blot analysis of mRNA from the different
clones showed correlation of levels of transcripts with fluorescenceactivated cell sorting analysis data. Wound migration assays, utilizing
the different clones, showed that the low-expressing clones manifested
less motility than did cells showing high levels of CD44. Homotypic
aggregation of cells was increased in those cells expressing high levels
of CD44, and these variants were also better able to adhere to hyaluronate
substrates. All of these activities were inhibited by the presence of antiCD44 antibody. When injected i.v. into nu/nu BALB/c mice, the lowexpressing clones gave significantly fewer lung nodules than the highexpressing clones, although the two variant types did not differ in their
capacity to form s.c. tumors in similar mice. These results suggest that
the CD44 molecule, possibly as a function of its activities as a hyaluronate
receptor, may play a vital role in determining the fate of hematogenously
disseminating melanoma cells.
INTRODUCTION
CD44 is a widely distributed integral cell membrane glyco
protein that exists in a variety of forms, the most prevalent
being M, 85,000-95,000 in size, arising from posttranslational
modifications of a M, 37,000 core protein (1, 2). Monoclonal
antibodies raised against the Hermes antigen (3), polymorphic
glycoprotein Pgp-1 antigen (4), extracellular matrix receptor
ECM-III (5), and the human CD44 antigen all recognize the
same molecule (6), and suggest that this glycoprotein has a
pleiotropic role in various cell processes. One of the major
functions of CD44 appears to be in regulating lymphocyte
adhesion to the cells of high endothelial venules during lym
phocyte migration (7, 8), a process which has many similarities
to the metastatic dissemination of solid malignancies (9).
Cloning of CD44 cDNA revealed significant homology in the
NH2-terminal domain of the molecule with the tandem repeat
domains of the cartilage link and proteoglycan core proteins
(10, 11), suggesting that the molecule might have a more
ubiquitous role in cell adhesion than simply determining lym
phocyte-high endothelial venule interactions. This possibility
was underscored by the recent demonstration that CD44 is the
principal cell surface receptor for hyaluronate (12, 13). It is
known that this glycosaminoglycan plays a major role in diverse
cellular processes including cell-cell adhesion, embryonic de
velopment, and cell migration and angiogenesis (14, 15).
Many tumors are characterized by the production and accu
mulation of high levels of hyaluronate and neoplastic cells often
Received 4/11/91; accepted 10/4/91.
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.
' To whom requests for reprints should be addressed.
exhibit substantial capacity to bind to this glycosaminoglycan
(16, 17). Accordingly, it was of interest to note that elevated
levels of CD44 were detected in carcinomas relative to normal
epithelium, a finding which was thought to be consistent with
the possibility that the molecule could play a part in regulating
the invasive and metastatic process (11).
In the present study we have sought to investigate the putative
involvement of CD44 in the hematogenous spread of melanoma
cells by isolating a series of stable CD44-deficient and CD44sufficient variants from a single human melanoma cell line. The
technique used was that of iterative fluorescence-activated cell
sorting by using monoclonal anti-CD44 antibodies, followed by
single cell cloning, as described by Schreiner et al. (18) for the
isolation of fibronectin receptor-deficient variants. Isolated
clones have then been assessed for tumorigenic potential, ca
pacity to colonize the lungs of athymic mice after i.v. injection,
growth characteristics, cell adhesion and migration abilities,
and mRNA levels for the CD44 molecule.
MATERIALS
AND METHODS
Cell Culture. The DX3.LT5.1 human melanoma line (19) was grown
on plastic in Dulbecco's modification of Eagle's minimal essential
medium supplemented with L-glutamine and 10% heat-inactivated fetal
calf serum. Cultures were maintained at 37°Cin a humidified atmos
phere of 95% air, 5% CO2.
Variant Selection. DX3.LT5.1 cells were harvested following a brief
incubation with 0.25% trypsin, washed once in DEM,2 followed by two
washes in ice-cold PBS containing 1% BSA. Cells were stained by direct
fluorescence with the use of anti-CD44 antibody (Hermes-3, 10 Mg/ml;
generously provided by Dr. E. Butcher, Stanford University, Stanford,
CA) and fluorescein isothiocyanate-conjugated rabbit anti-mouse IgG
(1:40 dilution; Dako Immunoglobulins F232). Background controls
were established by omitting the primary antibody. Cells were separated
by fluorescence-activated cell sorting on 2 Facstar Plus (Becton Dick
inson) flow cytometer equipped with an argon ion 488 nm laser. Cells
were run through three selection cycles, where the brightest and dim
mest 5% of the population were selected and amplified between each
cycle, before the respective bright and dim populations were plated
directly to 96-well plates at a concentration of 0.5 cells/well and grown
to confluence. The individual clones were then harvested with trypsin
and replicate dishes were established which were then screened by using
the Hermes-3 antibody (10 ^g/ml) in the enzyme-linked immunosorbent assay technique as detailed by Durbin and Bodmer (20). The four
clones expressing the highest level of CD44 expression (high clones 1,
2, 6, and 7) and the four clones expressing the lowest level of CD44
expression (low clones 3, 5, 6, and 7) were then expanded for further
analysis from the original unassayed duplicate 96-well dish. These
clones were rescreened regularly over a period of 4 months, to ensure
the stability of CD44 expression, using both fluorescence-activated cell
sorting and enzyme-linked immunosorbent assays. Evaluation of VLA3«chain expression, an integrin molecule unrelated to CD44, was
performed by utilizing the monoclonal antibody J143 (saturation dilu
tion; kindly provided by Dr. M. Horton, ICRF Dominion House, St.
Bartholemew's Hospital).
2The abbreviations used are: DEM. Dulbecco's modified Eagle's medium:
BSA. bovine serum albumin: PBS. phosphate-buffered saline; SDS. sodium
dodecyl sulfate: cDNA, complementary DNA.
6660
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
Cell Aggregation Assay. The aggregation assay described by St. John
et al. (21) was used. Briefly, adherent cells were harvested with trypsin
solution as described above and washed once in DEM. Cells were
resuspended to a concentration of IO6 cells/ml in PBS containing I
mg/ml BSA and 40 mm piperazine-iY,/V"-bis(2-ethanesulfonic acid) (pH
7.0). Three hundred n\ of this single cell suspension (as determined by
microscopic examination) were placed into each well of a 24-well tissue
culture plate which was then rotated on a gyratory shaker at 4°Cfor 1
h. After this time cell mixtures were fixed by the addition of 1 ml of
PBS containing 2% paraformaldehyde. Aggregation was assessed by
counting cells in a Neubauer chamber. Cells in three independent fields
were counted, including those in aggregates and those present as single
cells. The degree of aggregation was determined as the fraction of cells
in clumps of a size greater than five cells. The effects of anti-CD44
antibodies on cell aggregation was examined by incubating cells with
Hermes-3 (10 Mg/ml) for l h at 0°Cprior to addition to the plates.
Adhesion Assay. A modified version of the assay utilized by Miyake
et al. (13) was used to determine cellular adhesion. Ninety-six-well
flexible plates (Falcon 3912 Assay Plate. Becton Dickinson) were
treated for 2 h at 37°Cwith hyaluronate solution (5 mg/ml in PBS)
before being washed with PBS and blocked with PBS/1% BSA. Cells
were labeled with Na"CrO4 (Amersham International, Little Chalfont,
United Kingdom). Specific activity. 250-500 mCi/mg chromium, as
detailed previously (22). Labeled cells (5 x IO4) in 100-/¿1
aliquots
(PBS/0.02% BSA) were added to each well and incubated at 37°Cfor
70 min before unbound cells were removed by gentle washing (3 times)
with PBS/0.02% BSA. The flexible plates were then cut up into
individual wells and the radioactivity associated with each well was
counted by using a gamma counter (LKB 1261 Multigamma). The
percentage of attached cells was calculated from the formula.
lane), and blotted onto a filter (Hybond N; Amersham Inc., Little
Chalfont). Filters were hybridized at 42°C50% formamide, 5 x Denhardt's solution [0.1% (w/v) BSA, 0.1% (w/v) Ficoll, 0.1% (w/v)
polyvinyl pyrrolidone], 5 x 0.9 M NaCl, 50 mM sodium phosphate, pH
7.7, 5 mM sodium EDTA), 0.5% SDS, and 10% dextran sulfate with a
[-"PjdCTP nick-translated CD44 cDNA probe (11) which was gener
ously supplied by Dr. B. Seed, Massachusetts General Hospital, Boston,
MA. Filters were washed at 65°Cto a final stringency of 0.2 x 0.9 M
NaCl, 50 mM sodium phosphate, pH 7.7, 5 mM sodium EDTA, and
0.2% SDS.
In y ivo Behavior of Selected Variants. Specific pathogen-free athymic
nude mice (BALB/c background) were obtained from the Animal
Breeding Unit, ICRF, Clare Hall, South Mimms, United Kingdom.
Animals were housed within plastic isolators and were used when 810 weeks old.
To assess tumorigenic capacity of the clones, cells were harvested,
adjusted to a concentration of 5 x IO6 cells/ml PBS, and mice were
given injections s.c. in the flank region of an inoculum of 0.2 ml (1 x
IO6cells). Animals were examined three times a week for the appearance
of palpable tumors and were killed when the size of this lesion ap
proached 1.5 cm at the greatest diameter.
In order to determine the lung colonization capacity of the clones,
cells were harvested as above, adjusted to a concentration of 1 x IO6
cells/ml, and the viability was assessed by trypan blue exclusion. All
cell suspensions had viabilities of 90% or greater. Mice received inoc
ulum volumes of 0.2 ml via the lateral tail vein and were killed 8 weeks
later when the lungs were removed and assessed for pulmonary tumor
nodule formation under a dissecting microscope.
RESULTS
Residual cpm x 100
% of attached cells =
Input cpm
Variant Selection and Growth Characteristics. In the study of
Schreiner et al. (18) the isolation of Chinese hamster ovary cell
Cell Migration Assay. The wound assay of Goodman et al. (23) was variants deficient in fibronectin receptor expression was
achieved by a protocol on which the present investigation was
used to determine the migratory capacity of the various clones. Har
vested cells (5 x IO1cells/ml) in DEM were plated into the wells of 24based. These workers found that a single flow cytometric sort
well plates and cultured at 37°Cfor 24-48 h until the cultures were
resulted in a population of cells with unstable levels of receptor
subconfluent. A wound track (approximately 5 mm in size) was scored
expression (18). Consequently in our experiments we utilized
in each well with the use of a plastic scraper. Replicate wells were
three batch sorts, followed by limiting dilution cloning, to
terminated at 16. 24, and 48 h after wounding by fixing and staining
obtain variants which expressed different levels of CD44 at
the cell cultures with 1% crystal violet in methanol. The stained cells
their cell surface. Fluorescence intensity histograms of four
were then examined under an inverted microscope.
high-expressing and four low-expressing clones are presented
Immunoprecipitation. Cells from the individual clones were surface
in
Fig. \A. The population means of log-integrated green fluo
labeled by the lactoperoxidase iodination procedure (24). Harvested
cells were suspended (0.2-1 x IO7 cells/ml) in PBS to which were
rescence for the high clones ranged from 40 to 60 fluorescent
unitsv whereas those of the low clones varied from 7 to 10
added, in sequence, 5 /jl sodium iodide (Sigma, United Kingdom; 5 x
10~4 M), 10 ß\glucose (0.5 M), 5 /¿Ilactoperoxidase (Calbiochem,
fluorescent units, indicating a 4- to 6-fold difference in levels
Cambridge Bioscience, Cambridge, United Kingdom; 1.8 mg/ml PBS)
of expression between the two populations.
100 fiCi sodium |'-sl]iodide (Amersham, United Kingdom), and 5 n\
These differences were very stable over an 8-week period in
glucose oxidase (20 units/ml); the suspension was mixed and incubated
tissue culture and have continued to be maintained over 4
for 15 min at room temperature and then 10 n\ sodium iodide (0.1 M) months of continuous culture (data not shown).
were added. Cells were washed 3 times with PBS and then extracted
All clones, whether low or high CD44 expressers, showed
twice in Penman's buffer (20 mM 4-(2-hydroxyethyl)-l-piperazineethcomparable levels of VLA-3« chain on their cell surface as
anesulfonic acid, 3 mM MgCl;, 50 mM NaCl, 1 mM CaCI2, 300 mM
detected by the monoclonal antibody J143 as illustrated in Fig.
sucrose, 0.5% Nonidet P-40, 1 m,Mphenylmethylsulfonyl fluoride, 0.5%
IB.
sodium-azide). Paniculate debris was removed by centrifugation; ali
quots of lysate were incubated, on ice, with Hermes-3 antibody (10 ¿ig/
Growth curves, using both the incorporation of tritiated
ml) for 45-60 min, followed by 5 ß\rabbit antiserum to mouse IgG
thymidine into DNA and cell counts as the parameter meas
(DAKO Corp., Santa Barbara, CA) for a further 45-60 min. Complexes
ured, did not identify any difference in growth rates, with all
were then removed with 50-100 M' protein A-Sepharose [50% (w/v)
clones exhibiting population doubling times of 12-14 h. There
suspension] for 60 min [Pharmacia (GB) Ltd., Milton Keynes, United
is therefore no correlation between level of receptor expression
Kingdom). The bound complexes were washed as described elsewhere
and growth capacity.
(25) and then boiled for 10 min in Laemmli buffer (26). and analyzed
Immunoprecipitation Analysis. The result of surface iodinatby SDS-polyacrylamide gel electrophoresis (8% gel). Gels were dried
ing cells from CD44 high-expressing and CD44 low-expressing
and autoradiographed with Kodak XAR-5 film at -70°C.
clones and immunoprecipitating the Nonidet P-40 extracted
Northern Blot Analysis. Total RNA was prepared by the guanidinium
isothiocyanate technique (27), and polyadenylated RNA isolated by lysates with the Hermes-3 antibody is presented in Fig. 2A.
From molecular weight standards it was clear that the immuoligodeoxythymidylate cellulose (type 7, Pharmacia, Uppsala, Sweden),
then electrophoresed through a 1% formaldehyde-agarose gel (10 Mg/ noprecipitated CD44 was of the M, 80,000-90,000 species.
6661
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
HK3HSORTS
LOW SORTS
AACLONE
LT5.1 CD44 LOW
LT5.1 CD44 HIGH
+3
90kD
II.*
B
CLONE
ABCD
I-
CLONE *6
CLONE
CLONE f7
CLONE *7
RELATIVE FLUORESCENCE
LOW 3
RELATIVE FLUORESCENCE
EFGH
Fig. 2. Immunoprccipitation of representative LT5. l CD44 high- and lowexpressing clones. A. immunoprecipitations of CD44 by Hermes-3 monoclonal
antibody (10 Mg/ml) were performed as described in "Materials and Methods,"
using I2'l-labeled cell lysates. Equal cpm «ereimmunoprecipitated and each cell
type was loaded on to an 8rr acrylamide gel and run at 50 V overnight. Gels »ere
fixed in acetic acid/glycerol before drying and were autoradiographed for 24 h.
Lanes A-D, CD44 low-expressing clones 3, 5, 6, and 7; Lanes E-H CD44 highexpressing clones I. 2. 6. and 7. B, immunoprecipitation of same clones with
JI43. and SDS-polyacrylamide gel electrophoresis under reducing conditions,
reveals approximately equal expression of VLA-3«.Ordinale, molecular weight
in thousands.
Since all lysates contained equal amounts of radioactivity (cpm)
the clones have variable amounts of labeled species at their cell
surface, with the high clones having approximately 8-fold
higher levels than the low-expressing clones as determined by
densitometric scanning. Immunoprecipitations
of the same
clones with control antibody J143 and analysis under reducing
conditions revealed the expected band at a molecular weight of
about 130,000 («,and i3¡),
showing approximately equal expres
sion by high- and low-expressing clones for this unrelated
surface molecule (Fig. 2B).
Northern Blot Analysis. RNA blot hybridization revealed
three message sizes of approximately 1.6, 2.2, and 5.5 kilobases
(Fig. 3) as reported for other melanoma cell lines elsewhere
(10, 11). The 5.5-kilobase transcript, scarcely discernible in the
example shown, always was expressed to a lesser degree than
the 1.6- and 2.2-kilobase size messages. Verification that equal
loading had been achieved was obtained by stripping the nylon
filters (30 min boiling in 0.1% standard saline citrate) and
reprobing with a ['-PjdCTP nick-translated actin cDNA probe
(Oncor probe, Hybaid Ltd., Middlesex, United Kingdom).
From densitometric measurements it was calculated that the
high-expressing clones had up to a 7-fold increase in steady
state mRNA levels for CD44 relative to the low-expressing
clones (Fig. 3).
Cell Aggregation. When single cell suspensions were incu
bated at 4°Cfor set time periods the CD44 high-expressing
clones self-aggregated to a much greater extent than the lowexpressing clones. Thus, whereas this latter group did show the
occasional cluster consisting of approximately 2-6 cells, the
high-expressing groups regularly formed large aggregates cornFig. 1. A, selection of LT5.1 CD44 high- and low-expressing clones. LTS.l
cells were surface labeled with anti-CD44 monoclonal antibody (Hermes-3, 10
ng/ml). followed by fluorescein isothiocyanate-labeled secondary antibody, batch
sorted three times, and finally single sorted into 96-well plates, as described in
"Materials and Methods." The four highest and four lowest CD44-expressing
clones were grown to confluency after enzyme-linked immunoabsorbent assay
analysis of duplicate plates. For each panel 5000 cells were analyzed. Full scale
in the >'-axis is 400 cells, with hatch marks indicating every 100 cells. The axis
spans four logs relative fluorescence intensity. B. expression of VLA-3a (detected
by J143) by both CD44 high- and low-expressing clones.
6662
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
LT5.1 CD44
5.5kb —¿
2.2kb —¿
1.6kb —¿
B
ACTIN
«
LOW
LT5.1 CD44
HIGH
•¿Mill
•¿IV*
B
D
E
H
Fig. 3. Northern blot analysis of LT5.1 CD44 RNA. A, RNA was prepared
from LT5.1 CD44 high- and low-expressing clones as described in "Materials
and Methods." Ten ¿igof mRNA/lane were loaded onto an 0.8r< agarose gel,
electrophoresed. transferred to Hybond N membranes (Amersham. Inc.. Buck
inghamshire. United Kingdom) and hybridized with "P-labeled CDW44 probes.
The transcripts seen are 5.5. 2.2, and 1.6 kilobases (kh). Lanes A-l), low CD44
expressing clones 3, 5, 6. and 7: Lanes K-H, high-expressing clones I, 2. 6. and
7. /?, filter was striped and reprobed with actin to verify equal loading.
posed of 20-40 cells. Representative fields illustrating this
phenomenon with high clone 7 and low clone 3 are presented
in Fig. 4. The observed self-aggregation was abrogated by the
presence of anti-CD44 antibody (Hermes-3, saturation dilu
tion). The degree of aggregation observed, defined by the frac
tion of cells in clumps larger than 5 cells, in the presence and
absence of monoclonal antibody, is presented in Fig. 5. Highexpressing clones I and 7 manifested aggregation of 52 and
50%, respectively, whereas the low-expressing clones 3 and 5
only had percentage aggregation values of 13 and 10%. Aggre
gation of the high-expressing clones could be reduced to the
same value as that of the low-expressing clones by the addition
of Hermes-3 antibody, although comparable treatment of the
low-expressing clones had no effect on aggregation character
istics (Fig. 5). Control antibody (J143, saturation dilution) had
no effect on the self-aggregation phenomenon (data not shown).
Cell Migration. The migration of the different clonal lines
was analyzed by using the in vitro "wound" system. Wounds of
approximately 5 mm were made in subconfluent monolayers of
(+/-
ANTI-CD44
LTS.l CD44 CLONES
MONOCLONAL
ANTMiODY)
Fig. 5. Quantitative analysis of aggregation of LTS.l CD44 high- and lowexpressing clones. Data from aggregation assays using high clones 1 and 7 low
clones 3 and 5 are presented. Aggregaties containing five cells or more were
counted in triplicate samples. Addition of anti-CD44 antibody (ah) (Hermes 3,
10 ^g/ml) to cells followed by incubation at 4°Cfor l h prior to initiation of
assay abrogated aggregation. All standard deviations were less than 10%.
the different clones and cells were allowed to migrate into the
cell-free area over a 24-h period. Representative experiments
using four clones are illustrated in Fig. 6. Here the two clones
which expressed low levels of CD44 migrated less than did the
two clones expressing high levels of CD44. The results from
the assay, which was repeated at least 3 times with virtually
identical results, indicate a correlation between CD44 expres
sion and the migrating ability of DX3.LTS.l melanoma cells.
Quantitation of data obtained from a single experiment with a
representative high and low clone (Fig. 7) showed that, in the
24-h period examined, roughly 3 times the number of cells
from the high-expressing clones migrated into the cleared area.
Moreover, incubation in the presence of anti-CD44 antibody
(Hermes-3) reduced the migratory activity of the high clone to
that of the low clone (Fig. 7).
Cell Adhesion. When two of the high-expressing clones (1
and 7) and two of the low-expressing clones (3 and 5) were
assessed for their ability to bind to a hyaluronate substrate, it
was found that there was a substantial difference in the adhesive
Fig. 4. Aggregation of LTS.l CD44 highand low-expressing clones. Representative
fields of cells from one of a number of similar
aggregation assays are presented. A, a clone of
LT5. l CD44 high-expressing cells (high clone
7) showing extensive homotypic cell aggrega
tion; B, a clone of LT5.1 CD44 low-expressing
cells (low clone 3): C. the high clone 7 in the
presence of anti-CD44 monoclonal antibody
(Hermes 3, 10 ng/ml): lì,low clone 3 in pres
ence of Hermes 3 antibody. Testicular hyaluronidase (5000 units/ml) also inhibited cellular
aggregation (data not shown).
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
o,
Fig. 6. Migratory ability of the LT5.1 C"D44 high- und low-expressing clones. Subconfluent monolayers of representative LT5. l CD44 high and low clones were
"wounded" at time 0 as described in "Materials and Methods." The cells were allowed to migrate into the cell-free area for 24 h then fixed and stained with crystal
violet. A and B. CD44 low-expressing clones 3 and 5; C and D. CD44 high-expressing clones I and 7.
weeks after s.c. injection large progressively growing tumors
were apparent in virtually all animals (Table 1) with no obvious
difference in take or growth rate apparent between those re
sulting from the high-expressing clones or low-expressing
clones. When lungs were removed from animals which had been
killed after being given injections i.v. of 5 x IO5 cells of any of
three of the CD44 low-expressing clones (clones 3, 6, and 7) or
with cells from any of three of the CD44 high-expressing clones
(clones 1, 2, and 7) a very different picture was apparent. Of a
total of 18 animals given injections of cells from the highexpressing clones, 16 showed gross evidence of lung tumor
Fig. 7. Quantitation of cell migration in LT5.1 CD44 high- and low-expressing
cells. A, wound assay was carried out as described in "Materials and Methods."
The columns show the number of cells that had migrated into a 5-mm wound
over a 16-h period in the presence or absence of Hermes 3 antibody (10 /jg/ml).
The results are derived from three randomly chosen areas of the wound. Bars,
SD. A, high-expressing clone 7: B, low-expressing clone 3; (, high-expressing
clone 7 in the presence of antibody: D, low-expressing clone 3 in presence of
antibody.
60i
u
u
at
£
Q
capabilities of these variants (Fig. 8). Thus, clones 1 and 7
achieved attachment values of 50% or greater within the culture
period, whether sulfated or nonsulfated hyaluronate was uti
lized as substrate, whereas clones 3 and 5 only exhibited less
than 20% adherence. The addition of testicular hyaluronidase
(Sigma Chemco., Poole, Dorset) to the wells (Fig. 8) or the
addition of anti-CD44 antibodies (data not shown) virtually
abolished all adherence (to 10% or less) of both the variant
populations.
In Vivo Behavior of Variant Lines. The tumorigenic capacity
of six clonal lines was assessed in athymic nude mice. By 8
50-
HYALURONATE
(sulfated)
40 -
HYALURONATE
30-
HYALURONATE
-f hyaluronidase
20 -
PBSA
10 -
LOW 3
LOW 5
(LTS.l
CD44
HIGH 1
HIGH 7
CLONES)
Fig. 8. Adhesion of LT5.1 CE44 high- and low-expressing clones to hyaluron
ate. Hyaluronate. either sulfonated or nonsulfonated (5 mg/ml), was coated onto
the surface of 96-well plates. Radiolabeled LT5.I CD44 high- and low-expressing
clones were added to the wells as described in "Materials and Methods." The
columns represent the percentage of cells added which adhered to the 96-well
plates. Standard deviation of quadruplicate determinations was lO^r or less. The
results shown arc representative of three separate experiments.
6664
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
Fig. 9. Lung colonizing capacity of LT5.1
CD44 high- and low-expressing clones. Gross
appearance of representative lungs obtained
from groups of BALB/c nude mice given injec
tions of 5 x 10' LT5.1 CD44 high-expressing
cells (high clone 7) (A) and LT5.1 CD44 lowexpressing cells (low clone 3) (B) 8 weeks
previously. Arrows indicate pulmonary tumor
colonies which stand out in contrast to the
lung parenchyma.
B
Table I Tumor incidence resulting from injection ofLT5.l-CD44highLT5.I-CD44 low-expressing clones
and
of lung
no. of lung
incidence"10010010010010060%
colonization*838310033400Mean
CloneHighlHigh
nodules
±SD43
2High?Low
±1315
±717
±92
±0.51
±0.70
3Low
6Low
7%oftumor
" Five mice/group were given injections s.c. of IxlO6 cells/animal. Animals
were killed when tumors reached 1.5 cm.
'Six mice/group were given injections i.v. of 5x10* cells/animal. Animals
were killed 8 weeks later.
nodules (median number of nodules, 30; range, 15-45), whereas
of 17 animals given injections of cells from the low-expressing
clones, only 4 showed evidence of what was a lower level of
pulmonary involvement (median number of tumor nodules, 1;
range, 0-2). Not only the number of nodules varied between
the two populations of DX-3.LT5.1 cells, but also the size of
the neoplastic lesions, as can be seen in Fig. 9. Thus the highexpressing clones produced large, very evident lung nodules,
whereas those produced by the low-expressing clones were
markedly smaller, being scarcely discernible by the naked eye.
DISCUSSION
Metastatic dissemination of solid tumors is a complex pathophysiological process consisting of several steps (28). Many of
these steps, such as entrance into small vascular or lymphatic
channels, survival during circulation, arrest at distant sites, and
extravasation, would seem to be very comparable to the steps
involved in lymphocyte migration (9, 29). It was these apparent
similarities of function that caused us to wonder whether cell
surface molecules shown to have a major role in determining
the trafficking of normal lymphocytes also had a part to play
in regulating the dissemination of cells from solid tumors.
CD44 was characterized as an adhesion molecule determining
the interaction of lymphocytes with specialized endothelium in
lymphoid tissue (1, 30); antibodies against varied epitopes of
this "homing receptor" were able to abrogate selectively this
on the analogy outlined above, CD44 was a good candidate
molecule to examine for a possible contribution to metastatic
dissemination; a feeling strengthened by the fact that CD44
expression is not limited to cells of the hematolymphoid series
(10,11,32).
Our approach has been to utilize the fluorescence-activated
cell sorter to select out populations of cells, from a single
parental line, which vary in the levels of CD44 expressed at the
cell surface. Like Schreiner et al. (18), who isolated variants
lacking the fibronectin receptor, we found that variations in
surface expression were a reflection of variations in steady-state
mRNA levels (Fig. 3), and that the selected clones exhibited a
stable phenotype. Upon injection into immunoincompetent
athymic nude mice there were substantial differences between
the CD44 high- and CD44 low-expressing clones such that the
high-expressing clones produced substantially more, and dra
matically larger, pulmonary nodules than did the low-express
ing clones (Fig. 9). That each of the three clones from either
category manifested similar behavior made it unlikely that these
differences were simply a random consequence of préexistent
cellular heterogeneity (33), and this fact was underlined by the
precise duplication of these findings when similar variants from
the murine B16 melanoma system were obtained and analyzed.1
These results suggest strongly that, as predicted, the CD44
molecule may play an important part in affecting the dissemi
native capacity of the DX3.LT5.1 melanoma cells.
The determination that CD44 functions as the principal cell
surface receptor for the glycosaminoglycan, hyaluronate (12,
13) offers several possibilities for explaining the mechanistic
basis of this observed increase in lung-colonizing capacity.
Tumors of many types have been characterized as exhibiting
high levels of hyaluronate accumulation (16), and transformed
cells frequently manifest high levels of hyaluronate-binding
activity (17). Hyaluronate is widely distributed throughout the
body but has been found to be present in greatest amounts in
lung tissue (34), raising the possibility that the observed de
crease in pulmonary colonization by the low-expressing clones
was a reflection of decreased attachment of i.v. introduced
tumor cells. Certainly in in vitro assays the clones selected for
higher levels of CD44 expression were better able to adhere to
binding propensity (31). Accordingly, we reasoned that, based
' Unpublished observations.
6665
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
hyaluronate substrata than their low-expressing counterparts
(Fig. 8).
Equally, it has long been known that hyaluronate plays an
important role in mediating cell to cell interaction (35-37).
Interestingly, when St. John et al. (21) transfected mouse Lcells with CD44 cDNA they found an increase in self-aggrega
tion in high-expressing clones. It was felt highly likely that this
aggregation phenomenon probably was a consequence of the Lcells expressing at least one ligand for the transfected primate
CD44 (21). Whether the interaction observed in our studies
results from a receptor-ligand-receptor interaction or a homophilic receptor-receptor interaction is not known but, indisput
ably, the clones expressing greater amounts of CD44 aggregated
to a considerably greater extent than did those expressing low
levels of CD44 (Fig. 5). Since homotypic clumping has been
shown to increase the experimental metastatic capacity of tu
mor lines (38), it seems quite probable that this effect could
have contributed to the in vivo results presented here.
Finally, it has been known that the hyaluronate receptor is
linked to the actin filaments (39) and Pgpl has been associated
with cell movement of fibroblasts (5, 40), making it seem likely
that the CD44 molecule may facilitate the locomotory behavior
of cells. In our wound migration assay we have noted that the
high-expressing clones are indeed more motile than the lowexpressing clones (Fig. 7). Given the importance of cellular
motility in the invasive and metastatic process (41), it is tempt
ing to speculate that this characteristic also may well have
contributed to the in vivo behavior of the various clones.
The binding of lymphocytes to high endothelial venules of
lymphoid tissue is regulated by more homing receptors than
just the CD44 molecule (42). Both the "selectin" LECAM-1
and the integrin VLA-4 molecules are known to play an impor
tant role in determining lymphocyte adhesion, and it is probable
that the binding behavior of melanoma cells also is a conse
quence of coordinate regulation of a variety of surface receptors
(43). Interestingly, therefore, the results obtained with the J143
antibody suggest that the selection procedure utilized herein
did not result in variation of all cellular adhesion molecules.
Equally, as we have tried to indicate by reference to homotypic
aggregation and migratory capacity, it is possible that the results
obtained have little to do with initial adhesive interactions with
endothelium. Currently, we are trying to determine the exact
role that CD44 expression plays in regulating melanoma dis
semination to explain the strong correlations reported herein.
REFERENCES
1. Jalkanen, S.. Jalkanen, M.. Bargatze. R., Tammi. M., and Butcher. E. C.
Biochemical properties of glycoproteins involved in lymphocyte recognition
of high endothelial venules in man. J. Immunol.. 141: 1615-1623. 1988.
2. Omary, M. B.. Trowbridge, I. S.. Letarte. M.. Kagnoff. M. F., and Isacke. C.
M. Structural heterogeneity of human Pgp-1 and its relationship with p85.
Immunogenetics. 27: 460-464. 1988.
3. Jalkanen. S.. Bargatze. R. F., Herrón. L. R., and Butcher, E. C. A lymphoid
cell surface glycoprotein involved in endothelial cell recognition and lympho
cyte homing in man. Eur. J. Immunol.. 16: 1195-1202. 1986a.
4. Trowbridge. I. S. Identification and characterisation of the human pgp-1
glycoprotein. Immunogenetics. 23: 326-332, 1986.
5. Carter. W. G., and Wayner. E. A. Characterisation of the class III collagen
receptor, a phosphorylaled transmembrane glycoprotein expressed in nu
cleated human cells. J. Biol. Chem.. 263: 4193-4201, 1988.
6. Picker, L. J., de los Tojos. J., Telen. M. J.. Haynes. B. F.. and Butcher. E.
C. Monoclonal antibodies against the CD44 [In (Lu)-related p80], and Pgp1 antigens in man recognise the Hermes class of lymphocyte homing recep
tors. J. Immunol.. 142: 2046-2051. 1989.
7. Gallatin. M.. St. John, T. P., Siegelman, M.. Reichert, R., Butcher, E. C.,
and Veissman. I. L. Lymphocyte homing receptors. Cell, 44:673-680, 1987.
8. Jalkanen. S., Reichert. R. A.. Gallatin, W. M., Bargatze, R. F., Weissman,
I. L., and Butcher, E. C. Homing receptors and the control of lymphocyte
migration. Immunol. Rev., 91: 39-60, 1986.
9. Sher, B. T., Bargatze, R.. Holzmann, B., Gallatin, W. M, Mathews. D., Wu,
N., Picker. L., Butcher, E. C.. and Weissman, I. L. Homing receptors and
metastasis. Adv. Cancer Res., 51: 361-390. 1988.
10. Goldstein, L. A., Zhou, D. F. H., Picker, L. J., Minly, C. N., Bargatze. R.
F., Ding, J. F.. and Butcher, E. C. A human lymphocyte homing receptor,
the hermes antigen, is related to cartilage proteoglycan core and link proteins.
Cell. 56: 1063-1072, 1989.
11. Stamenkovic, I., Amiot, M., Pesando, J. M., and Seed, B. A lymphocyte
molecule implicated in lymph node homing is a member of the cartilage link
protein family. Cell, 56: 1057-1062, 1989.
12. Aruffo, A., Stamenkovic, I., Melnick, M.. Underbill, C. B., and Seed. B.
CD44 is the principal cell surface receptor for hyaluronate. Cell, 61: 13031313, 1990.
13. Miyake, K., Underbill, C. B., Lesley. J.. and Kincade, P. W. Hyaluronate
can function as a cell adhesion molecule and CD44 participates in hyaluron
ate recognition. J. Exp. Med., 172: 69-75. 1990.
14. Toóle, B. P. Glycosaminoglycans in morphogenesis. In: E. Hay (ed.). Cell
Biology of the Extracellular Matrix, pp. 259-294. New York: Plenum Press,
1981.
15. West, D. C., Hampson, I. N., Arnold, F., and Kumar. S. Angiogenesis
induced by degradation products of hyaluronic acid. Science (Washington
DC), 228: 1324-1326. 1985.
16. Toóle, B. P.. Biswas, C., and Gross, J. Hyaluronate and invasiveness of the
rabbit V2 carcinoma. Proc. Nati. Acad. Sci. USA, 76: 6299-6305, 1979.
17. Nemec, R. E., Toóle, B. P.. and Knudson, V. The cell surface hyaluronate
binding sites of invasive human bladder carcinoma cells. Biochem. Biophys.
Res. Commun., 149: 249-257. 1987.
18. Schreiner, C. L.. Bauer, J. S.. Danilov, Y. N.. Hussein, S., Sczekan, M. M.,
and Juliano. R. L. Isolation and characterisation of Chinese hamster ovary
cell variants deficient in the expression of fibronectin receptor. J. Cell Biol.,
109: 3157-3167. 1989.
19. Ormerod. E. J.. Everett, C. A., and Hart, I. R. Enhanced experimental
metastatic capability of a human tumor line following treatment with 5azacytidine. Cancer Res., 46: 884-890, 1986.
20. Durbin, H., and Bodmer, W. F. A sensitive micro-immunoassay using ßgalactosidase/anti-fÃ-- galactosidase complexes. J. Immunol. Methods, 97:
19-27, 1987.
21. St. John, T.. Meyer, J., Idzerda, R., and Gallatin, W. M. Expression of CD44
confers a new adhesive phenotype on transfected cells. Cell, 60:45-52, 1990.
22. Ormerod, E. J., Everett, C. A., and Hart, I. R. Adhesion characteristics of
human melanoma cell lines of varying metastatic potential. Int. J. Cancer,
41: 150-154, 1988.
23. Goodman. S. L., Vollmers, H. P.. and Birchmeier, W. Control of cell
locomotion perturbation with an antibody directed against specific glycopro
teins. Cell. 14: 1029-1038, 1985.
24. Hubbard. A. L., and Cohn, Z. A. The enzymatic iodination of the red cell
membrane. J. Cell Biol.. 55: 390-405. 1972.
25. Kellie, S., Patel. B.. Mitchell. A.. Critchley, D. R., Wigglesworth. N. M.. and
Wyke, J. Comparison of the relative importance of tyrosine-specific vinculin
phosphorylation and the loss of surface-associated fibronectin in the mor
phology of cells transformed by Rous sarcoma virus. J. Cell Sci.. 82: 129142, 1986.
26. Laemmli, U. K. Cleavage of structural proteins during assembly of the head
of bacteriophage T4. Nature (Lond.), 227: 680-685. 1970.
27. Maniatis. T., Fritsch, E. F., and Sambrook, J. Molecular Cloning: A Labo
ratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory,
1982.
28. Hart, I. R., Goode. N. G., and Wilson, R. E. Molecular aspects of metastatic
spread. Biochem. Biophys. Acta. 989:65-84, 1989.
29. Cowans, J. L., and Knight, E. J. The route of recirculation of lymphocytes
in the rat. Proc. R. Soc. Lond. B Biol. Sci., 159: 257-282. 1964.
30. Haynes. B. F.. Telan. M. J., Hale, L. P., and Denning, S. M. CD44—a
molecule involved in lymphocyte homing, leukocyte cellular adherence and
T cell activation. Immunol. Today. 10:423-428. 1989.
31. Jalkanen, S., Bargatze, R. F., de los Toyos, J., and Butcher. E. C. Lymphocyte
recognition of high endothelium: antibodies to distinct epitopes of an 85-95
kd glycoprotein antigen differentially inhibit lymphocyte binding to lymph
node, mucosal, or synovial endothelial cells. J. Cell Biol., 105: 983-990,
1987.
32. Picker, L. J., Nakache, M.. and Butcher, E. C. Monoclonal antibodies to
human lymphocyte homing receptors define a novel class of adhesion mole
cules on diverse cell types. J. Cell Biol.. 109: 927-937, 1989.
33. Fidler, I. J.. and Hart, I. R. Biological diversity in metastatic neoplasms:
origins and implications. Science (Washington DC). 217: 998-1003. 1982.
34. Green. S. J., Tarone. G.. and Underbill. C. B. Distribution of hyaluronate
6666
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
CD44 VARIANTS FROM A HUMAN MELANOMA
and hyaluronate receptors in the adult lung. J. Cell Sci., 89: 145-156. 1988.
35. Pessac, B., and Defendi, V. Cell aggregation: role of acid mucopolysaccharide.
Science (Washington DC) /75-898-900
1972
36. Underbill. C. B., and Dorfman. A. The role of hyaluronic acid in intercellular
adhesion of cultured mouse cells. Exp. Cell Res., // 7: 155-164, 1978.
,
,
37. Underbill. C. B., and Toóle, B. P. Binding of hyaluronate to the surface of
cultured cells. J. Cell Biol., 82:475-481, 1979.
actin filaments. J. Cell. Biol.. 105: 1395-1404, 1987.
40. Hughes, E. N., Mengood. G., and August, J. T. Murine cell surface glycoprotein. Characterisation of a major component of 80,000 dallons as a
^SSf^^î^*
^^
°fmesenchymal Ce"S' J' Bi°''Chem"
, ,,,•'„,„
.
«_ . •¿ 41. Liotta, L. A., and Shiffman. E. Tumour motility factors. Cancer Surv. 7:
63I-6S2 1988
42 Springer,' T. A. Adhesion receptors of the immune system. Nature (Lond.),
38. Fidler. I. J. The relationship of embolie homogeneity number size and
viability to the incidence of experimental metastasis. Eur. J. Cancer, 9: 223227, 1973.
39. Lacy, B. E., and Underbill, C. B. The hyaluronate receptor is associated with
34^: 425-434, 1990.
43. R|ce, G. E., and Bevilacqua, M. P. An inducible endothelial cell surface
glycoprotein mediates melanoma adhesion. Science (Washington DC), 246:
1303-1306, 1989.
6667
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
Isolation and Characterization of Human Melanoma Cell
Variants Expressing High and Low Levels of CD44
Mary Birch, Stephen Mitchell and Ian R. Hart
Cancer Res 1991;51:6660-6667.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/51/24/6660
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 15, 2017. © 1991 American Association for Cancer Research.