TNFα Cooperates with IFN-γ to Repress Bcl-xL Expression to Sensitize Metastatic Colon Carcinoma Cells to TRAIL-mediated Apoptosis

Background TNF-related apoptosis-inducing ligand (TRAIL) is an immune effector molecule that functions as a selective anti-tumor agent. However, tumor cells, especially metastatic tumor cells often exhibit a TRAIL-resistant phenotype, which is currently a major impediment in TRAIL therapy. The aim of this study is to investigate the synergistic effect of TNFα and IFN-γ in sensitizing metastatic colon carcinoma cells to TRAIL-mediated apoptosis. Methodology/Principal Findings The efficacy and underlying molecular mechanism of cooperation between TNFα and IFN-γ in sensitizing metastatic colon carcinoma cells to TRAIL-mediated apoptosis were examined. The functional significance of TNFα- and IFN-γ-producing T lymphocyte immunotherapy in combination with TRAIL therapy in suppression of colon carcinoma metastasis was determined in an experimental metastasis mouse model. We observed that TNFα or IFN-γ alone exhibits minimal sensitization effects, but effectively sensitized metastatic colon carcinoma cells to TRAIL-induced apoptosis when used in combination. TNFα and IFN-γ cooperate to repress Bcl-xL expression, whereas TNFα represses Survivin expression in the metastatic colon carcinoma cells. Silencing Bcl-xL expression significantly increased the metastatic colon carcinoma cell sensitivity to TRAIL-induced apoptosis. Conversely, overexpression of Bcl-xL significantly decreased the tumor cell sensitivity to TRAIL-induced apoptosis. Furthermore, TNFα and IFN-γ also synergistically enhanced TRAIL-induced caspase-8 activation. TNFα and IFN-γ was up-regulated in activated primary and tumor-specific T cells. TRAIL was expressed in tumor-infiltrating immune cells in vivo, and in tumor-specific cytotoxic T lymphocytes (CTL) ex vivo. Consequently, TRAIL therapy in combination with TNFα/IFN-γ-producing CTL adoptive transfer immunotherapy effectively suppressed colon carcinoma metastasis in vivo. Conclusions/Significance TNFα and IFN-γ cooperate to overcome TRAIL resistance at least partially through enhancing caspase 8 activation and repressing Bcl-xL expression. Combined CTL immunotherapy and TRAIL therapy hold great promise for further development for the treatment of metastatic colorectal cancer.


Introduction
Recent advance in chemotherapeutic and biological agents for metastatic colorectal cancer, combined with liver resection, has dramatically increased the survival of patients with advanced colorectal cancer [1]. However, metastasis is still the primary cause of mortality of colorectal cancer patients and there are currently very limited treatment options for patients with metastatic colorectal cancer. Therefore, novel therapeutic approaches are in urgent need. Over the past decade, accumulating experimental data from both animal models and human patients suggest that the host immune system functions as an extrinsic tumor suppressor [2] that might be developed into effective therapies against metastatic human cancer [3,4]. Molecular analysis of large cohorts of human colorectal cancers revealed that the level of T lymphocytes and immune effector molecules in the tumor microenvironment are positively correlated with the growth, metastasis and recurrence of human colorectal tumors [5,6,7,8]. Therefore, both immune cells and immune effector molecules are potentially effective anti-tumor biologic agents.
TNF-related apoptosis-inducing ligand (TRAIL, also known as TNFSF10 or APO2L) is expressed on the surface of several subsets of immune cells. TRAIL activates the extrinsic apoptosis signaling pathways upon binding to its death domain-containing receptors and has been under intense study ever since its discovery because it preferentially induces apoptosis in a wide variety of tumor cells but not in normal cells [9,10,11]. However, TRAIL only works in TRAIL-sensitive tumors and most tumor cells often exhibit a TRAIL-resistance phenotype, which is currently a major obstacle in TRAIL-based cancer therapy [12,13,14]. To overcome tumor resistance to TRAIL, various therapeutic agents are used in combination with recombinant TRAIL or TRAIL receptor agonist mAbs and have shown to be effective in enhancing TRAIL efficacy against tumor cells [15,16,17,18,19,20,21,22,23,24,25,26,27]. In addition to therapeutic agents, immune modulating cytokine IFN-c has been shown to induce TRAIL expression in various tissue NK cells and IFN-c-activated TRAIL plays a significant role in IFN-c-dependent tumor suppression [28]. Furthermore, IFN-c also regulates the expression of apoptosis-related genes to overcome tumor cell apoptosis resistance [29,30,31,32,33], suggesting that IFN-c might modulate both TRAIL expression in immune cells and TRAIL sensitivity in tumor cells. Recently, it has been shown that TRAIL receptors and caspase 8 are significantly down-regulated in high grade and metastatic head and neck squamous cell carcinoma [34], suggesting that the level of tumor cell resistance to TRAIL might increase with tumor progression. Indeed, it has been shown that metastatic human colon carcinoma cells are more resistant to TARL than the primary colon carcinoma cells [35] and we have observed that metastatic colon carcinoma cells become resistance to IFN-c sensitization (Fig. 1). Therefore, IFN-c alone is insufficient for sensitizing metastatic colon carcinoma cells to TRAIL-mediated apoptosis (Fig. 1). Because TNFa is also an inflammatory cytokine that regulates expression of apoptosis mediators [26,36], we examined whether TNFa could overcome TRAIL resistance in metastatic colon carcinoma cells. We observed that TNFa alone exerted minimal sensitization effect on metastatic colon carcinoma cells. However, when combined with IFN-c, TNFa dramatically sensitized the metastatic colon carcinoma cells to TRAIL-induced apoptosis in vitro. Furthermore, we demonstrated that TRAIL therapy and TNFa/IFN-cproducing T cell immunotherapy, when used in combination, can effectively suppress colon carcinoma metastasis in vivo. Thus, our data revealed a synergistic cooperation between TNFa and IFN-c in sensitizing metastatic colon carcinoma cells to TRAILmediated apoptosis in vitro and in suppressing colon carcinoma metastasis in vivo.

Results
TNFa cooperates with IFN-c to sensitize metastatic human colon carcinoma cells to TRAIL-induced apoptosis IFN-c has been shown to modulate TRAIL-mediated apoptosis pathways [32,33]. However, it has recently been shown that metastatic tumor cells often develop greater degree of TRAIL resistance [34,35] and we observed that metastatic colon carcinoma cells are not sensitive to IFN-c sensitization (Fig. 1A&B). TNFa has been shown to induce TRAIL expression in breast cancer cells [34]. Therefore, we hypothesized that TNFa might cooperate with IFN-c to modulate TRAIL-induced apoptosis in metastatic colon carcinoma cells. To test this hypothesis, the TRAIL-resistant metastatic human colon carcinoma SW620 cells were pre-treated with recombinant TNFa, IFN-c or both TNFa and IFN-c, and tested their sensitivity to TRAIL-induced apoptosis. SW620 cells exhibited resistance to TRAIL treatment (Fig. 1A&B). TNFa or IFN-c pre-treatment alone did not dramatically increase the tumor cell sensitivity to TRAIL-induced apoptosis (Fig. 1A&B). However, combined TNFa and IFN-c pre-treatment significantly increased the tumor cell sensitivity to TRAIL-induced apoptosis (p,0.01, Fig. 1A&B).
It has been shown that therapeutic agents can sensitize tumor cells to TRAIL-initiated apoptosis through mediating TRAIL receptor expression and function [20,37,38,39,40,41,42]. We next sought to determine whether TNFa and IFN-c regulate TRAIL receptor expression in SW620 cells. TNFa and IFN-c treatment exhibited no effect on DR4 and DR5 expression level (Fig. 1C&D). The decoy receptors T-R3 and T-R4 are undetectable on SW620 cell surface. TNFa and IFN-c treatment did not alter T-R3 and T-R4 expression (Fig. 1C&D). Therefore, TNFa and IFN-cmediated sensitization of colon carcinoma cells to TRAIL-induced apoptosis does not depend on increasing DR4 and DR5 expression or decreasing T-R3 and T-R4 expression.
TNFa and IFN-c modulate survivin and Bcl-xL expression in metastatic colon carcinoma cells Chemotherapeutic sensitization agents have been shown to alter the expression level of key apoptosis regulators in tumor cells [21,43,44,45,46,47]. Next, we analyzed the effects of TNFa and IFN-c on the expression and/or activation of apoptosis mediators. We observed that TNFa treatment decreased survivin protein level and combined treatment of TNFa and IFN-c decreased Bcl-xL protein level in the metastatic SW620 cells (Fig. 2A). The expression levels of Bcl-2, FLIP, cIAP1 and xIAP were not altered by TNFa and IFN-c (Fig. 2B). Analysis of mRNA level of survivin and Bcl-xL revealed that TNFa and/or IFN-c regulate survivin and Bcl-xL in the gene expression level ( Fig. 2A). To determine whether survivin and Bcl-xL contribute to TRAIL resistance in SW620 cells, survivin and Bcl-xL were silenced in the tumor cells by transfection with survivin-and Bcl-xL-specific siRNAs, respectively. RT-PCR analysis indicated that introduction of siRNAs dramatically reduced survivin and Bcl-xL expression level in the tumor cells (Fig. 3A). Silencing Bcl-xL significantly increased SW620 cell sensitivity to TRAIL-induced apoptosis (Fig. 3B). However, silencing survivin failed to overcome TRAIL resistance in SW620 cells (Fig. 3B). To further determine the roles of Bcl-xL and survivin in TRAIL resistance, SW620 cells were transfected with Bcl-xL and survivin-expressing plasmid, respectively, and analyzed their sensitivity to TRAIL-induced apoptosis. Overexpression of Bcl-xL significantly decreased TNFa and IFN-csensitized and TRAIL-induced apoptosis in SW620 cells (Fig. 3C&D). However, although silencing survivin did not alter the tumor cell sensitivity to TRAIL-induced apoptosis (Fig. 3A&B), overexpression of survivin also significantly decreased TNFa and IFN-c-sensitized and TRAIL-induced apoptosis in SW620 cells (Fig. 3C&D). Taken together, our observations suggest that TNFa and IFN-c sensitize the metastatic colon carcinoma cells to TRAIL-induced apoptosis at least partially through repressing Bcl-xL expression.
Caspase 8 is required for TRAIL-induced apoptosis [48,49], and it is known that chemotherapeutic agents modulate caspase 8dependent and mitochondrion-mediated apoptosis pathway to sensitize tumor cells to TRAIL-initiated apoptosis [50,51,52]. It is also known that IFN-c can regulate caspase 8 expression to mediate apoptosis [29,53]. Therefore, we reasoned that IFN-c and/or TNFa might also mediate the intrinsic apoptosis pathway to sensitize colon carcinoma cells to TRAIL-induced apoptosis. Analysis of SW620 cells revealed that TRAIL induced undetectable to weak caspase 8 activation (Fig. 2C). IFN-c or TNFa treatment alone exhibited some effects on caspase 8 activation. In contrast, combined IFN-c and TNFa pre-treatment dramatically increased TRAIL-induced caspase 8 cleavage in the SW620 cells as compared to IFN-c or TNFa treatment alone (Fig. 2C). Consistent with enhanced caspase 8 activation, cytochrome C release, an activation indicator of the mitochondrion-mediated apoptosis pathway, and PARP cleavage, a biochemical indicator of apoptosis, were dramatically increased in IFN-c and TNFapretreated cells after TRAIL treatment (Fig. 2C). Taken together, our data suggest that IFN-c and TNFa sensitize human colon carcinoma cells to TRAIL-induced apoptosis also through modulating caspase 8 activation.

TNFa simultaneously induces NF-kB activation and TRAIL-mediated apoptosis sensitization
Our above data demonstrated that TNFa, when used in combination with IFN-c, can sensitize metastatic human colon carcinoma cells to TRAIL-induced apoptosis. However, TNFa is also a potent activator of NF-kB [54] and NF-kB has been shown to play a important role in TRAIL resistance [55,56]. Thus, TNFa may simultaneously activate apoptosis and cell survival pathways, two conflicting biological processes, in human colon carcinoma cells. To determine whether these two conflicting pathways coexist and interferes with each other, we examined TNFa-induced NF-kB activation and the effects of blocking NF-kB activation on TRAIL-induced apoptosis in human colon carcinoma cell line SW480. SW480 cell line was chosen since we have a wellestablished NF-kB activation model in this cell line. SW480 cells exhibited spontaneously activated NF-kB activity, albeit at low level. Treatment of the tumor cells with recombinant TNFa rapidly and transiently activated NF-kB (Fig. 4A). Although IFN-c cooperates with TNFa to enhance TRAIL-induced apoptosis, IFN-c did not alter TNFa-mediated NF-kB activation (Fig. 4A). It has been shown that it is IKKb that activate the canonical NF-kB to promote tumor [57]. Next, we stably transfected SW480 cells with empty vector (SW480.Vector) and a vector expressing IKKb mutant IKKb-K44A (SW480.IKKb-KA) [58], and examined the effects of inhibition of NF-kB activation on colon carcinoma cell sensitivity to TRAIL. EMSA analysis indicated that ectopic expression of the IKKb mutant blocked both constitutively and TNFa-induced NF-kB activation (Fig. 4A).
SW480.IKKb-KA cells exhibited a slight increase in sensitivity to TRAIL-induced cell death than SW480.Vector cells under our culture conditions (approximately 1.4% more) (Fig. 4B&C). However, the TRAIL-induced cell death in IFN-c, TNFa and both IFN-c and TNFa treatment groups of SW480.IKKb-KA cells is significantly higher as compared to those in of SW480.Vector cells (Fig. 4B&C), suggesting that TNFa-activated NF-kB does interfere with TNFa-sensitized apoptosis. Nevertheless, TNFa-mediated apoptosis sensitization function apparently overpowers TNFa-induced and NF-kB-mediated cell survival effect to result in an overall apoptosis sensitive phenotype in human colon carcinoma. Our data thus suggest that blocking NF-kB activity might increase human colon carcinoma cells to IFN-c/ TNFa-sensitized and TRAIL-induced apoptosis.

TRAIL plays a significant role in immune cell-mediated colon carcinoma rejection
To translate the above findings to TRAIL-based therapy against colon carcinoma metastasis, we next examine the function of TRAIL in suppression of colon carcinoma in preclinical mouse models. Because immune cells express TRAIL [28,59,60], we used mouse colon carcinoma model [55] to determine whether colon carcinoma cell-activated immune cells express TRAIL. Mouse colon carcinoma CT26 cells were transplanted to BALB/c mice to develop lung metastases. Approximately 21 days after tumor transplant, tumor-bearing lungs were excised to make single cell suspension. Infiltrating immune cells were identified in the tumor-bearing lungs (Fig. 5A). Macrophage consists of the largest population of tumor infiltrating immune cells (3.45%), followed by NK cells Analysis of protein levels of antiapoptotic genes. Cells were treated as described in A and analyzed by Western blotting analysis for the indicated proteins. C. Caspase activation and apoptosis. SW620 cells were treated with IFN-c, TNFa or both IFN-c and TNFa overnight, followed by incubation with recombinant TRAIL protein (100 ng/ml) for the indicated time. Total cell lysates were then prepared and analyzed by Western blotting for activated caspase 8. Cytosol fractions were also prepared from cells as treated above and analyzed for cytochrome C release and PARP cleavage. doi:10.1371/journal.pone.0016241.g002 (1.44%), CD8 + T cells (1.08%) and CD4 + T cells (0.23%). Flow cytometry analysis revealed that 79-96% of these infiltrating immune cells express TRAIL protein on their surface (Fig. 5A). RT-PCR analysis confirmed that TRAIL is expressed in these four subsets of tumor-infiltrating immune cells (Fig. 5B). To validate TRAIL expression in immune cells in a more defined system, we then stained TRAIL protein in a CT26 tumorspecific cytotoxic T lymphocyte (CTL) line. The CTLs were stimulated with irradiated tumor cells and analyzed for TRAIL protein level on the cell surface. It is clear that these tumorspecific CTLs express high level of TRAIL (Fig. 5C).
To determine whether TRAIL plays a significant role in tumor rejection, CT26 cells were mixed with IgG control mAb and TRAIL neutralizing mAb, respectively, and injected to syngeneic mice. Analysis of lung metastasis revealed that blocking TRAIL function significantly increased CT26 tumor cell metastasis rate (p,0.001) (Fig. 5D). In summary, our data suggest that TRAIL protein is expressed in tumor-infiltrating immune cells and plays a

Level of tumor infiltrating CD8 + T cells inversely correlates with tumor progression stage
Our above observations suggest that both tumor-infiltrating CD8 + T cells and in vitro activated CD8 + T cells express TRAIL (Fig. 5). We next analyzed T cell infiltration in human colorectal cancer specimens using a colorectal cancer progression tissue microarray (TMA) and observed that CD8 + T cells are present in all 14 adenoma specimens examined with an average of 79 cells per section (Fig. 6A, a&b). The average CD8 + T cell number is 37 per section in the adenocarcinoma (Fig. 6A, c&d). Six of the 7 distal metastases specimens (4 liver metastases and 3 lung metastases) exhibited few than 22 CD8 + T cells per section (Fig. 6A, e), whereas one of the 7 metastases specimens (lung metastases) has 125 CD8 + T cells per section (Fig. 6B). Post hoc pair-wise comparisons showed that the adenoma has significantly higher mean CD8 + T cells than adenocarcinoma (p = 0.0051) and Liver/Lung metastases (p = 0.0036). Although most of the adenocarcinoma specimens have more CD8 + T cells than the liver/lung metastases (37 vs 26), no statistically significant difference was found between adenocarcinoma and the liver/lung metastases (p = 0.4957), probably due to the higher CD8 + T cell number in one of the lung metastases specimen (Fig. 6B).

CTL adoptive immunotherapy in combination of TRAIL therapy effectively suppresses colon carcinoma metastasis
Our data suggest that IFN-c and TNFa when used in combination, are effective sensitizers for TRAIL-induced apoptosis in metastatic colon carcinoma cells (Fig. 1), however, system infusion of exogenous IFN-c or TNFa are often highly toxic to the host [36], thereby limiting their clinical use. CD8 + T cells rapidly up-regulate IFN-c and TNFa expression upon activation (Fig. 7A), therefore, the locally produced IFN-c and TNFa in the tumor microenvironment by tumor-infiltrating T cells should be non-toxic and yet effective sensitizers in TRAIL therapy. Because CD8 + T cells also express TRAIL (Fig. 5C), the function of tumor-specific CD8 + T cells can be two folds: first, CD8 + T cells may infiltrate inside tumor (Fig. 6Aa b&d) and utilize IFN-c, TNFa and TRAIL to induce tumor cell apoptosis; second, CD8 + T cells may not effectively infiltrate the advanced tumor, especially the metastatic tumor (Fig. 6A, c&e), however, the activated CD8 + T cells can still secrete IFN-c and TNFa. IFN-c and TNFa may move inside the tumor through peripheral blood circulation to sensitize the tumor cells. In that case, exogenous TRAIL may be applied to treat the TRAIL-resistant cancer. To test this hypothesis, we first sought to determine whether TRAIL plays a significant role in CTLmediated suppression of colon carcinoma metastasis. CT26 cells were transplanted to syngeneic mice to establish lung metastases. Five days later, tumor-specific and perforin-deficient pfpCTLs were incubated with IgG control mAb and TRAIL neutralizing mAb, respectively, and adoptively transferred to the tumor-bearing mice. The CT26 cells are Fas-resistant. Use of perforin-deficient pfpCTLs and Fas-resistant CT26 tumor cells will eliminate the function of perforin and Fas/FasL effector mechanisms of the CTLs, thus reducing the CTL cytotoxicity background for optimal TRAIL function evaluation. It is clear that blocking TRAIL function significantly decreased CTL-mediated tumor rejection efficacy (p = 0.03) (Fig. 7B), suggesting that tumor-specific CTLs at least partially use TRAIL to suppress colon carcinoma development in vivo. Next, we used CT26 lung metastasis mouse model to determine whether combined therapy of TRAIL and CTL adoptive transfer is more effective in suppressing established lung metastases than single agent therapy. CT26 cells were transplanted to syngeneic mice for 5 days to establish extensive lung metastases. pfpCTL and TRAIL protein were then injected to the tumor-bearing mice either alone or in combination. The prediction is that if CTLs indeed can secrete IFN-c and TNFa to sensitize the tumor cells, then combinational therapy of CTL adoptive transfer and TRAIL therapy should exhibit greater anti-tumor efficacy than CTL or TRAIL alone. Indeed, TRAIL therapy alone exhibit no significant efficacy against the TRAIL-resistant CT26 colon carcinoma (p = 0.15). Although pfpCTL alone showed a significantly antitumor cytotoxicity (p,0.001), combinational therapy of CTLs and TRAIL exhibits significantly greater tumor rejection efficacy against the established colon carcinoma lung metastases than CTL alone (p,0.001) (Fig. 7C). In summary, our data suggest that TRAIL therapy alone is ineffective in suppressing TRAIL-resistant colon carcinoma in vivo. Combined TRAIL therapy and CTL adoptive transfer immunotherapy is significantly more effective than CTL adoptive immunotherapy alone for the treatment of metastatic colon cancer.

Discussion
We demonstrated here that combined treatment of TNFa and IFN-c, two physiologic cytokines of the host immune system, effectively sensitized metastatic human colon carcinoma cells to TRAIL-induced apoptosis (Fig. 1). Therefore, TNFa and IFN-c is a pair of sensitizers that can effectively overcome TRAIL resistance in metastatic colon carcinoma cells.
The molecular mechanisms underlying TRAIL resistance in tumor cells have been an active research area. It has been shown that decreased TRAIL receptor level or increased decoy TRAIL receptor level can lead to enhanced TRAIL resistance [12]. Similarly, the altered expression of anti-apoptotic Bcl-2 family proteins can confer the tumor cells with TRAIL resistance [12,21,24,44,45,46,61,62,63]. It has also been shown that antiapoptotic protein survivin is highly expressed in colon carcinoma cells [64,65]. These observations suggest that TRAIL resistance mechanisms might be tumor type and stage-dependent. In this study, we found that TNFa decreased the expression level of survivin ( Fig. 2A). However, although survivin is a protein that inhibits apoptosis and promote cell survival [66], silencing survivin failed to overcome TRAIL resistance in metastatic colon One-way analysis of variance was used with a Bonferroni multiple comparison procedure to test for pair-wise post hoc differences between the three types of tumor specimens. Adenoma has a significantly higher mean CD8 + T cells than adenocarcinoma (p = 0.0051) and liver/lung metastases (p = 0.0036). Adenocarcinoma was not significantly different than liver/lung metastases in mean CD8 + T cells (p = 0.4957). doi:10.1371/journal.pone.0016241.g006 carcinoma cells, suggesting that Bcl-xL might be a limiting determinant of TRAIL resistance in the metastatic colon carcinoma cells. Indeed, silencing Bcl-xL expression significantly increased the tumor cell sensitivity to TRAIL-induced apoptosis and overexpression of Bcl-xL significantly decreased the tumor cell sensitivity to TNFa and IFN-c-sensitized and TRAIL-induced apoptosis. Therefore, it seems that Bcl-xL is at least partially responsible for TRAIL resistance in the metastatic colon carcinoma cells. In addition, we observed that TNFa and IFN-c cooperatively enhance TRAIL-induced caspase 8 activation (Fig. 2C). It is known that enhanced caspase 8 recruitment to the DISC is essential for overcoming TRAIL resistance in human hepatocellular carcinoma cells [49], therefore it is possible that TNFa and IFN-c may enhance caspase 8 association with the DISC or mediate DISC conformation to alter the caspase 8 cleavage kinetics in human colon carcinoma cells, which remains to be determined.
TNFa is a potent inducer of NF-kB activation [55,67]. Exposure of human colon carcinoma cells to TNFa rapidly activated NF-kB (Fig. 4A), and blocking NF-kB activation significantly increased human colon carcinoma cells to TRAILinduced apoptosis (Fig. 4B), suggesting that NF-kB does counteract with TRAIL-induced apoptosis. However, even though TNFa induces NF-kB activation, TNFa functions primarily as a sensitizer of TRAIL-induced apoptosis and it seems that TNFamediated apoptosis sensitization function apparently overpowers the TNFa-induced and NF-kB-mediated tumor cell survival effects. In the literature, it has been shown that NF-kB activation promotes inflammation-mediated tumor cell survival and progression [54] and blocking NF-kB activation can convert inflammation-induced tumor progression mediated by TNFa to TRAILmediated tumor regression in an experimental metastasis mouse model [55]. It is very likely that blocking NF-kB activation might enhance tumor cell sensitivity to TRAIL-induced apoptosis [56] and inhibit inflammation-mediated tumor promotion in the tumor microenvironment, thus enhancing TNFa function in sensitization of metastatic colon carcinoma cells in TRAIL therapy, which requires further study.
The major subsets of immune cells, including T cells, NK cells and myeloid cells, all express TRAIL (Fig. 5) and activated T cells produce IFN-c and TNFa (Fig. 7). Therefore, IFN-c, TNFa and TRAIL of these immune cells may cooperate to effectively induce tumor cell apoptosis when infiltrated into tumor. However, as demonstrated in this study and reported in the literature, the CT26 cells (1610 5 cells/mouse) were injected into mice. Five days later, perforin-deficient CTLs (pfp CTL, 1610 6 cells/mouse) were mixed with IgG or TRAIL neutralizing mAb (50 mg/mouse) and injected into the tumor-bearing mice. IgG and TRAIL neutralizing mAb (100 mg/mouse) were injected again into the tumor-bearing mice 2 days later. Mice were sacrificed 14 days after CTL treatment and analyzed for lung metastasis. Images of lungs from representative mice are shown. The number of lung tumor nodules was enumerated in a single-blinded fashion. Each dot represents total tumor counts from a single mouse. The difference between CTL+IgG control group and CTL+Anti-TRAIL mAb group is statistically significant (p = 0.03). C. Combined TRAIL and CTL adoptive immunotherapy effectively suppressed colon carcinoma metastasis. CT26 cells (1610 5 cells/mouse) were injected into mice. Five days later, pfp CTLs (1610 6 cells/mouse), TRAIL (200 mg/mouse), or both pfpCTL and TRAIL were injected into the tumor-bearing mice. Mice groups that received TRAIL or both pfpCTL and TRAIL were injected with TRAIL (200 mg/mouse) again every 2 days for 4 more times. Mice were sacrificed 16 days after tumor transplantation and analyzed for lung metastasis. The number of lung tumor nodules was enumerated in a singleblinded fashion. Each dot represents total counts from a single mouse. Counts greater than 250 are expressed as $250. The difference between control and TRAIL treatment group is not statistically significant (p = 0.15). The differences between control and pfpCTL treatment alone group and between control and pfpCTL+TRAIL treatment group are both statistically significant (p,0.001). doi:10.1371/journal.pone.0016241.g007 tumor-reactive immune cells (i.e. CD8 + T cells) are frequently unable to infiltrate the colorectal tumor, especially metastatic colorectal carcinoma (Fig. 6). Because TRAIL needs to directly contact the TRAIL receptors on the tumor cell surface to induce apoptosis, lack of infiltration of immune cells into the tumor may result in lack of TRAIL-induced apoptosis of the tumor cells. IFNc, TNFa and TRAIL are soluble molecules and thus should be able to penetrate into the tumor through the blood circulation. However, systemic use of IFN-c and TNFa is highly toxic. Therefore, we reasoned that use of adoptive transfer of tumorspecific CTLs to produce IFN-c and TNFa locally in the tumor microenvironment in combination with TRAIL protein/mAb therapy should effectively induce colon carcinoma cell apoptosis, and thereby suppressing colon cancer metastasis. In our prove of concept study, we demonstrated that tumor-specific CTL adoptive transfer immunotherapy, when combined with TRAIL therapy, achieved significantly greater metastasis suppression efficacy against the TRAIL-resistant colon carcinoma than either therapy alone (Fig. 7C). Taken together, our results suggest that combined CTL immunotherapy and TRAIL therapy hold great promise for further development for the treatment of metastatic colon cancer.

Mice
Mice were purchased from the National Cancer Institute (Frederick, MD) and housed in the Medical College of Georgia animal facility. All Experiments with mice and care/welfare for mice used in this study were in agreement with National Institutes of Health regulations and were carried out with a protocol (Protocol # 05-12-728*B) approved by the Medical College of Georgia Institute Animal Care and Use Committee.

Tumor cells and specimens
All colon carcinoma cell lines were obtained from ATCC (Manassas, VA). De-identified human colon carcinoma specimens were provided by the Cooperative Human Tissue Network (CHTN) sponsored by the National Cancer Institute. All studies with human tumor specimens were carried out in accordance with NIH and MCG guidelines.

Measurement of TRAIL-induced apoptosis
Recombinant TRAIL protein was expressed and purified as described [68]. Cell death was measured by Propidium Iodide (PI, Trevigen Inc. Gaithersburg, MD) staining and flow cytometry analysis as described [68]. IFN-c and TNFa were obtained from R&D System (R&D Systems, Minneapolis, MN).

Western Blot Analysis
Total cell lysates and cytosol fractions were prepared and analyzed by Western blotting analysis as previously described [68].

Electrophoresis Mobility Shift Assay (EMSA) of NF-kB activation
NF-kB activation was analyzed using EMSA with NF-kB probe (AGT TGA GGG GAC TTT CCC AGG C, Santa Cruz Biotech) as previously described [69]. Briefly, the end-labeled probes were incubated with nuclear extracts for 20 min at room temperature. For specificity controls, unlabeled probe was added to the reaction at a 1:100 molar excess. Anti-p50 subunit antibody (Santa Cruz Biotech) was also included to identify NF-kB-specific DNA binding. DNA-protein complexes were separated by electrophoresis in 6% polyacrylamide gels and identified using a phosphoimage screen (Molecular Dynamics) and the images were acquired using a Strom 860 imager (Molecular Dynamics).

RT-PCR analysis
Total RNA was isolated from cells or tissues using Trizol (Invitrogen, San Diego, CA) and used for RT-PCR analysis of gene expression as described [70].

Tumor cell transfection
Bcl-xL plasmid (pSFFV-neo.Bcl-xL) [71] was obtained from Addgene. Survivin plasmid [64] was kindly provided by Dr. Michael Brattain (University of Nebraska Medical Center). Tumor cells were transient transfected with the vector control plasmid, Bcl-xL-or Survivin-expressing plasmid, respectively, using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Tumor cells were harvested approximately 20 h after transfection and used for the indicated studies.

Experimental lung metastasis mouse model and CTL immunotherapy
Tumor-specific CTLs were generated from perforin-deficient BALB/c mice as previously described [73]. The experimental lung metastasis mouse model and CTL adoptive transfer immunotherapy was carried out as previously described [73].

Statistical Analysis
All statistical analyses for T cell infiltration in the tumor specimens were performed using SAS 9.2. A one-way analysis of variance (ANOVA) and a Tukey-Kramer multiple comparison procedure was performed to analyze correlation between receptor level and apoptosis rate.