(Received for publication, July 12, 1996, and in revised form, December 30, 1996)
From The interference of tumor necrosis factor- Tumor necrosis factor- The cytotoxic effect of TNF toward tumor cells can be affected by both
intrinsic and acquired cell resistance. However, the current
understanding of the molecular mechanisms critical for tumor resistance
to TNF and for subsequent tumor progression remains limited. Cell
surface expression of TNF receptors is necessary but not sufficient to
induce a biological response, and post-receptor mechanisms are
important in controlling the susceptibility to the cytotoxic action of
TNF. The elucidation of the TNF signaling transduction pathway is
particularly challenging because of the extremely wide variety of TNF
responses. It is established that TNF regulates the transcription of
several genes, many of which are regulated by NF- Recently, ceramide was reported to be an important lipid messenger in
various pathways of TNF action (16, 17). Ceramide can be
generated from sphingomyelin (SM) hydrolysis by two types of early
TNF-responsive sphingomyelinases (SMases), a membrane-associated neutral (N-)SMase and an endosomal acidic (A-)SMase (18, 19). Ceramide
targets may include a membrane-associated ceramide-activated protein kinase (20), a cytosolic ceramide-activated protein phosphatase
(21), the mitogen-activated protein kinase cascade (22), and the
stress-activated protein kinases (23). Recent studies established that
ceramide generated by N-SMase directed the activation of
proline-directed serine/threonine protein kinase(s) and phospholipase
A2 (PLA2), while ceramide generated by A-SMase triggers the activation of NF- Highly
purified (>99%) recombinant TNF- TNF-resistant R-A1 cells were derived from a
TNF-sensitive human breast carcinoma MCF7 cell line after continuous
exposure to increasing doses of recombinant TNF- Cell
viability was determined using the crystal violet staining method as
described previously (25). Absorbance (A), which was
proportional to cell viability, was measured at 540 nm. Cell viability
(%) = 100 × (A1/A0), cell lysis (%) = 1 Indirect immunofluorescence was
performed by incubating of 1 × 106 cells with TNF
receptor antibodies (htr-9 and utr-1) for 1 h on ice in PBS, 1%
FCS. Cells were then washed and stained with 1:50 dilution of
affinity-purified biotinylated goat anti-mouse IgG for 30 min. After
three washes with PBS, 1% FCS, cells were incubated with 50 µl of
streptavidin-phycoerythrin solution for 30 min. After additional
washing with PBS, stained cells were analyzed using an EPICS profile II
Coulter (Coultronic, Margency, France). Fluorescence data were
collected on 5 × 103 viable cells, as determined by
forward light scatter intensity. Background fluorescence was determined
using PBS, 1% FCS instead of the TNF-R monoclonal antibody.
TNF ligand internalization and degradation were estimated, as
described previously by Tsujimoto et al. (30). Cells (1 × 106) were exposed to 125I-TNF (Amersham) at
4 °C for 2.5 h and then washed with ice-cold medium and shifted
to 37 °C by adding prewarmed medium and further incubated at
37 °C. At the times indicated, culture fluids were harvested, and
trichloroacetic acid was added to a final concentration of 10% to
quantitate degradation of internalized TNF by the cells. The soluble
counts were determined after removal of the precipitate by
centrifugation at 1500 × g for 20 min. The cells were
washed once with ice-cold PBS and incubated for 5 min at 4 °C with 2 ml of 50 mM glycine-HCl buffer (pH 3.0) containing 150 mM NaCl. After removal of the glycine buffer, cells were
washed twice and solubilized in 0.1% SDS. 125I
radioactivity found in the glycine buffer and that found in solubilized
cells, respectively, represented surface-bound TNF and internalized
intracellular TNF.
Human breast carcinoma MCF7 cells (15 × 106) were incubated for 90 min in the presence or absence
of TNF. Adherent cells were then trypsinized, and nuclear
extracts were prepared according to the procedure of Dignam
et al. (31). Gel mobility shift assays were performed with a
synthetic double-stranded 31-mer oligonucleotide containing the Total RNA was
extracted from the tumor cell lines according to the method of
Chomczynski and Sacchi (33). RNA (15 µg/lane) were electrophoresed in
a 1.2% agarose gel and transferred to nitrocellulose membrane Hybond-C
(Amersham). The membrane was then hybridized overnight at 42 °C with
the probe labeled with [ For phospholipid labeling, cells were incubated
with RPMI medium containing 5% FCS and 1 µCi/ml
[9,10-3H]palmitic acid (35.9 Ci/mmol, NEN du Pont,
France). After 48 h of incubation, the radioactive medium was
removed and cells were incubated for another 2-4 h in culture medium.
Cells (3 × 106) were resuspended in 1 ml of culture
medium supplemented with 10 mM HEPES and treated with TNF
at various times. Lipids were extracted by the method of Bligh and Dyer
(34) and were separated by thin layer chromatography (TLC) using
chloroform/methanol/water (100:42:6, by volume) followed by a second
step using petroleum ether/diethylether/acetic acid (80:20:1, by
volume) or hexane/diethylether/formic acid (55:45:1, by volume) as
developing solvent systems. Radioactive lipid spots, detected with a
Berthold radiochromatoscan and upon exposure to iodine vapor, were
scraped into scintillation fluid and counted. The positions of ceramide
on TLC plates were determined by comparison with concomitantly run
3H-lipid extracts from MCF7 cells treated with exogenous
bacterial SMase (100 milliunits/1.5 × 106 cells).
Statistical analysis was performed using Student's t test.
Cells (5 × 106) were washed twice in phosphate-buffered saline, and
lipids were extracted by the method of Bligh and Dye (34) and separated
on TLC using chloroform/methanol/water (70:35:5, by volume). The
various spots detected after exposure to iodine vapor were determined
for phosphorus content according to Böttcher et al.
(35).
The SMase assay was performed, as
described previously by Wiegmann et al. (24). Cells were
incubated with or without TNF for various times, and the stimulation
was stopped by placing 2-3 × 106 cell aliquots in a
methanol-dry ice bath. To measure acidic SMase activity, cell pellets
were resuspended in 0.1% Triton X-100 and incubated for 15 min at
4 °C before homogenization. About 100 µg of cellular lysate
protein were incubated for 1 h at 37 °C in a buffer containing
250 mM sodium acetate (pH 5.0), 1 mM EDTA, and
[choline-methyl-14C]SM (54.5 mCi/mmol, NEN
DuPont; 100,000 dpm/assay). To measure neutral SMase activity, cell
pellets were resuspended in 0.1% Triton X-100, 20 mM HEPES
(pH 7.4), 10 mM MgCl2, 2 mM EDTA, 5 mM dithiothreitol, 0.1 mM
Na3VO4, 0.1 mM
Na2MoO4, 10 mM
According to the method of Mutch et al. (37),
cells (2 × 105) were plated in 35-mm dishes
containing 1 ml of culture medium to attach overnight and were then
incubated with 0.5 µCi/ml [3H]arachidonic acid
([5,6,8,9,11,12,14,15-3H]) (209 Ci/mmol, NEN du Pont,
France) at 37 °C for 18 h. Unincorporated [3H]arachidonic acid was removed and the cells were
washed twice with PBS supplemented with 0.1% BSA. Radiolabeled cells
were incubated with medium or TNF. The medium was then removed and
centrifuged at 2000 × g for 5 min to separate cell
debris. 0.1 ml of the supernatant fluids was mixed with 3 ml of
scintillation fluid and counted. The amount of 3H release
was determined in triplicate. The relative release of 3H
was calculated as the percentage of arachidonic acid released in the
supernatant with respect of total [3H]arachidonic acid
content of cells.
In an attempt to examine the mechanism of
TNF-resistance acquisition by tumor cells, we established a
TNF-resistant variant R-A1, derived from MCF7 cells (25). As shown in
Fig. 1A, R-A1 cells were resistant to TNF
compared to the parental MCF7 cells. Flow cytometry analysis (Fig.
1B) indicates that while p55 TNF-R was highly expressed in
parental MCF7 cells (80%), a lower level of p55 expression (30%) was
observed in R-A1. Both cell lines displayed marginal expression of the
p75 receptor (10%). Data of binding experiments (Fig. 1C)
using 125I-radiolabeled TNF show that receptor-bound TNF
was rapidly internalized by TNF-sensitive MCF7 cells. In contrast, very
little TNF binding and no TNF internalization were detected in R-A1
cells. Electrophoretic mobility shift assays for NF-
Based on the
above observations and as the p55 TNF receptor was reported to be
responsible for TNF cytotoxicity signaling in most cellular models, we
attempted to correct TNF signaling by transfecting R-A1 cells with a
wild-type p55 expression vector, pMPSVEH-hup55 TNF-R. Following
screening by fluorescence-activated cell sorting analysis using
anti-TNF-R-p55 monoclonal antibody (htr-9), the stable transfected
cells expressing a high level of cell surface p55 receptor (more than
70%) were selected for further study. Gel shift experiments were first
performed to examine early response to TNF, i.e. NF-
Strong evidence has been
provided about the role of MnSOD and A20 in cell protection against the
cytotoxic action of TNF (12, 38). We therefore investigated the
mRNA expression of these genes in the p55-transfected R-A1 cells to
determine whether they were involved in the resistance to TNF
cytotoxicity observed in these cells. Northern blot analysis showed
that TNF treatment induced a similar increase in MnSOD and A20 mRNA
expression in sensitive MCF7 and in p55-transfected R-A1 cells (Fig.
3). As expected, no increase in the mRNA expression
level of either gene was detectable in control R-A1 cells due to the
absence of TNF signaling. These data suggest that the resistance of
these transfected cells was not directly related to MnSOD and A20 gene
expression and further emphasize TNF signaling efficiency in these
p55-transfected cells.
Ceramide, the product
resulting from the breakdown of sphingomyelin, serves as the second
messenger in the apoptotic signaling pathway (17). Recently, ceramide
was reported to be an essential mediator in TNF-induced cell killing
(39, 40). We therefore investigated the possible interference of the
sphingomyelin/ceramide pathway with the mechanism of TNF resistance in
MCF7-derivative cells. As shown in Fig. 4A, a
concomitant increase in intracellular ceramide (~2.4-fold increase)
was detected in MCF7 at 10-20 min of TNF incubation, preceded by a
rapid SM hydrolysis (>30% decrease in SM content) that reached the
maximum level within 5-10 min after TNF treatment. The measurement of
neutral and acidic SMases indicates that the SM breakdown and ceramide
generation in MCF7 cells was correlated with a significant induction of
these two SMase activities (20% and 40% of control for neutral and
acidic SMases, respectively) after 5-15 min of TNF incubation (Fig.
4B). In contrast, TNF did not induce neither ceramide
generation nor SM hydrolysis in R-A1 and p55-transfected cells (clones
1001 and 3024) (Fig. 4A). This is consistent with the
failure of TNF to stimulate both SMase activities in these cells (Fig.
4B). However, the defect of the SMase activation was not
related to a decrease in basal SMase activities in the resistant cells,
since the basal SMase activities were not significantly different in
the three cell lines tested (e.g. N-SMase activities were
231, 385, and 258 pmol/h/mg proteins for MCF7, R-A1, and clone 1001 cells, respectively).
In order to compare the basal level of SMase substrate in
TNF-sensitive and resistant cells, the analysis of cellular SM content was performed. As shown in Table I, the percentage of SM
(as compared to total phospholipids) was significantly higher in
parental MCF7 cells (28-30% of increase) than in R-A1 and clone 1001 cells. Moreover, the basal SM content was 2-fold higher in MCF7 than in
the two resistant counterparts (20.9 nmol for MCF7 versus
10.7 nmol and 12.4 nmol for R-A1 and clone 1001 cells, respectively). This result suggests that the different responses induced by TNF in
these cells may be dependent on the cellular SM content.
Cellular SM content in TNF-sensitive and -resistant MCF7 cells
INSERM Contrat Jeune Formation 94-11 "Cytokines et Immunité Antitumorale,
CNRS Unité de Recherche
associée 1283,
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
Acknowledgments
REFERENCES
(TNF) signaling processes with the acquisition of tumor resistance to
TNF was investigated using the TNF-sensitive human breast carcinoma
MCF7 cell line and its established TNF-resistant variant (R-A1). The resistance of R-A1 cells to TNF correlated with a low level of p55 TNF
receptor expression and an absence of TNF signaling through TNF
receptors. Stable transfection of wild-type p55 receptor in R-A1
resulted in enhancement of p55 expression and in partial restoration of
TNF signaling, including nuclear factor-
B (NF-
B) activation.
However, the transfected cells remained resistant to TNF-induced
apoptosis. Northern blot analysis revealed a comparable induction of
manganous superoxide dismutase and A20 mRNA expression in
p55-transfected cells and in sensitive MCF7 cells, making it unlikely
that these genes are involved in the resistance to TNF-mediated cytotoxicity. While TNF significantly stimulated both neutral and
acidic sphingomyelinase (SMase) activities with concomitant sphingomyelin (SM) hydrolysis and ceramide generation in MCF7, it
failed to trigger these events in TNF-resistant p55-transfected cells.
In addition, the basal SM content was significantly higher in sensitive
MCF7 as compared to the resistant counterparts. Furthermore, the
TNF-resistant cells tested could be induced to undergo cell death after
exposure to exogenous SMase or cell-permeable C6-ceramide. This study also shows that TNF failed to induce arachidonic acid release in p55-transfected resistant cells, suggesting that an alteration of phospholipase A2 activation may be associated
with MCF7 cell resistance to TNF. Our findings strongly suggest a role of ceramide in the mechanism of cell resistance to TNF-mediated cell
death and may be relevant in elucidating the biochemical nature
of intracellular messengers leading to such resistance.
(TNF),1
originally described for its antitumor activity, is now recognized as
one of the most pleiotropic cytokine to act as a host defense factor in
immunological and inflammatory responses (1-3). Like other biological
activities of TNF, its antitumor activity is exerted through binding to
two distinct but structurally related cell surface receptors, p55 (TNF-R1) and p75 (TNF-R2). Gene knockout experiments and the use of
receptor-specific agonistic antibodies confirmed that the two TNF
receptors generate nonoverlapping signals (4-7). Although there is now
considerable evidence that p55 is the TNF receptor that directly
mediates cytotoxicity in a wide variety of cell types (8), recent
studies have indicated that p75-mediated signals may cooperate with p55
to facilite cell death in some cell types (9). Structure function
analysis of p55 TNF-R signaling demonstrated that an 80-amino acid
region within the cytoplasmic domain is required for initiation of
apoptosis and NF-
B activation (10).
B (11). It is also
clear that the activation of this transcription factor is a pivotal and
integral event for the transfer of the TNF signal to the nucleus.
Several mechanisms have been reported to contribute to cellular
resistance to TNF-induced cell killing, including the constitutive
expression of several protective proteins in resistant tumor cells,
such as MnSOD, endogenous TNF, major heat shock protein hsp70, A20 zinc
finger protein (12-15). However, these proteins confer only partial
protection against TNF cytotoxicity, suggesting that additional
resistance mechanisms exist.
B, suggesting that the two SMases control important yet dissociable and nonoverlapping pathways of TNF
receptor signaling (24). Because of the potential role of ceramide in
mediating the cytotoxic effect of TNF, we examined its possible
involvement in cell resistance to TNF. Our findings indicate that
stable transfection of p55 TNF receptor restores TNF signaling
including NF-
B activation in TNF-resistant MCF7 variant R-A1, but
does not restore the susceptibility of these cells to TNF cytotoxicity.
The data presented in this study further support the notion that the
apoptotic effect of TNF is probably dissociated from NF-
B
activation, and suggest that an alteration of sphingomyelinase
activation and subsequent ceramide generation may represent an
important additional mechanism by which human tumor cells may escape
TNF-mediated apoptosis.
Cytokines, Monoclonal Antibodies, and Reagents
(TNF, specific activity 6.63 × 106 units/mg protein) was kindly provided by A. G. Knoll
(BASF, Ludwigshafen, Germany). Monoclonal antibodies htr-9 and utr-1
directed against TNF receptors p55 and p75, respectively, were
generously provided by Dr. M. Brokhaus (Hoffman-La Roche, Ltd., Basel,
Switzerland). Sphingomyelinase (Staphylococcus aureus) and
phospholipase D (cabbage) were obtained from Sigma.
N-Hexanoyl-D-sphingosine
(C6-ceramide) and C6-dihydroceramide were
purchased from Matreya (Pleasant Gap, PA).
(25). The wild-type human p55 TNF-R cDNA cloned in a mammalian expression vector
pMPSVEH (26) was used to transfect R-A1 cells by the calcium phosphate precipitation method (27). Briefly, 1000 cells/10-cm tissue culture
plate were plated. After 10-14 days of selection in growth medium
containing 200 µg/ml G418 (Sigma), 4-5 resistant
colonies were isolated from each plate and examined for human p55 TNF-R expression by fluorescence-activated cell sorting. The positive clones
were subsequently maintained in medium with 100 µg/ml G418 for more
than 2 months. The sensitivity of clones to TNF was tested every 2 weeks during culture. All cell lines were routinely cultured in
RPMI 1640 medium containing 5% FCS, 1% penicillin-streptomycin, 1%
L-glutamine at 37 °C in a humidified atmosphere with 5%
CO2.
cell viability (%), where A1 and
A0 were the absorbance obtained from treated and
untreated cells, respectively. The mean value of quadruplicate was used
for analysis. Quantitative DNA fragmentation was determined as
described previously (28). TNF- or ceramide-treated and untreated cells
(1 × 106) were pelleted and washed in PBS. Cells were
then resuspended in lysis buffer (0.5% v/v Triton X-100, 20 mM EDTA and 5 mM Tris-HCl, pH 8.0) and
centrifuged at 27,000 × g for 20 min to separate the chromatin pellet from fragmented DNA. Both the pellet (resuspended in 1 mM EDTA and 10 mM Tris-HCl, pH 8.0) and
supernatant were assayed to determine DNA by the spectrofluorometric
DAPI procedure (29).
B
sequences of the human immunodeficiency virus long terminal repeat,
5
-end-labeled with [
-32P]ATP using the T4 kinase
(32).
-32P]dCTP using a Megaprime
DNA labeling system (Amersham). The hybridized membrane was washed and
exposed to Hyperfilm-MP (Amersham). The blot was stripped by boiling in
0.1% SDS and probed again with a
-actin probe as a control for
equal RNA loading.
-glycerophosphate, 750 µM ATP, 1 mM
phenylmethylsulfonyl fluoride, 10 µM leupeptin, and 10 µM pepstatin. After incubation for 5 min at 4 °C,
cells were homogenized and 100 µg of cellular lysate protein were
incubated for 2 h at 37 °C in 20 mM HEPES (pH 7.4),
1 mM MgCl2, and
[choline-methyl-14C]SM (100,000 dpm/assay).
Radioactive phosphocholine produced from
[choline-methyl-14C]SM was then extracted with
chloroform/methanol (2:1) and quantitated by scintillation counting
(36).
Lack of Signaling through TNF Receptors in the TNF-resistant
Variant of MCF7
B activation
were performed to further evaluate the response of these cells to TNF.
Treatment with TNF induced NF-
B translocation in MCF7 but had no
effect in R-A1 cells (Fig. 1D). The data of binding and
internalization experiments are consistent with the failure of TNF to
induce NF-
B activation in R-A1. The hypothesis that resistance to
TNF exhibited by R-A1 cells may be due to altered signaling through TNF
receptors was examined next.
Fig. 1.
Lack of TNF signaling in TNF-resistant
MCF7-derived R-A1 cells. A, effect of TNF on the viability
of parental MCF7 () and R-A1 (
) cells. Cells (7.5 × 103 cells/well) were incubated for 72 h with the
indicated doses of recombinant TNF-
. Cell viability was measured
using the crystal violet assay as described under "Experimental
Procedures." Data presented are the means ± S.D. of
quadruplicate. B, flow cytometric analysis of TNF receptor
expression on MCF7 and R-A1. Cells (1 × 106) were
detached from the culture and stained with the htr-9 and utr-1
monoclonal antibodies directed against the p55 and p75 TNF receptors,
respectively, as described under "Experimental Procedures." C, internalization and degradation of 125I-TNF
bound to MCF7 and R-A1 cells. 125I-TNF (50 pM)
was allowed to bind to cells for 2 h at 4 °C. Thereafter, cells
were washed free of TNF and shifted to 37 °C for indicated incubation times. Surface-bound (
), internalized (
), and degraded (
) 125I-TNF were determined, as described under
"Experimental Procedures." Data presented are representative of one
of three independent experiments. D, effect of TNF on
NF-
B activation in MCF7 and R-A1 cells. Cells (10 × 106) were incubated for 90 min in the presence or absence
of TNF (50 ng/ml). Nuclear proteins (15 µg) extracted from untreated cells (lane 1) or TNF-treated cells (lane 2) were
submitted to electrophoretic mobility shift assay as described under
"Experimental Procedures." TNF-treated extracts were competed with
100-fold excess of unlabeled oligonucleotide (lane 3).
[View Larger Version of this Image (32K GIF file)]
B Activation but Not TNF-induced Cell Lysis
B
activation, in the transfected cells. As shown in Fig.
2A, exposure of three representative R-A1 p55-transfected clones (clones 1001, 2101, and 3024) to TNF (50 ng/ml)
resulted in the activation of NF-
B, indicating that the TNF
signaling pathway leading to NF-
B activation was functional. Interestingly, despite NF-
B activation, these p55-transfected clones
remained resistant to TNF, even when a high concentration of TNF (200 ng/ml) was used (Fig. 2B). These data clearly indicate that
wild-type p55 receptor expression and the resulting NF-
B activation
in R-A1 cells are not sufficient to trigger TNF cytotoxic activity.
Fig. 2.
A, effect of TNF on NF-B activation
in vector-transfected control R-A1 cells and pMPSVEH-hup55-transfected
R-A1 clones (clones 1001, 2101, and 3024). Cells (10 × 106) were incubated for 90 min in the presence or absence
of TNF (50 ng/ml). Nuclear proteins (15 µg) extracted from untreated cells (
) or TNF-treated cells (+) were submitted to electrophoretic mobility shift assay as described under "Experimental Procedures." NF-
B activation by TNF was undetectable in control
vector-transfected R-A1 cells. B, effect of TNF on the
viability of parental MCF7 and transfected R-A1 cells. Cells (7.5 × 103/well) were incubated for 72 h with the
indicated doses of recombinant TNF-
. Cell viability was measured
using the crystal violet assay as described under "Experimental
Procedures." Data presented are the means of quadruplicate.
[View Larger Version of this Image (48K GIF file)]
Fig. 3.
Northern blot analysis of MnSOD and A20 gene
expression following TNF treatment. Cells (10 × 106) were incubated with medium () or 50 ng/ml TNF (+)
for 6 h at 37 °C. Total RNA (15 µg/lane) were electrophoresed
in a 1.2% agarose gel and hybridized with 32P-labeled
MnSOD and A20 specific cDNA probes.
-Actin probe was used to
confirm comparable loading of RNA sample.
[View Larger Version of this Image (59K GIF file)]
Fig. 4.
Induction of sphingomyelin metabolism in the
parental TNF-sensitive MCF7 and in its resistant counterparts.
A, ceramide and SM levels were evaluated in cells (3 × 106) prelabeled with [9,10-3H]palmitic acid
for 48 h. Cells were then treated with 50 ng/ml TNF for the
indicated time intervals. Aliquots were then collected to prepare lipid
extracts. Labeled ceramide and SM were resolved by TLC as described
under "Experimental Procedures." Results are expressed as a
percentage of untreated controls. Control ceramide counts were
11,076 ± 230 and 10,190 ± 540 cpm for MCF7 and R-A1, respectively. For MCF7, ceramide data are the mean ± S.E. of
three independent experiments. *, p < 0.045; **,
p < 0.002, relative to time zero. SM data are from one
representative of three independent experiments. For R-A1- and
p55-transfected cells (clones 1001, and 3024), data are from one
representative of three independent experiments. B, for
neutral (N-) and acidic (A-) SMase activities, aliquots were collected
after TNF treatment (50 ng/ml) and enzyme assays were performed as
described under "Experimental Procedures." Data are from one
representative of two independent experiments, and they are expressed
as a percentage of controls. The results for A-SMase are the mean ± S.D. of duplicate measurements. The data (not shown) for R-A1 and
clone 3024 are similar to that of clone 1001.
[View Larger Version of this Image (28K GIF file)]
Cells
SM
SM (compared
to PL)
nmol IP/mg protein
%
MCF7
20.9 ± 0.3
14.96
± 0.20
R-A1
10.7 ± 1.3
11.46 ± 1.60a
Clone 1001
12.4 ± 0.3
11.66 ± 0.50b
a
Decreased vs. % of MCF7 SM
(p < 0.02).
b
Decreased vs. % of MCF7 SM (p < 0.001).
To determine whether the activation of SMase and the
production of ceramide could overcome the resistance of transfected
R-A1 cells, we examined the susceptibility of these cells to exogenous bacterial-derived SMase and synthetic cell-permeable ceramide. The
addition of SMase (Fig. 5A) or
C6-ceramide (Fig. 5B) was able to induce the
killing of p55-transfected cells as well as that of control R-A1 and
parental MCF7 cells in a dose-dependent manner. The
cytolytic effect of SMase and C6-ceramide on these cells
was specific, since the addition of phospholipase D (data not shown) or
C6-dihydroceramide (Fig. 5B) at equivalent
concentrations failed to induce cell killing. Furthermore, DNA
fragmentation analysis indicates that C6-ceramide killed
both TNF-sensitive and TNF-resistant MCF7 cells through an apoptotic
pathway (Fig. 6). After 24 or 48 h of treatment of
TNF-sensitive MCF7 cells, more apoptotic cells were observed upon
C6-ceramide treatment than that upon TNF treatment,
indicating that ceramide triggers apoptosis of these cells in a more
direct manner than TNF.
TNF Induces the Release of Arachidonic Acid in Sensitive MCF7 but Not in p55-transfected R-A1 Cells
TNF has been described to be
capable of activating PLA2 and inducing the release of
arachidonic acid (AA) from membrane phospholipids in several sensitive
target cells (41, 42). In addition, ceramide was recently reported to
be capable of activating the PLA2/AA pathway (24). Initial
experiments using dexamethasone, an inhibitor of PLA2,
indicated that the addition of this component at a subtoxic concentration (100 ng/ml) efficiently inhibited (4-fold) the killing of
parental MCF7 cells by TNF (Fig. 7A).
Therefore, the involvement of PLA2 in TNF signaling in MCF7
and transfected R-A1 cells was assessed by measuring the release of AA.
As shown in Fig. 7B, TNF induced a significant release of
3H-labeled AA metabolites (165% of control) in sensitive
MCF7 cells after 18 h of incubation. No stimulation of AA release
was observed during short term (0-6 h) incubations (data not shown).
In contrast, the increase in the release of 3H-labeled AA
in R-A1- and p55-transfected clones was not detected at any of times
tested (Fig. 7B), suggesting that the resistance of these
cells to TNF may also be related to altered AA release.
In contrast to the rapid progress that has been made in defining
gene products capable of regulating TNF-induced cell death, the
knowledge of the molecular components involved in cell resistance to
TNF remains limited. We attempted to delineate the functional role of
some second messengers in the acquisition of tumor resistance to TNF by
comparing a TNF-sensitive human breast cancer cell line MCF7 with its
R-A1 variant selected for resistance to TNF. This TNF-resistant variant
derived from MCF7 was found to be susceptible to anti-Fas-induced cell
lysis as well as to the chemotherapeutic drug adriamycin (data not
shown), compelling evidence that the intracellular cell death pathway
is functional in these cells. As compared to the parental MCF7 cells,
the resistance of R-A1 cells to TNF correlated with a low level of p55
TNF receptor expression and an absence of TNF signaling through TNF-Rs.
Although functional wild-type p55 receptor expression was reestablished
in R-A1 cells by gene transfer as well as subsequent NF-B activation
in response to TNF, these cells remained totally resistant to the
cytotoxic action of TNF. It should be noted that transfection of rodent cells using the same vector efficiently triggered the cytotoxic effect
of TNF (26). Our observations suggest that NF-
B is not sufficient to
induce cell killing, and confirm that the activation of this nuclear
factor and apoptosis are coincidental but that these two activities are
separable. This is also in agreement with the report of Dbaibo et
al. (43), indicating that the growth inhibitory effect of TNF is
dissociated from the activation of NF-
B in Jurkat lymphoblastic
leukemia cells.
Cell resistance to the cytotoxic action of TNF is thought to be an active process requiring the synthesis of TNF-inducible proteins, since this resistance can be overcome by protein synthesis inhibitors in some experimental systems (44). Overexpression of several TNF-inducible early response genes, such as MnSOD and A20, has been reported to protect cells against TNF cytotoxicity (12, 15). MnSOD acts as a superoxide radical scavenger, and its presence correlates with cellular protection against TNF cytotoxicity. A20 zinc finger protein is the product of a cytokine-induced primary response gene, and its overexpression inhibits TNF-induced activation of PLA2 and TNF-mediated apoptosis (38). Although the involvement of MnSOD and A20 in cellular protection has been established, our data indicate that both genes can be induced by TNF in p55-transfected resistant cells at similar level as compared to that in TNF-sensitive MCF7 cells, suggesting that the resistance of these cells is not related to overexpression of these two genes.
Ceramide has emerged as a potent second messenger in TNF signaling, and
a substantial amount of evidence has been accumulated in favor of
ceramide functioning as a selective mediator of the cytotoxic/cytostatic effect of TNF (16, 17). Ceramide generated by the
activation of sphingomyelinase has been reported to mediate TNF-induced
apoptosis in the human monocyte-like U937, human leukemic HL-60, and
murine fibrosarcoma cell lines (39, 40). This study shows that TNF can
activate both neutral and acidic SMases in human breast tumor MCF7
cells. A defect in the activation of these enzymes is apparently
sufficient to confer resistance to TNF in R-A1 cells. Indeed, when this
defect was overcome by adding exogenous SMase or ceramide, the
susceptibility of R-A1 cells to apoptosis was restored. This suggests
that the stimulation of ceramide-activated enzymes may constitute an
important step in the regulation of programmed cell death. It is worthy
to note that in our study N-SMase was more rapidly activated than
A-SMase, confirming that the activation requirements of A-SMase differ
from that of N-SMase. This is also in agreement with the report by
Wiegmann et al. (24), which suggests that N- and A-SMase
activations may be triggered by distinct pathways. When specific
inhibitors of two SMases become available, the determination of the
nature of SMase involved in TNF-induced apoptotic cell death would be
of major interest. It is tempting to speculate that the failure of TNF
to induce DNA fragmentation and apoptosis in resistant cells may be
related to structural membrane organization of SM, which may constitute a limiting step for the generation of this second messenger. Our data
also demonstrate that the basal level of total SM content was higher in
sensitive parental MCF7 cells, as compared to the resistant
counterparts (R-A1 and clone 1001 cells). On the other hand, both
sensitive and resistant cells showed a similar basal level of SMase
activity. Thus, the lower level of basal SM content and the absence of
SMases activation in resistant cells may represent a double blockage
for ceramide generation. These results confirm the hypothesis from our
previous report indicating that the failure of TNF to induce either SM
hydrolysis or apoptosis in resistant myeloid leukemia KG1a cells was
correlated with the SM pool used for TNF signaling, which is
predominantly located in the inner leaflet of the plasma membrane (45).
One could speculate that the lower TNF-hydrolyzable SM pool in
p55-transfected R-A1 cells and the absence of SMase activation would
explain the failure of TNF signaling to induce SM hydrolysis and
ceramide generation in these cells. Although the sphingomyelin/ceramide
pathway was reported to be capable of signaling NF-B translocation
in HL-60 cells (46, 24), we showed here that TNF induced NF-
B
translocation in p55-transfected resistant cells in the absence of
ceramide generation. This is in agreement with other reports
demonstrating that exogenous addition of a short chain ceramide to
Jurkat cells or the inhibition of ceramide pathway had no effect on
NF-
B activation (47, 48), and further confirming that NF-
B
activation by TNF can be independent of endogenous cellular ceramide
generation.
Alternatively, non-induction of apoptosis in the p55-transfected cells could be due to an abnormal expression of other genes such as the proto-oncogene bcl2, involved in the regulation of apoptosis (49, 50). However, the involvement of bcl2 in the resistance of R-A1 cells can be excluded, since the parental TNF-sensitive MCF7 cells displayed a higher bcl2 expression level than the resistant R-A1 cells (data not shown), suggesting that there is no correlation between bcl2 expression and the magnitude of TNF-induced apoptosis in these cells. In addition, we and others have reported that bcl2 acts downstream of ceramide preventing ceramide-induced cell death but not ceramide accumulation in at least two models of chemotherapy-induced cell death (51).2 Our demonstration that TNF failed to generate ceramide in p55-transfected R-A1 clones overrides the implication of bcl2 in this phenomenon.
A complex pattern of integrated signals may be generated in response to an elevation of cellular ceramide content. Indeed, recent studies support the view that ceramide produced by N-SMase triggers the mitogen-activated protein kinase cascade via ceramide-activated protein kinase, which presumably results in the activation of PLA2 (24, 52, 53). This is consistent with several lines of evidence suggesting the involvement of PLA2 in the cytotoxic pathway of TNF (37, 41, 46, 54). In addition, the AA generated by the activation of PLA2 was reported to activate sphingomyelin hydrolysis in HL-60 cells (55). Data obtained in our studies indicate that the release of arachidonic acid is altered in p55-transfected TNF-resistant cells. This might be a consequence of the defect in ceramide generation. Although ceramide generation occurred rapidly (10-20 min) following TNF treatment in the sensitive cells, the activation of PLA2 and the AA release induced by TNF could not be detected in the first hours (0-6 h) after TNF treatment. The AA release probably does not precede TNF-stimulated SM hydrolysis in our system, but is involved as a later biochemical response to the action of TNF. This is consistent with the report of Wiegmann et al. (24), suggesting that PLA2 activation occurs as a later event in the TNF signaling pathway. Whether ceramide and AA function independently or in coordination to transduce TNF cytotoxic effect requires further investigation. Taken together, our data suggest that a selective defect in TNF signaling associated with an alteration in sphingomyelinase activation can, at least in part, confer resistance to TNF-induced cell death.
An insight into signal transduction by p55 TNF-R1 has resulted from the
identification of the TNF-R1-associated protein TRADD, which interacts
with the death domain of p55 and signals both cell death and NF-B
activation (56). Following TNF treatment, the association of TRADD and
TNF-R1 occurs rapidly in U937 cells (57). In addition, TRADD interacts
with TRAF2 (58) and FADD/MORT1 (59, 60) leading, respectively, to
NF-
B activation and apoptosis induction in the overexpression
systems. The importance of the signaling complex assembly is also
demonstrated by the fact that dominant-negative derivative of
FADD/MORT1 abrogated CD95(Fas)-induced apoptosis and ceramide
generation in a B lymphoma cell line (61). One can suggest that in the
case of TNF, ceramide generation may also be a downstream event,
e.g. post-TRADD and/or post-FADD/MORT1 activation. We have
obtained data indicating a comparable TRADD protein expression level in
parental MCF7 as well as in resistant R-A1 and clone 1001 cells, and
that transient TRADD-transfection induced apoptosis and NF-
B
activation in all these cells. However, no effect of TRADD
overexpression on ceramide accumulation could be detected (data not
shown). It seems unlikely that TRADD alone triggers cell death
signaling through ceramide pathway. Whether TRADD/FADD complex
formation occurs and interacts with the SM/ceramide pathway under
physiological conditions remains to be determined. It would be of major
interest to decipher the possible cross-talk between diverse TNF-R
associated signaling molecules, such as FAN protein (62), and the
sphingolipid messengers implicated in the TNF cytotoxic signaling
pathway. Understanding of the molecular and biochemical mechanisms of
tumor cell resistance to the cytocidal effect of TNF may ultimately
provide new approaches to enhance the therapeutic efficacy of TNF
against human malignancies.
We thank Drs. X. J. Ma (Wistar Institute), A. Harel-Bellan (CNRS URA 1156, Villejuif), G. Laurent and J. P. Jaffrézou (INSERM CJF 9503, Toulouse) for critical comments, and L. Saint-Ange (Institut Gustave-Roussy, Villejuif) for editing the manuscript. We are also grateful to Dr. D. Wallach (Rehovot, Israel) for providing pMPSVEH-hup55 TNF-R constructs.