Deficient Tumor Necrosis Factor-related Apoptosis-inducing Ligand (TRAIL) Death Receptor Transport to the Cell Surface in Human Colon Cancer Cells Selected for Resistance to TRAIL-induced Apoptosis*

Many tumor cell types are sensitive to tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-induced apoptosis. Incubation of TRAIL-sensitive cells with TRAIL invariably leads to resistant survivors even when high doses of TRAIL are used. Because the emergence of resistance to apoptosis is a major concern in successful treatment of cancer, and TRAIL survivors may contribute to therapeutic failure, we investigated potential resistance mechanisms. We selected TRAIL-resistant SW480 human colon adenocarcinoma cells by repeatedly treating them with high and/or low doses of TRAIL. The resulting TRAIL-resistant clones were not cross-resistant to Fas or paclitaxel. Expression of modulators of apoptosis was not changed in the resistant cells, including TRAIL receptors, cFLIP, Bax, Bid, or IAP proteins. Surprisingly, we found that DISC formation was deficient in multiple selected TRAIL-resistant clones. DR4 was not recruited to the DISC upon TRAIL treatment, and caspase-8 was not activated at the DISC. Although total cellular DR4 mRNA and protein were virtually identical in TRAIL-sensitive parental and TRAIL-resistant clones, DR4 protein expression on the cell surface was essentially undetectable in the TRAIL-resistant clones. Moreover, exogenous DR4 and KILLER/DR5 were not properly transported to the cell surface in the TRAIL-resistant cells. Interestingly, TRAIL-resistant cells were resensitized to TRAIL by tunicamycin pretreatment, which increased cell surface expression of DR4 and KILLER/DR5. Our data suggest that tumor cells may become resistant to TRAIL through regulation of the death receptor cell surface transport and that resistance to TRAIL may be overcome by the glycosylation inhibitor/endoplasmic reticulum stress-inducing agent tunicamycin.

II transmembrane protein, is highly homologous to other TNFrelated proteins, such as Fas ligand (1). TRAIL, unlike Fas ligand, appears to preferentially induce apoptosis in tumor cells over normal cells (2). TRAIL binds to five distinct TRAIL receptors including death receptor 4 (TRAIL-R1) (3), KILLER/ DR5 (TRAIL-R2, TRICK2) (4 -6), DcR1 (TRID, TRAIL-R3) (7), DcR2 (TRUNDD OR TRAIL-R4) (8), and osteoprotegerin (9). These receptors have been classified into two groups, deathinducing receptors (TRAIL-R1 and -R2) and death-inhibitory receptors (TRAIL-R3, TRAIL-R4, and osteoprotegerin). Both TRAIL-R1 and TRAIL-R2 contain a C-terminal death domain that signals downstream caspase activation to mediate TRAILinduced apoptotic cell death in a variety of tumor cell types. In contrast to these death-inducing receptors, TRAIL-R3, which shares homology with DR4/KILLER/DR5, is devoid of a cytoplasmic domain and exists as a glycophospholipid-anchored protein on the cell surface. TRAIL-R4 has a cytoplasmic domain containing a truncated death domain that cannot transmit a death signal but can weakly activate NF-B, which may protect cells from TRAIL-mediated apoptosis (10). It has been suggested that these two decoy receptors can protect cells from TRAIL-induced apoptosis by competing with the death-inducing TRAIL-Rs for TRAIL binding. Since TRAIL-R3 mRNA was preferentially found in normal cells but not in transformed cells, it is thought that TRAIL-R3 might be responsible for the cellular resistance of some normal cells to TRAIL-mediated cytotoxicity (11)(12)(13). These findings suggest a complex regulation of cellular susceptibility to TRAIL-mediated apoptosis at the level of receptor expression.
Apoptosis can be initiated by two distinct pathways: one is the "intrinsic pathway" mediated by the mitochondria, and the other is the "extrinsic pathway" mediated by death receptors. Activation of those two pathways ultimately results in cleavage of caspase-3 and induction of apoptosis. Cross-talk between the extrinsic and intrinsic pathways has been observed mainly as an amplification loop at the level of execution of each cascade. Depending on the relative contribution of mitochondria in death receptor-mediated apoptosis, tumor cells can be classified as type I or type II cells (14). Apoptosis in type I cells can be induced by TRAIL or Fas without early involvement of the mitochondria pathway, and this death is not blocked by Bcl-2, Bcl-XL, or caspase-9 inhibitors (15). In type II cells, the amount of DISCactivated caspase-8 is not sufficient, and the mitochondria act as "amplifiers" of the apoptotic signal. Type II cell death is sensitive to inhibition by Bcl-2, Bcl-XL, or caspase-9 inhibitors.
TRAIL is a promising anti-cancer agent because it can preferentially kill tumor cells. Sensitivity to TRAIL-induced apoptosis is a key factor influencing the efficacy of TRAIL treat-ment. However, the basis for the sensitivity and resistance of cells to TRAIL-induced apoptosis is not fully understood. In addition, TRAIL contributes to immune surveillance against tumor metastasis, which implies that tumors may gain more metastatic potential after acquiring resistance to TRAIL-induced apoptosis (16,17).
Thus far, most studies of TRAIL resistance have focused on natural resistance, such as resistance of normal cells to TRAILinduced apoptosis (18). Mechanisms of natural or intrinsic resistance may be different from those of acquired or adapted resistance. In the latter, tumor cells may survive after prolonged treatment with TRAIL in cancer therapy or escape from TRAIL-mediated immune surveillance in vivo. Tumor cells can acquire resistance to apoptosis through interference with either intrinsic or extrinsic apoptotic signaling pathways. Mutation of the Bcl-2 family member Bax can confer resistance to TRAIL induced apoptosis in HCT116 cells (19). It also has been reported that Jurkat cells acquire resistance to Fas-mediated apoptosis because their mitochondria lose the ability to release intermembrane proteins in response to Bid or Bax (20). Both HCT116 and Jurkat cells behave as Type II cells in which the mitochondrial pathway appears to play a major role in death receptor-mediated apoptosis. Type II cells may develop resistance by modulation of the intrinsic pathway instead of the extrinsic pathway. Resistance in these cells may be overcome in some cases by combined application of chemotherapeutic drugs (19), which might bypass the mitochondrial pathway (21). Elucidation of acquired TRAIL resistance may lead to new therapeutic approaches to overcome resistance in cancer therapy with TRAIL. In addition, it may improve our understanding of mechanisms underlying how tumor cells escape from surveillance by the immune system and thereby provide novel strategies to prevent tumor development.
In the present study, we investigated how tumor cells acquire resistance to TRAIL-induced apoptosis. TRAIL-resistant tumor cells were selected by eliminating apoptotic cells resulting from treatment with TRAIL. TRAIL-resistant clones were found not to be cross-resistant to Fas or paclitaxel-induced apoptosis. DISC formation was abrogated in TRAIL-resistant clones, apparently due to a defect in localization of DR4 in the plasma membrane. Strikingly, we found that pretreatment with the glycosylation inhibitor tunicamycin specifically and efficiently restored the susceptibility of resistant clones to TRAIL-induced apoptosis. Our studies provide a new mechanism through which tumor cells may acquire resistance to TRAIL-induced apoptosis. This mechanism suggests that pharmacological manipulation of cell surface expression of death receptors may be crucial to overcome resistance to TRAILinduced apoptosis or maintain tumor cell sensitivity in vivo.

MATERIALS AND METHODS
Cell Culture and Reagents-The human colon cancer cell line SW480 was obtained from the American Type Culture Collection (Manassas, VA). SW480 cells were cultured in Dulbecco's modified Eagle's medium. Cell culture media were supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, penicillin (100 units/ml) and streptomycin (100 units/ml). Human recombinant human TRAIL ligand, His 6 antibody, and anti-DR4 for immunofluorescence staining were obtained from R&D Systems (Minneapolis, MN). Anti-human poly(ADP-ribose) polymerase antibody was obtained from Roche Applied Science. Anti-DR4, anti-KILLER/DR5, anti-DcR1, and anti-DcR2 antibody used for flow cytometric detection and functional application (block TRAIL receptors-mediated killing) were obtained from Alexis Corp. (Lausen, Switzerland). Anti-Bax and anti-caspase-3 antibodies were obtained from BD Biosciences. Anti-caspase-8 and anti-FLIP antibodies were obtained from Cell Signaling Technology (San Diego, CA). Anti-KILLER/DR5, anti-Bcl-2, anti-Bcl-XL and anti-caspase-9 antibodies were purchased from IMGENEX (San Diego, CA). The mitochondriaselective carbocyanine dye 3,3Ј-dihexyloxacarbocyanine iodide was purchased from Molecular Probes, Inc. (Eugene, OR). Cycloheximide and tunicamycin were obtained from Calbiochem, The mitogen-activated protein kinase inhibitor U0126 was obtained from Cell Signaling Technology (San Diego, CA).
Generation of TRAIL-resistant SW480 Cells-Selection of TRAILresistant cells was performed as described previously (22). In brief, 1 ϫ 10 6 SW480 cells seeded in 100-mm dishes were treated with 200 ng/ml TRAIL for 24 h, resulting in ϳ99% cell death. The apoptotic cells (cells floating in the medium) were then removed. Surviving cells growing in medium containing 10 ng/ml TRAIL up to 60 -70% confluence were then treated with 200 ng/ml TRAIL for 24 h and allowed to regrow to 60 -70% confluence again in 10 ng/ml TRAIL-containing medium. After five cycles, SW480 pooled clones resistant to TRAIL were obtained. Individual resistant clones were isolated using sterile cloning disks (Scienceware). The sensitivity of individual clones to TRAIL-induced apoptosis was subsequently examined by PI staining and flow cytometry.
Colony Formation Assays-SW480 cells and TRAIL-resistant cells (5000/well) were plated in 12-well dishes and treated with TRAIL, Fas, or paclitaxel for 3 days. The cell culture medium was subsequently changed every 3 days, and after 2 weeks the resulting colonies were stained with Coomassie Blue (23).
PI Staining and Active Caspase-3 Assay-Cells were harvested at the indicated time periods (see figure legends) and prepared for detection of active caspase-3 by flow cytometry or stained with propidium iodide and analyzed by flow cytometry for sub-G 1 content as described previously (30). The fluorescence was measured using an Epics Elite flow cytometer (Beckman Coulter).
Transfections-Transfections were carried out as described previously (24). In brief, before transfection, 5 ϫ 10 5 cells were seeded per well in 6-well plates. SW480 cells were transfected with 2 g of DNA with a 1:10 ratio of EGFP-spectrin co-expression vector using the Lipofectin transfection reagent (Qiagen) according to the manufacturer's specifications, and transfection efficiency was about 15-20%. Twentyfour hours after transfection, cells were collected and analyzed as indicated in the figure legends.
Flow Cytometric Analysis of Mitochondrial Transmembrane Potential Change during Cell Death-Cells were washed with 2 ml of PBS, rinsed with PBS, and suspended following treatment with 0.5 ml of trypsin (trypsin-EDTA; 0.05% trypsin, 0.53 mM EDTA). Cells were transferred to 15-ml conical tubes and centrifuged at 1100 rpm for 5 min. Cells were resuspended in 300 l of media with 100 nM 3,3Јdihexyloxacarbocyanine iodide and incubated for 30 min at 37°C. The fluorescence was measured using an Epics Elite flow cytometer (Beckman Coulter).

Flow Cytometric Detection of Death Receptor Surface Expression-
The experiments were performed as described previously (25). In brief, 1 ϫ 10 6 cells were rinsed and detached following the addition of 0.5 ml of trypsin (trypsin-EDTA; 0.05% trypsin, 0.53 mM EDTA⅐4Na). Cells were transferred to 15-ml conical tubes and centrifuged at 1100 rpm for 5 min. Cells were washed with 1 ml of PBS containing 1% FBS and then resuspended in 50 l of fluorescence-activated cell sorting buffer (PBS with 1% FBS) containing the primary antibody (10 g/ml) against a death receptor or a nonspecific mouse IgG1 antibody as a control. After staining on ice for 60 min, cells were washed once with PBS containing 1% FBS and incubated with FITC-or phycoerythrin-labeled goat antimouse secondary antibody at 4°C for 45 min in the dark. Cells were analyzed on an Epics Elite flow cytometer (Beckman Coulter).
Northern Blot Analysis-RNA was extracted from parental and TRAIL-resistant SW480 cells and then electrophoresed on a 1.5% agarose gel and transferred to a nitrocellulose membrane. The mRNA expression level of human TRAIL death receptor DR4 and KILLER/ DR5 was determined through hybridization to the membrane using full-length cDNAs of the death receptors as probes. Each probe was labeled by random priming as described previously (5). Ethidium bromide staining of 28 and 18 S RNA was performed to evaluate the integrity of the RNA and to demonstrate equal RNA loading.
Reverse Transcriptase (RT)-PCR-Expression of DR4 and KILLER/ DR5 RNA was also evaluated using RT-PCR. Total RNA was extracted from the parental and TRAIL-resistant SW480 cells. cDNA was prepared using an oligo(dT) primer and Superscriptase II (Invitrogen) as previously described (26). Primers used in these experiments were obtained from R&D Systems.
Western Blotting-Western blot analyses were performed as described previously (5). In brief, a total of 20 g of protein was separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore Corp., Bedford, MA). Membranes were blocked with 10% Blotto (Carnation, Los Angeles, CA), incubated with different primary antibodies, and incubated with horseradish peroxidase-conjugated antirabbit or anti-mouse IgG antibody used at a 1:5000 dilution. Antibody-antigen complexes were detected by the Enhanced Chemiluminescence system (Amersham Biosciences).
Alanine-scanning Mutagenesis of DR4 -The DR4 cDNA was cloned in frame into the pcDNA3.1-Myc-HisA vector (Invitrogen) as an EcoRI/ HindIII fragment. This C-terminally tagged Myc-His plasmid was subsequently mutagenized using the QuikChange TM site-directed mutagenesis kit following the instructions of the manufacturer (Stratagene). Mutations were verified by sequencing, and in each case the entire cDNA was checked and verified for the absence of second site mutations.
TRAIL DISC Immunoprecipitation-DISC IP was performed as described previously (24). In brief, 1 ϫ 10 7 SW480 cells were plated to achieve 80% confluence in a T75 flask. Cells were then suspended following trypsin treatment, spun down, and resuspended in 2 ml of complete medium supplemented with 100 ng/ml His tagged-TRAIL and 1 g/ml anti-His 6 antibody (R&D Systems) for 15 min at 37°C. For untreated control samples, the TRAIL was excluded. Cells were washed twice with ice-cold phosphate-buffered saline and lysed for 30 min on ice in TRAIL DISC IP lysis buffer (30 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100). The lysates were cleared twice by centrifugation at 4°C. The supernatants were immunoprecipitated overnight with 30 l of Protein A/G Plus-agarose (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at 4°C to isolate the TRAIL DISC. The complexes were subsequently washed four times with TRAIL DISC IP lysis buffer and eluted with Immunopure Gentle Ag/Ab elution buffer (Pierce) with 0.1 M dithiothreitol at room temperature for 2 h. The protein complexes were methanol/chloroform (4:1)-precipitated and resolved on 15% SDSpolyacrylamide gels. Western blots were performed to measure recruitment of specific endogenous proteins to the TRAIL DISC.
Immunofluorescence Staining-This procedure was described previously (25). Briefly, SW480 or TRAIL-resistant cells were plated on coverslips. Coverslips were washed two times with PBS containing 1% FBS. To detect the intracellular level of death receptors, some samples were permeabilized with 0.25% saponin and fixed with 2% formaldehyde, and other samples were stained without permeabilization and fixation. Samples were incubated at 4°C with anti-DR4 or KILLER/ DR5 antibody overnight at 4°C in Dulbecco's modified Eagle's medium (Invitrogen). Cells were incubated with secondary mouse anti-goat antibody conjugated to FITC (1:400; Caltag) at 4°C for 1 h in Dulbecco's modified Eagle's medium. After three 10-min washes with PBS with 1% FBS, cells were mounted onto slides with Pro-Long (Molecular Probes) mounting medium. Cells were visualized with a ϫ20 objective lens using an Olympus BX60 epifluorescent microscope, and images were recorded with an Optronics CCD-camera (DEI-750).

TRAIL-resistant SW480 Colon Cancer Cells Selected after Prolonged Treatment with TRAIL Were Not Cross-resistant to
Fas-or Paclitaxel-induced Apoptosis-Although TRAIL has cytotoxic effects against most tumor cells, a fraction of a given sensitive tumor cell population cannot be killed even at high doses of TRAIL. To explore how these resistant survivors escape from TRAIL-induced death, we obtained TRAIL-resistant cell lines by subjecting SW480 human colon cancer cells to repeated exposure to different doses of recombinant TRAIL (see "Materials and Methods"). After five consecutive cycles of selection, we isolated 12 individual clones from a TRAIL-resistant pool of clones. Two of them remained as sensitive as the parental SW480 cells to TRAIL, and the other 10 clones showed resistance to TRAIL-induced apoptosis. We randomly chose clones R4, R6, and R10 for further experiments. As shown in Fig. 1, less than 20% apoptosis could be detected after exposure to 100 ng/ml TRAIL for 24 h in the three resistant clones, in contrast to more than 70% cell death observed in parental SW480 cells (p Ͻ 0.01). Extended exposure at higher doses or longer times of TRAIL treatment did not increase cell death, and there remained a high percentage of surviving cells. To ensure that resistance to TRAIL-induced apoptosis was not simply due to clonal variation, we further examined six subclones of the TRAIL-resistant SW480 clone R4 derived by limiting dilution. All of them showed similar resistance to TRAIL as their parental clone R4 (data not shown). It has been reported that in vitro selection for resistance to some apoptosis-inducing stimuli may result in cross-resistance to other cytotoxic agents (20,22). To determine whether our resistant clones also co-selected for resistance to Fas and paclitaxel, the parental SW480 cells and the resistant clones R4, R6, and R10 were treated with TRAIL, Fas, or paclitaxel. We found differences between parental and resistant cell clones only after TRAIL treatment but not following exposure to Fas or the chemotherapeutic drug paclitaxel (Fig. 1, A and B). We found a significant change in mitochondrial membrane potential in TRAIL-sensitive SW480 cells after TRAIL exposure (p Ͻ 0.01) but not in TRAIL-resistant cells, whereas the change was similar in response following either Fas or paclitaxel when TRAIL-sensitive and TRAIL-resistant cells were compared. (Fig. 1C). To confirm that R4, R6, and R10 became stably resistant to TRAIL-induced apoptosis, we also performed long term colony formation assays. We found that TRAIL-resistant clones formed colonies in medium containing 20 ng/ml TRAIL, whereas parental SW480 cells formed very few colonies. A low number of colonies was observed following exposure of either parental or TRAILresistant cells to either Fas or paclitaxel (Fig. 1D). Thus, TRAIL-resistant clones selected from prolonged incubation with TRAIL did not display cross-resistance to either Fas-or paclitaxel-mediated apoptotic signals.

Expression of Major Apoptosis Mediators Does Not Change in TRAIL-resistant Clones Isolated following Prolonged TRAIL
Exposure-Previous studies suggested that altered expression of proapoptotic or antiapoptotic genes in tumors often results in the avoidance of cell death induced by chemotherapeutic agents or TRAIL (14,19). To investigate the mechanism of TRAIL resistance in the selected clones, we first examined the protein levels of molecules that are involved in the death receptor signaling pathway. We tested the expression of death receptors, including DR4, KILLER/DR5, DcR1, and DcR2, and found that protein levels were similar between TRAIL-sensitive and TRAIL-resistant cell lines. DcR1 and DcR2 were expressed in TRAIL-sensitive SW480 cells, indicating that the relative expression of decoy receptors may not efficiently block apoptosis induced by TRAIL in some tumor cell lines ( Fig. 2A). IAP family members are known negative regulators of apoptosis (27). Thus, we compared the expression level of IAP1 and IAP2 in the parental SW480 cells and in resistant clones and found no difference (data not shown). We also examined the mRNA levels of the TRAIL death receptors by RT-PCR and Northern blotting, and no apparent changes were observed in the resistant clones (Fig. 2C). The antiapoptotic molecules Bcl-2, Bcl-XL, and FLIP can protect some tumor cell lines from TRAIL-induced apoptosis (23,28), and BAX is required for HCT116 cells to undergo apoptosis after treatment with TRAIL (19). Changes in expression of these molecules may not be responsible for resistance in R4, R6, and R10, because we did not detect any alteration in their protein levels ( Fig. 2A).
We evaluated the protein levels and processing of endogenous caspases in TRAIL-sensitive and TRAIL-resistant SW480 cells in response to TRAIL. As shown in Fig. 2B, processing of caspase-8, -3, and -9 was observed following exposure of parental SW480 cells to TRAIL, but not in the TRAIL-resistant clones. The procaspase protein levels were almost the same in the TRAIL-resistant clones as in the parental clones. Activation of caspase-9 after stimulation of parental SW480 cells with TRAIL may result from a mitochondrial amplification loop mediated by caspase-8-cleaved Bid. Poly(ADP-ribose) polymerase cleavage was also not observed in the TRAIL-resistant clones. These results demonstrate that caspases known to be involved in the TRAIL signaling pathway were not processed or activated in TRAIL-resistant SW480 cells.

Abrogation of DISC Formation in TRAIL-resistant Cell Lines following Prolonged Selection for TRAIL Resistance in
Culture-We compared DISC formation in TRAIL-sensitive and TRAIL-resistant cell clones. DISC formation is the initial step in the TRAIL-death receptor apoptosis signaling pathway. In addition to the receptor/ligand trimers, the TRAIL DISC also recruits the adaptor molecule FADD and caspase-8 to propagate the apoptotic signal (29). Parental SW480 and TRAILresistant clones were treated with His-tagged TRAIL and a cross-linking anti-His 6 antibody for 15 min at 37°C. The DISC was then immunoprecipitated and analyzed for the presence of caspase-8, FLIP, FADD, and death receptors. As shown in Fig.  3, treatment with TRAIL led to the recruitment of caspase-8 and FLIP into the DISC only in parental SW480 cells. Surprisingly we found that DR4 was not recruited into the DISC in TRAIL-resistant clones, but KILLER/DR5 was still present in DISC IP from both TRAIL-sensitive and TRAIL-resistant cells. The decoy receptors, DcR1 and DcR2, were not immunoprecipitated from the DISC IPs in either parental or TRAIL-resistant cells. Previous studies suggested that the level of DR4 expression correlated with TRAIL sensitivity in some tumor cell lines, including SW480 (26). The absence of DR4 in the TRAIL DISC apparently results in the failure to recruit caspase-8, and then the entire downstream signaling pathway cannot be activated. Unimpaired recruitment of KILLER/DR5 to DISC in TRAILresistant clones may be responsible for the low but detectable level of apoptosis after exposure to high doses of TRAIL. Therefore, it appears that the acquired resistance of the R4, R6, and R10 clones to TRAIL-induced apoptosis is probably due to impaired DISC formation resulting from the failure of DR4 recruitment to the DISC.
DR4 Cell Surface Expression Is Down-regulated in TRAILresistant Clones-The absence of DR4 in the DISC after TRAIL treatment implied that the surface expression of DR4 might be down-regulated in TRAIL-resistant cells. To test this hypothesis, we used flow cytometry to measure death receptor expression levels on the cell surface in both TRAIL-sensitive and TRAIL-resistant clones. Flow cytometric analysis revealed an apparent left shift of the DR4 peak that represents a significant reduction in the DR4 expression level on the cell surface in TRAIL-resistant clones, as compared with the parental SW480 cells. KILLER/DR5 did not decrease significantly on the sur- TRAIL-resistant SW480 cell lines R4, R6, and R10 were obtained as described under "Materials and Methods." In a short term assay, parental SW480 and resistant cell lines were treated with TRAIL (100 ng/ml), CH-11 antibody (2 g/ml), or paclitaxel (1 M) for 20 h, respectively. CON, control. A, apoptosis of parental SW480 and resistant cell lines was determined by PI staining and flow cytometry. B, apoptosis of parental SW480 and resistant cell lines was determined by an active caspase-3 assay. C, change of mitochondrial transmembrane potential (MTP) was detected by the 3,3Ј-dihexyloxacarbocyanine iodide assay as described under "Materials and Methods." The percentage of apoptotic cells from three independent determinations with the corresponding S.D. value is indicated. D, long term apoptosis assay. SW480 cells and resistant cells (5000/well) were plated in a 12-well dish and treated with TRAIL (20 ng/ml), Fas (400 ng/ml), or paclitaxel (0.2 M) for 3 days. The medium was subsequently changed every 3 days, and after 2 weeks, the resulting colonies were stained with Coomassie Blue. face of TRAIL-resistant clones (Fig. 4A).
Interestingly, we found that the reduced level of DR4 in TRAIL-resistant clones was similar to that in parental SW480 cells after treatment with TRAIL. This coincidence implies that reduction of DR4 cell surface expression in TRAIL-resistant clones is significant enough to possibly account for the failure of cell killing mediated by TRAIL. DcR1 and DcR2 were not detected on the cell surface in either TRAIL-sensitive or TRAIL-resistant clones.
Next, we investigated the subcellular localization and cell surface expression levels of DR4 and KILLER/DR5 by immu-nofluorescence staining. As shown in Fig. 4B, nonpermeabilized SW480 cells showed intense staining of DR4 on the plasma membrane, but very weak staining of DR4 on the plasma membrane was found in nonpermeabilized R4 cells. Studies with permeabilized SW480 and R4 cells showed similar staining of intracellular membrane structures with strong staining of the nuclear perimeter (most likely the endoplasmic reticulum-Golgi complex). The staining pattern of KILLER/ DR5 was similar in nonpermeabilized and permeabilized SW480 and R4 cells (Fig. 4B). R6 and R10 showed identical staining patterns as R4 (data not shown).
FIG. 2. The protein levels of molecules involved in the death receptor signaling pathway do not change, but caspases are not processed in TRAIL-resistant cell lines exposed to TRAIL. Parental SW480 and TRAIL-resistant R4, R6, and R10 cells were treated with or without TRAIL (100 ng/ml) for 4 h. Total cellular proteins were isolated from the untreated cells and the cells treated with TRAIL, as indicated, and aliquots of total cell lysates were subjected to SDS-PAGE and immunoblotting. A, DR4, KILLER/DR5, DcR1, DcR2, Bax, Bcl-2, Bcl-XL, and DAP3 protein expression levels are not changed in R4, R6, and R10 cells as compared with SW480. Western blot analysis of endogenous levels of these proteins was performed as described under "Materials and Methods." Actin was immunoblotted to confirm that an equivalent amount of protein was loaded in each lane. B, levels of procaspase-3, -8, and -9 and poly(ADP-ribose) polymerase are similar in SW480 cells and R4, R6, and R10 cells, but processed caspase-3, -8, and -9 and poly(ADP-ribose) polymerase could not be detected in TRAIL-resistant cell lines. C, mRNA levels of DR4 and KILLER/DR5 SW480 were not changed in TRAIL-resistant cells. Equivalent amounts (5 g) of total RNA from SW480 (control; C), R4, R6, or R10 cells were subjected to Northern blot analysis. Membranes were sequentially hybridized with DR4 and KILLER/DR5 probes and then exposed to autoradiography for 24 h. RT-PCR was also performed as described under "Materials and Methods." GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
To confirm that reduction in DR4 cell surface expression contributes to loss of susceptibility to TRAIL-induced apoptosis, we performed a functional assay with DR4 and KILLER/ DR5 antibody. Consistent with previous results (26), our data showed that KILLER/DR5 antibody conferred minor protection, whereas the anti-DR4 antibody completely blocked TRAIL-induced apoptosis in SW480 cells (Fig. 4C). Thus, DR4 seems to play a more important role than DR5 in TRAILinduced apoptosis in SW480 cells. Our data suggest that downregulation of DR4 on the cell surface may be responsible for the observed (acquired) resistance of R4, R6, and R10 clones to TRAIL-induced apoptosis.
Deficient TRAIL Death Receptor Transport to the Cell Surface following Prolonged Selection for TRAIL Resistance-Because TRAIL-resistant clones express less DR4 on the cell surface, we further investigated the underlying mechanism. It has been shown that mutation in the death domain of TNF R1 can affect translocation of the receptor to plasma membrane (30). DNA sequence analysis revealed that DR4 and KILLER/ DR5 had not acquired mutations in selected TRAIL-resistant clones (data not shown). To examine whether exogenous death receptors could be transported to the plasma membrane, we transfected wild-type and mutant DR4 and KILLER/DR5 cDNA expression vectors into parental SW480 and TRAILresistant cells, and then performed the flow cytometric assay to detect death receptor expression on the cell surface. Considering that artificial overexpression of death receptors results in apoptosis, which may interfere with the assay, we designed several DR4 and KILLER/DR5 mutants that lose the ability to induce apoptosis after transfection into cells. Fig. 5A shows alignment of the DD from TNF receptor superfamily members. Residues crucial for receptor/adaptor interactions in the case of DR5 have been identified in previous work (23). We generated four missense DR4 mutants, R385A, L389A, I394A, and W415A, using site-directed mutagenesis. As shown in Fig. 5B, these DR4 mutants, when overexpressed, display significantly reduced cell death. The KILLER/DR5 mutant R330A has been shown to lose function. Wild-type DR4, the DR4 mutant R385A, wild-type KILLER/DR5, or the KILLER/DR5 mutant R330A were co-transfected with GFP at a 10:1 ratio into parental and TRAIL-resistant clones. In Fig. 5C, only GFP-positive cells representing the specifically transfected cells are shown, and GFP-negative cells did not demonstrate any sig-nificant difference as compared with nontransfected cells (data not shown). After introduction of DR4, whether wildtype or mutant, into parental SW480 cell lines, the surface expression of DR4 was dramatically increased. However, in TRAIL-resistant clones, the surface expression of DR4 after introduction of either wild-type or mutated DR4 was not changed. The surface expression level of DR4 was even lower than in nontransfected parental SW480 cells. This observation supports the conclusion that the transport of death receptors became deficient in TRAIL-resistant cells selected from prolonged exposure to TRAIL.

The Glycosylation Inhibitor Tunicamycin Restores the Sensitivity of TRAIL-resistant Clones to TRAIL-induced Apoptosis by Increasing the Death Receptor Level on the Cell Surface-Because
TRAIL acts primarily by inducing apoptosis, defects in the apoptotic pathway may result in the failure of cancer cells to respond to therapy. Our selection scheme mimics what may occur in the application of TRAIL as an anti-cancer therapeutic drug. To explore ways of restoring the susceptibility of resistant clones to TRAIL, we have tested several different approaches. Our previous studies suggested that agents that can arrest cells in the G 0 /G 1 phase of the cell cycle may increase TRAIL-induced cytotoxicity (31). These agents included hydroxymethylglutaryl-CoA reductase inhibitors, serum starvation, or pretreatment with U0126 (a mitogen-activated protein kinase/extracellular signalregulated kinase kinase inhibitor). Additionally, it is well known that cycloheximide can sensitize cells to TNF superfamily ligandmediated apoptosis. In the present studies, we also found that a glycosylation inhibitor can enhance the sensitivity of tumor cells to TRAIL-induced apoptosis. We therefore combined these agents with TRAIL to determine whether any combination might resensitize the TRAIL-resistant clones to TRAIL-mediated apoptosis. Serum starvation and U0126 were not effective to resensitize the R4, R6, and R10 clones to TRAIL-induced apoptosis (Fig. 6A). Surprisingly, we found that pretreatment with the glycosylation inhibitor tunicamycin rendered TRAIL-resistant cells almost as susceptible as the parental SW480 cells to TRAIL-induced cell death. As expected, cycloheximide also could convert R4, R6, and R10 from TRAIL-resistant to TRAIL-sensitive cells. Because SW480 cells have mutant p53, they are resistant to many chemotherapeutic agents. Our preliminary data showed that TRAIL plus adriamycin was unable to induce apoptosis in TRAIL-resistant clones (data not shown). As shown in Fig. 6B, tunicamycin significantly enhanced the death receptor level on the cell surface, which may compensate for defects acquired in resistant clones following selection during prolonged TRAIL exposure. This effect may explain why tunicamycin can so efficiently senplasmids and harvested at 18 h for flow cytometric analysis. The percentage of apoptotic cells from three independent determinations with the corresponding S.D. is indicated. C, SW480 or R4 cells were transfected with a 1:10 ratio of GFP-spectrin/DR4 or KILLER/DR5 wild-type or mutant plasmids and then harvested at 18 h. Cells were then labeled with control IgG1, anti-DR4, or anti-KILLER/DR5 antibody followed by labeling with secondary phycoerythrin-conjugated antibody and analyzed immediately by flow cytometry. Only specifically transfected GFP-positive cells are shown. The fluorescence intensity of DR4 or KILLER/DR5 from three independent experiments with the corresponding S.D. is indicated. sitize the resistant clones to TRAIL-induced apoptosis. These findings provided a mechanism-based strategy to restore sensitivity to TRAIL-induced apoptosis. Application of TRAIL combined with a glycosylation inhibitor or similar agents, which may specifically regulate surface expression of death receptors, may overcome resistance of tumor cells that arise during cancer therapy with TRAIL or death receptor antibodies. DISCUSSION Elucidation of mechanisms underlying resistance of tumor cells to TRAIL-induced apoptosis has led to an improved understanding of regulation of the death receptor signaling pathway. Our results provide a new mechanism of how tumor cells may escape from the killing mediated by TRAIL. In the present study, we selected TRAIL-resistant cell lines by prolonged exposure of SW480 human colon cancer cells to TRAIL. TRAILresistant clones showed no cross-resistance to other apoptosisinducing agents, like Fas or paclitaxel. The mitochondrial pathway was apparently not affected by TRAIL selection. Moreover, the downstream signaling pathway from the DISC remained functional, but DISC formation was apparently impaired in the selected TRAIL-resistant clones. The level of death receptor 4 on the cell surface was significantly downregulated in TRAIL-resistant cells, as compared with parental SW480 cells. However, the total expression of DR4 mRNA or protein was not altered in the TRAIL-resistant versus the TRAIL-sensitive parental cells. Low expression of death receptors on the cell surface may be due to the failure of transport of death receptors to the plasma membrane. Pretreatment with the glycosylation inhibitor, tunicamycin, dramatically resensitized the TRAIL-resistant cells to TRAIL-induced apoptosis, thus providing a way to overcome resistance to TRAIL. Restoration of TRAIL sensitivity was accompanied by an increase in death receptor expression on the surface of previously TRAILresistant cancer cells.
It has been widely reported that in vitro selection for resistance to cytotoxic agents results in co-selection for resistance to other chemotherapeutic drugs (22). For example, cells selected for resistance to Fas were completely cross-resistant to VP-16 or staurosporin (20). However, our TRAIL-resistant clones were not cross-resistant to either Fas or chemotherapeutic drugs such as paclitaxel. R6 and R10 clones were slightly more resistant to FAS and to a much lesser extent paclitaxel, but the resistance was minor as compared with that for TRAIL. This minor cross-resistance could be due to some common death signaling intermediates, such as caspase-8, that may be shared by TRAIL, FAS, and paclitaxel. The slightly reduced expression of procaspase-8 in the TRAIL-resistant cell lines is shown in Fig. 2B. SW480 cells, which contain mutated p53 alleles, do not undergo efficient apoptosis after treatment with chemotherapeutic drugs such as adriamycin or VP16. Of several chemotherapeutic drugs we tested, including adriamycin, VP-16, etoposide, CPT-11, and paclitaxel, only paclitaxel induced significant cell death in SW480 cells. The anti-cancer activity of paclitaxel is ascribed to its unique p53-independent mechanism of action, involving mitotic arrest of cancer cells, subsequently leading to apoptosis through the inhibition of the depolymerization of microtubules (32). The level of Fas on the SW480 cell surface is very low, so we used a high concentration of CH-11 antibody to induce the apoptosis. Our results suggest that the selected TRAIL-resistant clones showed specific resistance to TRAIL-induced apoptosis.
Previous studies have suggested that SW480 cells behave as type I cells in response to TRAIL (15). Type I cells do not require early involvement of the mitochondrial pathway in death receptor-mediated apoptosis, whereas in type II cells, this amplification loop is crucial. HCT-116 cells, which behave as type II cells (15), can acquire resistance to TRAIL by developing mutation of BAX. Loss of functional BAX completely abrogates the sensitivity of HCT116 to TRAIL-induced apoptosis (19). However, in type I cells, additional mechanisms may be utilized to gain resistance to TRAIL. Although activation of caspase-9 and a change of mitochondrial membrane potential were observed in parental SW480 cells after treatment with TRAIL, blockage of the mitochondria was not observed in the TRAIL-resistant SW480 cells obtained by prolonged exposure to TRAIL.
Our data suggest that type I and type II tumor cells may avoid the cell death induced by Fas or TRAIL in different ways. The intrinsic and extrinsic pathways in type I cells are regulated though a complex network of protein interactions, and this may provide tumor cells many more choices to escape from apoptosis. It is possible that the function and/or expression of death receptors may influence the sensitivity of tumor cells to TRAIL-induced apoptosis. In addition, death receptors 4 and 5 may play a more or less important role in different kinds of tumors. For example, in melanoma, KILLER/DR5 may play a predominant role in TRAIL-induced apoptosis (33), whereas in ovarian, breast, and nasopharyngeal cancers, DR4 seems more important (26). A low level of KILLER/DR5 expression was detected, and thus it may not be actively involved in TRAILinduced apoptosis in SW480 cells. TRAIL-resistant clones also had a somewhat reduced expression of KILLER/DR5 on their cell surface as compared with the parental SW480. However, there was a striking reduction in DR4 cell surface expression when the TRAIL-sensitive and the TRAIL-resistant SW480 clones were compared. This finding suggests that DR4 may be the major determinant for TRAIL-induced apoptosis in SW480 cells and that a change of TRAIL sensitivity may occur due to modulation of its expression and localization.
The transcription factor NF-B influences apoptosis in many ways, depending on the stimulus and the cellular context. Some evidence suggests that TRAIL also activates the NF-B pathway in some tumor cell lines (34). However, modulation of the NF-B pathway appears not to be involved in the resistance to TRAIL-induced apoptosis described here. NF-B was translocated into the nucleus in both parental and TRAIL-resistant cell lines upon exposure to TRAIL (data not shown). The NF-B inhibitory peptide SN-50, which can block the activation of the NF-B pathway, had no differential effect on TRAIL-induced apoptosis in TRAIL-sensitive versus TRAIL-resistant SW480 cells (data not shown). These observations suggest that NF-B activation by TRAIL may not be involved in the acquired resistance to TRAIL studied here.
It is possible that a conformational change of DR4 results in loss of recognition of DR4 by either anti-DR4 antibody or TRAIL. However, introduction of exogenous death receptors also resulted in a significantly reduced expression on the cell surface in TRAIL-resistant cell lines. Thus, it is likely that death receptor transport is impaired in selected TRAIL-resistant SW480 clones. This conclusion was also confirmed by a receptor biotinylation assay, in which detection of denatured DR4 revealed reduced expression of DR4 on the cell surface in TRAIL-resistant clones following cell fractionation (data not shown). Reduced receptor expression on the cell surface may result from either a lower level of expression in the cell or failure to deliver the receptor to the cell surface. Western blotting, RT-PCR, and Northern blotting showed that the mRNA and protein levels of DR4 are almost identical in parental and TRAIL-resistant cell lines. Thus, it appears that the defect in TRAIL-resistant clones involves defective transport of the DR4 protein to the cell surface following prolonged exposure to TRAIL in culture. Sequence analysis revealed no mu-tation of DR4 or KILLER/DR5, so a change involving posttranslational modifications or inappropriate protein folding may account for the intracellular retention of the death receptors. The mechanism of deficient DR4 cell surface transport remains unclear, but it may involve pathways of glycosylation or trafficking.
Because tunicamycin can reverse the phenotype of TRAILresistant clones, we studied how glycosylation is involved in the delivery of death receptors to the cell membrane. We mutated candidate N-linked glycosylation sites in DR4 and compared these mutants with wild-type DR4 in terms of their ability to be transported to the cell surface. Our results showed that with the mutants we generated, there was no significant difference in TRAIL sensitivity. This information will be useful for future efforts aimed at further understanding this acquired TRAIL resistance.
Another possibility we considered in understanding the mechanism of acquired TRAIL resistance is that TRAIL-resistant cells may have changes in other molecules that may ultimately affect the transport of DR4 to the cell surface. To examine this hypothesis, we performed a microarray analysis to investigate the global gene expression changes in the TRAILresistant clones. We found that several genes were dramatically up-regulated, including NGFR p75, PRL3, APRIL, cystatins, and GMF. We confirmed up-regulation of all of these genes in two TRAIL-resistant SW480 clones by Northern analysis and RT-PCR (data not shown). We then cloned the genes and made stable SW480 clones overexpressing these genes. However, introduction of these genes individually into TRAILsensitive SW480 cells did not confer TRAIL resistance and did not alter the surface expression of DR4. Two genes, BH3BGRL and AFAP, were down-regulated in TRAIL-resistant cells. We transfected these genes back into TRAIL-resistant cells, but they also did not reverse the phenotype of these clones. p75, PRL3, and cystatin 1 were introduced in combination, but this still did not confer TRAIL resistance. With respect to the use of tunicamycin, we did not expect this drug to be specific toward TRAIL-resistant clones. However, we think it is important that tunicamycin could restore sensitivity to TRAIL in a p53-independent manner, which may be useful in the clinical application of TRAIL as a cancer therapeutic drug. Normal cells generally do not express DR4, which is different from TRAILresistant tumor cells. Tunicamycin appears to sensitize the TRAIL-resistant clones to TRAIL-induced apoptosis mainly by increasing the cell surface level of DR4, not the basal expression of DR4. Therefore, the use of tunicamycin merits further study, since it may provide a tumor-selective cytotoxicity when combined with TRAIL.
Acquired resistance of tumor cells to apoptosis is a major obstacle with cancer chemotherapy, and TRAIL is no exception. Although numerous mechanisms involved in resistance to chemotherapeutic drugs were discovered in recent years (14), little is known about how tumor cells might acquire resistance to TRAIL. In our study, the emergence of resistance to TRAIL in SW480 resulted from decreased expression of TRAIL-death receptors, particularly DR4, on the cell surface. The glycosylation inhibitor tunicamycin increased the surface expression of the TRAIL-death receptors, apparently circumventing resistance. Our studies provide a new mechanism of acquired resistance to TRAIL-induced apoptosis. Future strategies to exploit our understanding of tumor resistance to TRAIL-induced cell death may include the identification of agents that up-regulate the surface expression of death receptors by overcoming an acquired deficiency in cell surface transport of death receptors.