| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Tumor Biology |
University of Pittsburgh Cancer Institute and Department of Pathology, University of Pittsburgh, Pittsburgh, Pennsylvania 15213 [Z. Z., X. H., Q. Z., E. G.]; Ontario Cancer Institute, Toronto, Ontario, M5G 2M9 Canada [O. S-S., R. K.]; Experimental Therapeutics Section, Laboratory of Experimental Immunology, National Cancer Institute-Frederick Cancer Research Development Center, NIH, Frederick, Maryland 21702 [R. W.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(CD49d) gene. In parallel, they
showed a reduction in homotypic aggregation and binding to endothelial
cells, increased Matrigel invasiveness, and decreased matrix
metalloproteinase-2 and matrix metalloproteinase-9 activity that
paralleled up-regulation of the TIMP-1 gene. S91Nadh
cells also manifested changes in cell surface carbohydrates, such as
appearance of -galactosyl epitopes as a result of up-regulation of
the 1,3-galactosyltransferase gene and concomitant reduction in cell
membrane sialylation. Thus, selection of S91 melanoma cells for anoikis
resistance resulted in an increase in their metastatic potential in
parallel with multiple alterations in their phenotypic properties. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although these findings are based on the in vitro study of cultured cell lines, it is believed that anoikis also occurs in vivo and might have important biological significance. Cell adhesion to extracellular matrix provides an important survival signal that also might induce expression of various genes, leading to cell proliferation and differentiation. Cells that have lost such contact and detach from the tissue are programmed to die via induction of anoikis (1) . Anoikis was probably an important evolutionary mechanism that emerged in multicellular organisms to maintain the integrity and function of various tissues and organs by elimination of cells that lost contact with the parental tissue or organ, thus preventing their survival and function in new anatomical locations. A basic rule for many cells in the body, then, is "attach or die." However, this is true only for most cells of epithelial origin, whereas migrating cells, such as blood cells, are anoikis resistant.
It is well documented that cell attachment is mostly mediated by integrins, although other cell adhesion molecules can be also involved (3 , 4) . Extracellular matrix-integrin interactions trigger intracellular signaling and activation of certain genes, leading to cell proliferation and differentiation (4) . However, it is unclear how the apoptotic signal is triggered in cells in which integrin-ligand interaction was abrogated. It is believed that anoikis is not activated via ligand activation of CD95 or other death receptors (5 , 6) . Nonetheless, death receptor activation is somehow involved in anoikis. This conclusion is based on the findings that dominant-negative Fas-associated death domain protein as well as silencers of death domains efficiently inhibited anoikis (5 , 6) . It was found that anoikis is mediated via activation of caspases, particularly caspase-8 and caspase-3 (5) . It was shown that inhibitors of apoptosis, crmA or bcl-2 and bcl-XL, also inhibit anoikis (7) . Some evidence also indicates that the endogenous galactoside-binding lectin, galectin-3, plays a role in inhibition of anoikis (8) . It was demonstrated that the cytoplasmic tail of galectin-3 contains a four-amino acid motif (NWGR) that is also present in the BHI domain of the bcl-2 gene, and it was found to be crucial for the antiapoptotic effect of both bcl-2 as well as galectin-3 (9) .
Various protein kinases that are activated/inactivated during cell attachment/deattachment were implicated in the prevention or induction of anoikis. It was shown that integrin-ligand interaction leads to activation of focal adhesion kinase. Some data indicate that activation of focal adhesion kinase might prevent induction of anoikis (10) . Similarly, attachment-mediated activation of phosphatidylinositol 3-kinase and protein kinase B/Akt protects cells from anoikis (11) . In contrast, inhibition of phosphatidylinositol 3-kinase and protein kinase B/Akt results in an induction of anoikis (11) . It was also found that disruption of the cultured cell-matrix interaction leads to activation of JNK and p38, a related stress-activated protein kinase (7) . Possible involvement of these kinases in anoikis is suggested by experiments in which transfection and overexpression of JNK4 and p38 resulted in anoikis (7) , although some reports showed no correlation between activation of JNK and induction of anoikis (12) . Recently, the importance of mitogen-activated protein kinase kinase kinase in anoikis has been demonstrated (13) .
Anoikis has been mostly investigated using normal cultured cell lines.
Normal cells usually fail to grow under anchorage-independent
conditions in semisolid agar. In contrast, malignantly transformed
cells are able to grow in semisolid agar (14)
. The ability
of the transformed cells to grow anchorage independent in a semisolid
agar became a classical assay for in vitro evaluation of
malignant transformation of various cultured cell lines
(14)
. Analysis of high and low metastatic tumor cell lines
showed no correlation between their ability of anchorage-independent
growth in semisolid agar and their metastatic potential
(15)
. Anchorage-independent growth of malignant but not
normal cells in semisolid agar might indicate that tumor cells are
resistant to anoikis. However, the sensitivity of malignant cells to
anoikis was not extensively investigated. Some data indicate that tumor
cells could be sensitive to anoikis. Human melanoma cells lacking
vß3 integrin showed a
high level of apoptotic death when they were cultured in
three-dimensional dermal collagen. Transfection of these cells with an
v cDNA restored
vß3 expression and
prevented their deaths (16)
.
It was shown that prevention of B16F1 melanoma cells adhesion to plastic by culture them on PolyHema-coated flasks affects their cell shape and ability to form metastases (17) . Anoikis in these cells was not tested, but melanoma cells survived and proliferated, although their doubling time increased from 16 to 24 h. B16F1 melanoma cells under this condition showed round morphology and grew in clusters as spheroids. B16F1 melanoma cells cultured on PolyHema-coated plates showed an increase in homotypic aggregation and a marked increase in their propensity to establish metastases. However, all of these changes were unstable and reversed 24 h after returning them to uncovered flasks (17) .
Numerous studies demonstrated that integrin-substrate interactions are important for tumor cell arrest, migration, and growth in different anatomical locations and formation of distant metastases (3 , 18, 19, 20) . Tumor cells inoculated i.v. are usually eliminated quickly. About 0.1–0.01% of inoculated cells develop metastatic foci in the lungs (21) . Although the mechanisms responsible for tumor cell destruction in the blood remain unclear, it was found that natural killer cells play a role in tumor cell elimination. In natural killer cell-depleted mice, tumor cell survival increased, but a vast majority of tumor cells were destroyed (22) . During hematogenic migration, metastatic cells are unable to adhere, and under these conditions, they may die as a result of anoikis. Thus, tumor cells that are unable to survive would be unable to form distant metastasis. If so, selection of tumor cells for adhesion independence might increase their resistance to anoikis and their ability to survive in the blood and develop metastatic tumors in distant anatomical locations.
In the present study, we tested our hypothesis. To study the possible role of anoikis in metastasis formation, we selected a stable subline of Cloudman S91 melanoma with high resistance to anoikis. The phenotypic and metastatic properties of anoikis-sensitive and anoikis-resistant S91 melanoma cells have been investigated.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Tumor Cells.
Cloudman S91 melanoma cells (a gift from Dr. Vincent Hearing, National
Cancer Institute, NIH) were cultured in RPMI 1640 supplemented with
10% fetal bovine serum, L-glutamine, streptomycin, and
penicillin (all from Life Technologies, Inc., Gaithersburg, MD), These
cells are strongly adherent to the plastic. Cells were usually
transferred after a short trypsin-EDTA (Life Technologies, Inc.,
Gaithersburg, MD) digestion. To select a nonadherent S91 cell
population, cells found floating in the cultured medium were
centrifuged, pelleted, resuspended in 5 ml of complete RPMI 1640, and
transferred into a T-25 culture flask. When cells in these flasks
become saturated, floating cells were again harvested and transferred
into a new T-25 culture flask. Such transfers of the floating cells
were repeated for seven generations, when a stable subline of
nonadherent S91 (S91Nadh) cells was established. These cells were
always transferred as nonadherent cells that were found floating in the
medium.
Anoikis Induction.
To induce anoikis, cells were prevented from adhering to the plastic of
the cultured flasks. This was achieved by two methods: rocking of the
cells or culturing them in a dish coated with PolyHema. Adherent and
nonadherent S91 cells (1 x 106
cells/3 ml of culture medium) were transferred into a Petri dish (60
mm). To prevent cell adhesion, dishes were placed on the platform of
the Orbitron Rotator 260250 (Boekel Industries, Inc.) for 18 h at
37°C. The platform was set up at about 10° angle, and it moved at a
speed of 30 cycles/min. Under these conditions, the dishes were gently
rocked in different directions so that the constant movement of culture
medium prevented cell adhesion to the plastic.
PolyHema coating was performed as described (8 , 17) . PolyHema (Sigma) was dissolved in 95% ethanol at a final concentration 120 mg/ml, diluted 1:10 in ethanol, and pipetted into a Petri dish (60 mm in diameter) at 0. 95 µl/mm2. After drying, dishes were washed three times with PBS, and 1 x 106 cells in 3 ml of culture medium were transferred into the dish. The PolyHema-coated and uncoated dishes were cultured for 18 h at 37°C.
Apoptosis Assay.
Apoptotic cells from PolyHema-coated or rocked dishes were analyzed
using the Vybrant Apoptosis Assay kit #4 (Molecular Probes, Eugene,
OR). This kit contains the green fluorescent YO-Pro-1 dye that can
enter apoptotic but not normal viable cells. Necrotic cells were
stained with PI. Cells were washed, resuspended in 0.5 ml of PBS, and 5
µl of the YO-Pro-1 and PI (diluted 1:10) were added. Cells were
incubated for 20–30 min on ice and were analyzed by flow cytometry,
measuring the fluorescence emission at 530 nm (FL1) and >575 nm (FL3).
Cell Proliferation.
To test for tumor cell proliferation under normal and anoikis-inducing
conditions, S91 adherent and nonadherent cells (1 x 106 cells/3 ml of medium/well) were plated into
six-well plates that were coated or uncoated with PolyHema. The plates
were cultured at 37°C under rocking conditions or under regular
stationary conditions, and the number of tumor cells/plate was
determined during 3–4 days of culture.
Cell Cycle Analysis.
Adherent and nonadherent S91 cells (1 x 106) cultured under normal or rocking conditions
were fixed in ethanol at -20°C. After washing, the pellet was
incubated with RNase A (75 units/ml) at 37°C for 30 min. Cells were
washed and resuspended in 0.5 ml of PBS containing 15 µg/ml of PI.
Cell cycle distribution analysis was performed by flow cytometry and
was analyzed using software Modlif LT (Verity Software House,
Topsham, ME).
Cell Adhesion Assay.
The murine endothelial cell line was originally established from the
lungs by Auerbach et al. (23)
and was kindly
provided by Dr. C. Johnson (University of Pittsburgh Cancer Institute).
Endothelial cells were distributed into 96-well plates (2 x 105/well), and 24–48 h later, S91
melanoma cells (4 x 105) were
added into each well containing a monolayer of endothelial cells (four
wells/group). After 45 min of incubation at 37°C, nonadherent cells
were removed by vigorous washing three times with PBS containing 2% of
FBS. Removed cells were plated into 24-well plates, incubated at 37°C
for 2 h, and their numbers were determined using the
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as
described previously (24)
. The percentage of S91 cells
adherent to endothelial cell monolayer was calculated:
 |
Homotypic Aggregation.
Single-cell suspensions of S91 melanoma cells were resuspended in 0.5
ml of complete RPMI 1640 supplemented with 10% FCS (2.5 x 106 cells/per tube) and incubated at
37°C for 1 h with intermittent agitation. Some cells were
incubated in the presence of PS/2 hybridoma supernatant containing
anti-VLA-4 mAb (final dilution, 1:10). Cells were washed by adding 2
ml of complete RPMI 1640, and pellets were resuspended in 1 ml of RPMI
1640 with a large-bore plastic Pasteur pipette (the bulb was squeezed
10 times; Ref. 24
). Single, nonaggregating cells were
counted in a hemocytometer. The percentage of aggregated cells was
calculated as (1 - Ne/Nc x 100, where Ne
is the number of single cells after incubation at 37°C and
Nc is the number of single cells before incubation
(24)
.
Flow Cytometric Analysis.
To test MHC class I expression, S91 melanoma cells (5 x 105 cells/tube) were incubated with
anti-H-2Kd (hybridoma 31-3-4) and
anti-H-2Dd (hybridoma 34-5-8) for 30 min at
4°C, washed, and stained with antimouse IgG-PE as a second antibody
(25)
. To analyze expression of -galactosyl epitopes and
SBA binding carbohydrates, S91 melanoma cells were incubated with the
biotinylated GS1B4 or SBA lectin (10 µg/ml) for 30 min at 4°C,
washed, and stained with avidin-PE. To evaluate the masking effect of
sialic acid on expression of SBA binding carbohydrates, S91 melanoma
cells were treated with 0.1unit/ml of neuraminidase type V from
Clostridium perfingens (Sigma Chemical Co., St. Louis, MO)
for 30 min, washed, and stained with biotinylated SBA lectin and
avidin-PE (26
, 27) . Integrin expression by S91 melanoma
cells was analyzed using the following anti-integrin antibodies:
hamster antimouse VLA-1 (CD49a), hamster antimouse VLA-2 (CD49b;
PharMingen, San Diego, CA), rat antimouse VLA-4 (CD49d) PS/2 mAb
(American Type Culture Collection, Rockville, MD), and rat antihuman
VLA-6 (CD49f; Amac, Inc., Westbrook, ME). After washing, cells were
stained with species-specific secondary FITC-labeled antibodies.
Analysis of fluorescence was performed on at least 5000 cells using a
FACStar or FACStar-Plus flow cytometer (Becton Dickinson
Immunocytometry, San Jose, CA; Refs. 26
and
27
).
Northern Blot Analysis.
To test TIMP-1 gene expression, total cellular RNA was
prepared from cell lines using a RNA extraction kit (RNAzol B;
Rel-Test, Inc., Friendswood, TX). The extracted RNA (10 µg/lane) was
electrophoresed on 0.7% agarose-formaldehyde gel and transferred to
nylon membranes and UV fixed before hybridization. The
32P-labeled probe used for hybridization was the
cDNA of TIMP-1 prepared by random priming (Life Technologies, Inc.,
Grand Island, NY) as described previously (24)
.
RT-PCR Analysis.
To perform RT-PCR analysis of VLA-4 and 1,3GT message in S91
melanoma cells, total RNA was extracted as described (24)
.
Four µg were subjected to a one-step RT-PCR reaction (Life
Technologies, Inc., Grand Island, NY) using the VLA-4 specific
primers and Taq polymerase (Life Technologies, Inc., Grand Island,
NY). The primers used were: upstream 5'-GTCTTCATGCTCCCAACAGC-3'; and
downstream 5'-ACTTCTGACGTGATTACAGGAAGC-3' (19)
. PCR was
run in the sequence of 94°C l min for denaturation, 55°C 1 min for
annealing, and 72°C 1 min for extension with up to 30 cycles in
total. Similarly, RT-PCR analysis of 1,3GT message was performed
using the following primers: upstream 5'-GTGGTTGTCGTGTTTTGCGAATAT-3';
and downstream 5'-ATCTGAAGGCAGGCCTATCTGATA-3'. RT-PCR was also
performed with the primers specific for the
ß-actin gene under the same conditions. After
amplification, PCR products were separated by 1.5% agarose gel
electrophoresis, stained with ethidium bromide, and visualized with UV
light (26)
.
In Vitro Invasion Assay.
Tumor cell invasiveness was tested using a modification of the Matrigel
method (24
, 28)
. Melanoma cells (5 x 104
in 0.2 ml of medium) were loaded on the
Matrigel-coated polycarbonate membrane in the upper compartment of
Transwells (Collaborative Research, Inc., Waltham, MA) in a
24-well plate. In the control group, melanoma cells were loaded into
upper part of the inserts not coated with Matrigel. After 48 h,
the nonpenetrating cells on the top of the filter were removed by
"scrubbing" with a cotton-tipped swab. The membranes were cutoff
and stained with the Diff-Quik stain, and the total number of invaded
cells on the back of the filter was counted under the microscope. After
removal of the inserts, cells that had penetrated the filters and
sedimented to the bottom of the well were cultured for an additional
48 h, and then they were stained and photographed. The percentage
of Matrigel invasion was:
 |
Gelatinase Activity of S91Adh and S91Nadh Melanoma Cells.
Melanoma cells were cultured to near-confluence in complete RPMI 1640.
The cell monolayers were rinsed with PBS and cultured with serum-free
RPMI 1640 for 24 h. Supernatants were collected and centrifuged to
remove floating cells and cell debris. Gelatinase activity of tumor
cell supernatants was tested as described previously (24)
.
The protein concentration in each conditioned medium was determined by
Bradford’s protein assay (Bio-Rad, Richmond, CA). Protein-standardized
samples were resolved in 10% SDS-PAGE containing 1 mg/ml gelatin
(Sigma). For this purpose, gelatin was solubilized in water by warming
to 60°C while stirring. After gel electrophoresis, incubation with
2.5% Triton X-100 for 30 min was followed by enzymatic degradation of
the substrate in a buffer containing 0. 1 M Tris-HCl (pH
8.0), 5 mM CaCl2, and 0.04%
NaN3 for 48 h at 37°C. Gels were stained
with 2.5% Coomassie Blue for 30 min and destained with methanol:acetic
acid:water (40:10:50). To verify equal protein loading in each lane,
parallel gels were silver stained (Silver Stain Plus kit; Bio-Rad,
Hercules, CA).
In Vivo Survival of S91 Melanoma Cells.
S91Adh and S91Nadh melanoma cells were labeled with
[111In]Ox (Amersham Life Science Inc.,
Arlington Heights, IL) as described previously (29)
. Tumor
cells (10 x 106/ml) were
incubated with 10 µCi of [111In]Ox for 10 min
at room temperature in complete RPMI 1640. The cells were then washed
three times in 15 ml of medium, counted, and resuspended at 1 x 106 cells/ml. Radiolabeled cells were
inoculated i.v. into DBA/2 mice (2.5 x 105 cells/mouse); 5 min, 5 h, and 24 h
after tumor cell inoculation, five mice/time point were sacrificed, and
lungs were removed. The level of radioactivity remaining in the lungs
was detected using the gamma counter (29)
.
Metastatic Properties of Melanoma Cells.
DBA/2 or B6D2F1 mice were inoculated i.v. with
1 x 106 S91 adherent and
nonadherent cells (five mice/group). Lungs were harvested 30 days
later, and metastatic foci in the lungs were counted under a dissecting
microscope (26)
. The experiments were repeated twice. The
significance of the differences between numbers of metastases was
determined by the Mann-Whitney U test.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Similarly, prevention of cell adhesion in PolyHema-coated, six-well
plates also resulted in an increase of apoptosis of S91Adh cells.
However, some cells growing on PolyHema formed clumps and survived in
these clumps. When PolyHema-coated plates were rocked, clumping of
cells was prevented, and the number of apoptotic S91Adh cells increased
further (Fig. 1)
. In contrast, no changes in survival of S91Nadh cells
were observed when these cells were cultured under rocking conditions
and/or on PolyHema (Fig. 1)
.
|
), indicating the involvement of caspases in the death of
these cells. We next tested whether selection of S91Nadh cells for
resistance to anoikis was associated with resistance to other apoptotic
signals. We showed previously that WGA lectin is highly efficient for
inducing apoptosis in various normal and malignant cells
(17)
. S91Adh and S91Nadh cells showed equal sensitivity to
the apoptotic effects of WGA (10 µg/ml). Survival of these cells
equally increased in the presence of the caspase inhibitor Z-VAD-fmk
(Fig. 2B)
. Similarly, both sublines were equally sensitive
to the apoptotic effect of VP16 (data not shown). Thus, resistance of
S91Nadh cells to anoikis was not a result of a general blockage of the
apoptotic cascade.
|
A, under normal culture conditions S91Nadh cells manifested a
significantly higher rate of proliferation than S91Adh cells.
Similarly, S91Nadh cells inoculated into DBA/2 mice showed a higher
rate of tumor growth than S91Adh cells (Fig. 3B)
. Prevention
of cell adhesion by rocking or PolyHema-coated culture dishes did not
significantly affect growth of S91Nadh cells, whereas growth of S91Adh
cells was prevented completely. In fact, their numbers actually
reduced, and very few survived (Fig. 3A)
.
|
. When adhesion of cells was prevented by constant rocking of the
plates, the percentage of apoptotic (hypodiploid cells, R1 region) in
S91Adh cells increased from 13 to 24%. Analysis of the cell cycle
status of the remaining nonapoptotic cells was performed using Modfit
software (Verity Software House). The results indicate that S91Adh
cells have more cells in
G0-G1 phase than S91Nadh
cells (53.7 and 29.6%, respectively), whereas S91Nadh cells had higher
proportion of cells in G2 + M phase
(33.3%) than S91Adh cells (10.16%). These differences in cell cycle
might be responsible for the observed differences in their cell
proliferation. Under anoikis-induced conditions, the vast majority
(82.5%) of rocking S91Adh cells was blocked in
G0-G1 phase, with
concomitant reduction of cells in S (from 36.2 to 12%) and
G2 + M phase (from 10.2 to 5.5.%).
Rocking of S91Nadh cells showed no change in the proportion of cells in
G0-G1 phase and an increase
up to 65.1% of cells in the S-phase. Thus, disruption of S91Adh cell
adhesion to the substrate induced blockage of transition from the
G1 to S-phase that was not observed in anoikis
resistance of S91Nadh cells.
|
. These data demonstrate that selection of S91 for resistance to
anoikis was associated with a significant increase in their ability to
form metastasis.
|
4% of cells in the lungs (3276 ± 430 and 3164 ± 178 cpm, respectively). At the 24-h time
point, only 1% of inoculated S91Adh or S91Nadh remained in the lungs
(1200 ± 304 and 939 ± 206 cpm,
respectively). Because these cells are different in their
susceptibility to anoikis, these results indicate that resistance to
anoikis did not increase survival of S91Nadh cells in the blood. It is
most likely that anoikis was not induced, because shortly after
entering the blood, S91Adh cells were trapped in the capillary bed of
the lungs and quickly adhered to the endothelial cells that prevented
anoikis in these cells.
Thus, cell survival in the blood cannot be responsible for the observed
increase in metastatic ability of S91Nadh cells. Therefore, we
investigated other mechanisms that might be responsible for the
observed differences in the metastatic potentials of these cells.
Because anoikis resistance of S91Nadh cells was associated with a loss
of cell adhesion and because integrins play a key role in cell adhesion
to various substrates as well as in metastasis formation (2
, 3)
, we analyzed the expression of various integrins by S91Adh
and S91Nadh cells. Neither S91Adh nor S91Nadh cells expressed VLA-1
and VLA-2 (data not shown). Both sublines expressed VLA-6, and the
only detectable difference in expression of integrins was found for
VLA-4, where S91Nadh cells lost expression of VLA- 4 (Fig. 5A)
. RT-PCR analysis revealed that S91Adh cells express
VLA-4 (CD49d), whereas no message for VLA-4 was found in S91Nadh
cells (Fig. 5B)
. Thus, loss of adhesion of S91Nadh cells was
associated with a loss of VLA-4 (CD49d) that was attributable to
down-regulation of VLA-4 gene expression.
|
(CD49d) subunit. VLA-4 is a
heterodimer consisting of 4 (CD49d) and ß1 (CD29) units
(32)
. Therefore, VLA-4-positive cells could adhere to each
other via VLA-4/ß1 interaction with VLA-4 on another cells,
leading to homotypic and heterotypic adhesion or aggregation. Such
aggregation and formation of the clumps in the blood stream might
increase tumor cell survival, capillary embolization, and metastasis
formation (33)
. Therefore, we compared the ability of
S91Adh and S91Nadh cells to form homotypic aggregation. S91Adh cells
showed a high level of homotypic aggregation that was significantly
blocked by anti-VLA-4 mAb. S91Nadh cells showed minimal aggregation,
and this was not significantly affected by anti-VLA-4 mAb (Fig. 6A)
. When the ability of S91 sublines to adhere to endothelial
cells was analyzed, it was found that S91Adh cells effectively bind to
endothelial cells, and this binding was reduced in the presence of
anti-VLA-4 mAb. In contrast, S91Nadh cells showed a low property for
adhesion to endothelial cells (Fig. 6B)
.
|
; Table 2
). S91Nadh cells were also more efficient in traversing uncoated
membranes. Calculation of the percentage of invasion revealed that only
6.5% of S91Adh versus 16% of S91Nadh cells were able to
invade Matrigel-coated membranes. Anti-VLA-4 mAb significantly
increased the ability of S91Adh cells to traverse Matrigel-coated and
uncoated membranes. As a result, 33% of S91Adh cells were able to
invade Matrigel. Anti-VLA-4 mAb had no significant effect on invasive
properties of S91Nadh cells (Fig. 7A
; Table 2
). Some of the
invaded cells detached from the back of the filter into the bottom of
the well. After removal of the inserts, these cells were allowed to
grow for an additional 48 h. The number of S91Nadh cells that
reached the bottoms of the wells was higher than for S91Adh cells (Fig. 7B)
. Therefore, it seems that invasion of the
Matrigel-coated filters depends not only on the ability of tumor cells
to digest Matrigel but also on their motility, particularly their
ability to move and cross the pores in the membrane. Thus, the observed
differences in the Matrigel invasion could be explained, at least in
part, by the higher ability of S91Adh cells to adhere to Matrigel, and
their adherence reduced their motility and ability to traverse the
membrane. Incubation of S91Adh cells with anti-VLA-4 mAb prevented
their adhesion and thus increased their ability to traverse
Matrigel-coated and uncoated membranes.
|
|
. Specifically, reduction in the bands corresponding to gelatinase A
(MMP-2) at Mr 70,000,
gelatinase B (MMP-9) at Mr 93,000,
Matrilysin (MMP-7) at Mr 28,000, and
collagenase 3 (MMP-13) or stromelysins (MMP-3 or MMP- 10) at
Mr 55,000 was found. Taking into
account the results of the zymogram, it is possible to conclude that
low invasiveness of S91Adh cells assessed by Matrigel assay was not
attributable to their low MMP enzymatic activity but probably was a
result of their high adhesiveness and low mobility. In contrast,
S91Nadh cells had lower MMP enzymatic activity but higher mobility that
resulted in their higher invasiveness of the Matrigel-coated membranes.
|
. These results suggest that lower gelanolytic activity in the
supernatant of S91Nadh cells might be attributed to TIMP-1 expression
in these cells.
|
show that S91Adh cells express a very low level of
H-2Kd and a higher level of
H-2Dd molecules. The expression of these
molecules was lower in S91Nadh cells. Similar to other murine
melanomas, S91Adh melanoma does not express SBA and
GSIB4 lectin-binding carbohydrates. However, in
S91Nadh cells, high levels of these carbohydrates were found (Fig. 10A)
. The ligand for GSlB4 lectin has
been shown previously to be the carbohydrate structure
Gal1–3Galß1–4GlcNAc-R (termed the -galactosyl epitope; Ref.
37
). This epitope, which is abundant on mouse normal and
malignant cells, is synthesized in the Golgi apparatus by the enzyme
1,3GT (38)
. It seems that differences in expression of
-galactosyl epitopes in S91Adh and S91Nadh cells might be
attributable to the differences in expression of the
1,3GT gene. Indeed, our RT-PCR analysis showed
that S91Adh cells do not contain message for
1,3GT, whereas in S91Nadh cells selected for
resistance to anoikis in the 1,3GT gene was
up-regulated (Fig. 10B)
.
|
. S91Nadh cells
express SBA-binding carbohydrates, and their expression was further
increased after neuraminidase treatment. These data indicate that
S91Adh cells are heavily sialylated, and sialic acid completely masks
SBA lectin-binding carbohydrates in these cells. In contrast, S91Nadh
cells are less sialylated and express SBA binding carbohydrates
(Fig. 10C)
. |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Selection of S91 melanoma cells for anoikis resistance was associated
with an increased metastatic potential. It is of note that this
increase in metastasis formation was not directly attributable to the
increase in survival of anoikis-resistant S91Nadh cells after i.v.
inoculation. Tumor cells entering into the blood trigger hemostatic
cascade, resulting in a fibrin deposition, platelet aggregation around
tumor cells, and embolization of the small blood vessels that allow
tumor cell adhesion to the endothelium (21)
. All of these
events could prevent anoikis. Indeed, our data showed that
anoikis-sensitive S91Adh and anoikis-resistant S91Nadh melanoma cells
had a similar rate of elimination from the blood, suggesting that
anoikis is not responsible for tumor cell elimination. Tumor cell
survival in the blood is an important step in the metastatic cascade,
but their further ability to extravasate and proliferate in a new
environment is also important for metastasis formation
(21)
. Thus, the increased metastatic potential of S91Nadh
cells could be attributed to some phenotypic changes that emerged in
these cells as a result of selection for anoikis resistance. Selection
of S91 cells for nonadherence resulted in a loss of VLA-4 integrin
expression, decrease in homotypic aggregation, adhesion to endothelial
cells, MMP-2 and MMP-9 activities, increase in Matrigel invasiveness,
TIMP-1 expression, up-regulation of the 1,3GT
gene and -galactosyl epitopes, and reduction in cell membrane
sialylation. Numerous experimental data showed that these phenotypic
changes are very important for the metastatic ability of tumor cells
(17
, 28
, 34, 35, 36
, 39)
.
It is unclear whether the anoikis-resistant cells with altered
phenotypic properties preexisted or were induced during selection for
independence on substrate adhesion. It is difficult to completely prove
or disprove this. However, we were unable to isolate VLA-4-negative and
-galactosyl-positive cells from S91Adh cells using flow cytometric
cell sorting. In addition, RT-PCR analysis of S91Adh cells showed no
message for the 1,3GT gene that could be
detectable if 1,3GT-positive cells preexisted. These results support
the possibility that the genetic and phenotypic changes were induced in
cells selected for resistance to anoikis. It was shown that cell
adhesion triggers the expression of various genes (2, 3, 4)
.
It is possible that in the cells capable of survival without adhesion
to the substrate, another set of genes is activated. Therefore,
numerous phenotypic changes have been found in anoikis-resistant cells.
Some of these changes affected the metastatic potential of
anoikis-resistant S91Nadh cells.
It remains to be determined whether loss of VLA-4 is crucial for loss
of adherence and prevention of anoikis. However, it is likely that the
loss of VLA-4 is probably responsible for the observed decrease in
homotypic aggregation and adhesion to endothelial cells. Consistent
with our findings, it was found that B16a melanoma cells that do not
express VLA-4 also showed higher Matrigel invasiveness than
VLA-4-positive B16 melanoma cells (18)
. Transfection of
B16a cells with the VLA-4 gene reduced their
invasiveness, and blocking VLA-4 by anti-VLA-4 mAb increased
invasiveness of B16 melanoma cells (18)
. We believe that
the increased in vitro invasiveness of the VLA-4-S91Nadh
cells is most likely based on their low adhesion and increased mobility
rather than a result of their increased proteolytic activities. In
fact, poorly invasive VLA-4-positive S91Adh cells have higher
proteolytic activity and produced more MMP-2 and MMP-9 as well as other
MMPs than did highly invasive VLA-4-S91Nadh cells.
Previous analysis of the possible involvement of VLA-4 in metastasis formation gave quite controversial results. Depending on the experimental model, VLA-4 expression decreased or increased metastasis formation (28 , 40 , 41) . Although in our study higher metastatic ability of S91Nadh cells was associated with loss of VLA-4 expression, we found concomitant changes in other phenotypic properties in these cells that might affect their metastatic ability.
In summary, our data indicate that anoikis resistance of S91Nadh melanoma cells did not provide them with an increased survival in the blood. It might be that time of free circulation of tumor cells in the blood as nonadherent cells is too short to trigger anoikis. Tumor cells adhere quickly to the endothelium, which could prevent anoikis. Selection of S91 cells for anoikis was associated with various phenotypic changes. Most of these changes coincide with rather than directly determine the anoikis resistance. However, some of these changes might be responsible for the observed increased metastatic ability of S91Nadh melanoma cells.
|
FOOTNOTES |
|---|
1 This study was supported by NIH Grant CA59903. 
2 Present address: Department of Biochemistry, Sun
Yat-sen University of Medical Science, Guangzhou 510089, China.
3 To whom requests for reprints should be
addressed, at University of Pittsburgh Cancer Institute, BST, W954,
Pittsburgh, PA 15213. Phone: (412) 624-0346; Fax: (412) 624-7736.
4 The abbreviations used are: JNK, c-Jun
NH2-terminal kinase; PolyHema,
polyhydroxyethylmethacrylate; PI, propidium iodide; mAb, monoclonal
antibody; PE, phycoerythrin; SBA, soybean agglutinin; TIMP,
tissue inhibitor of metalloproteinases; RT-PCR, reverse
transcription-PCR; 1,3GT, 1,3-galactosyltransferase; Z-VAD-fmk,
Z-Val-Ala-Asp-fluoromethyl ketone; WGA, wheat germ agglutinin; MMP,
matrix metalloproteinase; VLA, very late activation antigen.
Received 7/20/00. Accepted 12/12/00.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vß3 rescues melanoma cells from apoptosis in three-dimensional dermal collagen.. Proc. Natl. Acad. Sci. USA, 91: 8856-8860, 1994.vß5 to promote tumor cell dissemination in vivo.. J. Clin. Investig., 99: 1390-1398, 1997.[Medline]
4ß1 on melanoma cells can inhibit the invasive stage of metastasis formation.. Cell, 77: 335-347, 1994.[Medline]
4 integrin chain is a ligand for 4ß7 and 4ß1.. J. Exp. Med., 182: 345-355, 1995.1,3-galactosyltransferase gene transfection.. Cancer Res., 55: 4168-4173, 1995.-galactosyl epitopes on nucleated cells.. J. Biol. Chem., 263: 17755-17762, 1988. enhancement of experimental metastasis.. Cancer Res., 54: 3233-3236, 1994.This article has been cited by other articles:
|
|
 |
|
  Y. Liu, P. W. Smith, and D. R. Jones Breast Cancer Metastasis Suppressor 1 Functions as a Corepressor by Enhancing Histone Deacetylase 1-Mediated Deacetylation of RelA/p65 and Promoting Apoptosis Mol. Cell. Biol., December 1, 2006; 26(23): 8683 - 8696. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Rennebeck, M. Martelli, and N. Kyprianou Anoikis and Survival Connections in the Tumor Microenvironment: Is There a Role in Prostate Cancer Metastasis? Cancer Res., December 15, 2005; 65(24): 11230 - 11235. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Arumugam, D. M. Simeone, K. Van Golen, and C. D. Logsdon S100P Promotes Pancreatic Cancer Growth, Survival, and Invasion Clin. Cancer Res., August 1, 2005; 11(15): 5356 - 5364. [Abstract] |
) or under rocking conditions in the presence
(
) or absence (
) of Z-VAD-fmk (100 nM). The numbers
of viable melanoma cells/dish were determined. B,
sensitivity of S91Adh and S91Nadh cells to WGA lectin-induced
apoptosis. S91Adh and S91Nadh cells were cultured alone (
)
and S91Nadh (
) cells were cultured for 3 days under stationary
or rocking conditions, respectively. S91Adh (x) and S91Nadh (
)
cells were also cultured in PolyHema-coated plates.
Bars, SE. B, in
vivo growth of S91Adh and S91Nadh melanoma cells. S91Adh (
