IFN-γupregulates apoptosis-related molecules and enhances Fas-mediated apoptosis in human cholangiocarcinoma



Human cholangiocarcinoma is a malignancy with no effective therapy and a poor prognosis. Previously, we demonstrated that cultured human cholangiocarcinoma cell lines heterogeneously express Fas on their surface, resulting in 2 subpopulations, Fas-high and Fas-low cells. Fas-low cells are resistant to apoptosis induced by Fas antibody and the calmodulin antagonists tamoxifen and trifluoperazine and are tumorigenic in nude mice (Pan et al., Am J Pathol 1999;155:193–203). Here, we show that IFN-γ enhances apoptosis in both Fas-high and Fas-low cells. IFN-γ upregulates many apoptosis-related molecules, including Fas, caspase-3, caspase-4, caspase-7, caspase-8 and Bak, in both cell lines. Pretreatment with IFN-γ facilitated Fas-mediated caspase cleavage, cytochrome c release and Bax translocation. The ability of IFN-γ to inhibit tumorigenesis of Fas-low cells was demonstrated in nude mice. Intratumoral injection of IFN-γ decreased tumor volumes by 78%. These findings indicate that IFN-γ modulates the apoptotic pathway by upregulating apoptosis-related genes. This renders tumorigenic Fas-low cholangiocarcinoma cells nontumorigenic and sensitive to Fas apoptosis, thus representing a possible therapeutic modality. © 2002 Wiley-Liss, Inc.


FasL, fos-ligan; IFN, interferon; MAb, monoclonal antibody; NFκB, nuclear factor κB; NK cell, natural killer cell; PE, phycoerythrin; PI, propidium iodide; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TBS, TRIS-buffered saline; TNF, tumor necrosis factor; TNFR, TNF receptor; TRAIL, TNF-related apoptosis-inducing ligand.

Apoptosis plays a key role not only in embryonic development and maintenance of cellular and tissue homeostasis but also in immune disorders and carcinogenesis.1 Cells from a wide variety of human malignancies show decreased ability to undergo apoptosis in response to various physical stimuli, which contributes to the clonal expansion of cancer cells.2

Fas (CD95) is a type I membrane protein and a member of the TNFR family.3, 4 Activation of Fas by FasL initiates intracellular signals that result in cell death. There are an increasing number of reports that Fas and/or FasL are expressed in many tumor cells and tumor cell lines. Abnormal expression of Fas or FasL has been found in some malignancies,5–10 suggesting that Fas/FasL may be involved in the pathogenesis of malignant transformation.

The IFNs were initially recognized for their ability to interfere with viral replication and later demonstrated to affect many cell functions, such as inhibition of growth, induction of cell differentiation and modulation of the immune response.11 IFN-γ is a pleiotropic cytokine, which plays a central role in promoting innate and adaptive mechanisms of host defense.12, 13 However, the mechanism by which IFN-γ induces growth inhibition in tumor cells has not been fully elucidated. It has been suggested that endogenously produced IFN-γ forms the basis of a tumor surveillance system. Compared with wild-type mice, mice lacking sensitivity to IFN-γ developed tumors more rapidly and with greater frequency.14 Because IFN-γ promotes antigen processing and presentation via MHC class I and class II pathways, this cytokine may induce tumor-specific proteins involved in the antigen-processing/presenting process that help the immune system in the recognition and elimination of tumor cells.12, 13, 15 IFN-γ is able to stimulate tumor cells and immune cells to express Fas receptor16–22

Cholangiocarcinoma is a highly malignant tumor of bile ducts with no effective therapy and poor prognosis.23 This tumor is an increasingly frequent diagnosis worldwide,23–25 but information about its molecular pathogenesis is lacking. Several studies of cholangiocarcinoma suggest that this tumor may escape immune surveillance via dysregulation of the Fas/FasL system. Increased expression of FasL and FLICE-inhibitory protein (FLIP), a Fas signaling inhibitor, was reported to be one possible mechanism associated with progression of cholangiocarcinoma.26 A study of intrahepatic cholangiocarcinomas suggested that upregulation of FasL is important in preneoplastic and early neoplastic stages, inducing apoptosis of tumor-infiltrating lymphocytes.27 Downregulation of the Fas receptor significantly correlated with histologic de-differentiation, vascular invasion, tumor size and short survival of patients with intrahepatic cholangiocarcinoma.27 Previously, we reported that a human cholangiocarcinoma cell line, Sk-ChA-1, expresses Fas heterogeneously, resulting in 2 subpopulations, Fas-high and Fas-low cells.28 Fas-high cells are sensitive to Fas-mediated apoptosis, whereas Fas-low cells are completely resistant to Fas-mediated apoptosis.25 Interestingly, only Fas-low cells were tumorigenic in nude mice.28, 29

Here, we show the effect of IFN-γ on Fas-mediated apoptosis and gene transcription of a broad range of apoptosis-related molecules in Fas-high and Fas-low human cholangiocarcinoma cells. We also demonstrate that IFN-γ is antitumorigenic in nude mice engraftment experiments with Fas-low cells, suggesting a possible therapeutic modality for cholangiocarcinoma.


Cells and cell culture

The human cholangiocarcinoma cell line Sk-ChA-1 was provided by Dr. A. Knuth (Ludwig Institute for Cancer Research, London, UK). Cells were grown in RPMI-1640 (Life Technologies, Gaithersburg, MD) supplemented with 2 mM L-glutamine, penicillin (5 units/ml), streptomycin (5 μg/ml) and 10% heat-inactivated FBS.

Antibodies and reagents

Human activating Fas antibody (clone CH11) and Fas-associated death domain protein (FADD) antibody were obtained from Upstate Biotechnology (Lake Placid, NY). Human anti-Fas polyclonal antibody (C-20) for Western blotting was from Santa Cruz Biotechnology (Santa Cruz, CA). Antibody for FLIP, caspases, cytochrome c and Bax and PE–conjugated antihuman Fas antibody were purchased from Pharmingen (San Diego, CA). Recombinant human IFN-γ was purchased from R&D Systems (Minneapolis, MN). Tunicamycin were purchased from Sigma (St. Louis, MO).


Athymic (nu/nu) female BALB/c mice, 6–8 weeks of age, were purchased from Charles River (Wilmington, MA) for tumor inoculation and IFN-γ treatment. All animals were maintained in a sterile environment. Cages, bedding, food and water were autoclaved; and animals were maintained on a daily 12 hr light/12 hr dark cycle.

Measurement of surface Fas expression by flow cytometry

Cells were harvested, resuspended in FACS buffer (PBS with 5% FBS, 0.1% sodium azide) and labeled with PE-conjugated Fas antibody at room temperature for 20 min. After washing twice with FACS buffer, cells were fixed with 1% paraformaldehyde and analyzed by FACScan (Becton Dickinson, Mountain View, CA).

RNase protection assay

Total RNA was isolated using RNAgents (Promega, Madison, WI), and the Multi-probe RNase protection assay was performed according to the manufacturer's directions for RiboQuant (Pharmingen). 32P-labeled antisense RNA probes were prepared using the human apoptosis multiprobe sets hAPO-3c and hAPO-1c (Pharmingen).

Preparation of whole-cell lysates and cytosolic extracts

For whole-cell lysates, cells were washed with PBS and lysed in SDS lysis buffer [100 mM TRIS-HCl (pH 8.0), 150 mM NaCl, 1% SDS, 10% glycerol, 5 mM EDTA, 5 mM EGTA, 2 mM PMSF, 1 μg/ml pepstatin and leupeptin]. To extract cytosolic proteins, cells (4 × 106) were harvested, washed twice with ice-cold PBS and resuspended in 300 μl ice-cold buffer [20 mM HEPES-KOH (pH 7.0), 10 mM KCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 250 mM sucrose, 1 μg/ml leupeptin and pepstatin, 2 μg/ml aprotinin]. After incubation on ice for 15 min, cells were homogenized with Dounce homogenizer (B pestle/25 stroke) and centrifuged at 1,000g for 10 min to separate nuclei and unbroken cells. Then, supernatants were centrifuged at 14,000g for 15 min in a microcentrifuge to pellet membranes including mitochondria. The resulting supernatants were used as cytosolic extracts.

Western blotting

Whole-cell lysates or cytosolic extracts (20 μg) were separated by 12% or 15% SDS-PAGE and transferred to Immobilon P membranes (Millipore, Bedford, MA). Membranes were blocked in TBS containing 2% nonfat milk powder and 0.1% Tween-20 and incubated with primary antibodies, followed by incubation with antimouse or antirabbit horseradish peroxidase–conjugated antibodies (Amersham, Piscataway, NJ). Blots were developed by enhanced chemiluminescence (Amersham).

Detection of apoptosis

Apoptosis was determined by annexin V and PI staining using the annexin V-FITC apoptosis detection kit (Medical and Biological Laboratories, Nagoya, Japan). After incubations as indicated in the figure legends, 1 × 105 cells were harvested and resuspended in 200 μl of binding buffer (Medical and Biological Laboratories). Annexin V-FITC and PI were added, followed by incubation at room temperature for 5 min. Annexin V binding and PI staining were analyzed by flow cytometry.

In vivo tumor growth inhibition by IFN-γ treatment

Cultured Fas-low Sk-ChA-1 human cholangiocarcinoma cells were trypsinized, washed and resuspended in PBS. Mice were anesthetized by isoflurane inhalation; then, 5 × 106 cells were inoculated s.c. into the flank of mice using a 22-gauge needle in a total volume of 0.2 ml/site. Four days later, IFN-γ treatment was initiated. At each treatment, 5 × 104 units of IFN-γ in 0.1 ml PBS were injected into the area where cells had been injected. Control mice were injected with PBS. After 3 cycles of 3 daily injections with 1 injection-free day between cycles, tumor diameters were measured. After an additional week without IFN-γ injection, tumors were measured again. Tumor volume was calculated using the following formula: tumor volume = (π/6) × (diameter).3.


Human cholangiocarcinoma cells express Fas heterogeneously

As reported previously, the human cholangiocarcinoma cell line Sk-ChA-1 can be separated into 2 subpopulations, Fas-high and Fas-low.28 Fas-low cells express some Fas, as determined by flow cytometry and Western blot, but are functionally unresponsive to stimulation by Fas antibody. As shown in Figure 1, using flow cytometry, about 98% of Fas-high cells were shifted into the M2 region (PE-positive region), whereas only 45% of Fas-low cells were in that region.

Figure 1.

Heterogeneous Fas expression in human cholangiocarcinoma cell line Sk-ChA-1. (a) Fas-high and Fas-low cholangiocarcinoma cells were stained with PE-conjugated Fas antibody and analyzed by flow cytometry. Murine IgG was used for staining isotype control. M1, PE-negative region; M2, PE-positive region. (b) Fas-high and Fas-low cells were cultured with tunicamycin (1 μg/ml) or medium alone for 72 hr. Medium was changed every 24 hr to add fresh tunicamycin. Whole-cell lysates (20 μg protein) were separated by 12% SDS-PAGE and immunoblotted for Fas. *Glycosylated proteins, **nonglycosylated proteins.

There are multiple reports of active transcriptional repression by DNA methylation.30–32 Methylation of CpG dinucleotides as a mechanism of gene silencing has been reported and suggested to play an important role in tumorigenesis by inhibiting transcription of tumor-suppressor genes.32 In addition, several studies have shown that inhibition of DNA methylation restores Fas gene expression.33, 34 To address whether reduced Fas expression in Fas-low cells is due to DNA methylation, we treated Fas-low cells with 1–2 μM of 5-aza-2′-deoxycytidine, an inhibitor of DNA methylation, for 3 days and analyzed Fas expression by flow cytometry. Fas expression in Fas-low cells was not increased in the presence of 5-aza-2′-deoxycytidine (data not shown), suggesting that methylation does not account for loss of Fas expression in Fas-low cholangiocarcinoma cells.

Western blot analysis using anti-Fas sera showed several bands in the m.w. range 35–50 kDa, indicating that various glycosylated forms of Fas were present. Both Fas-high and Fas-low cells expressed Fas antigen, but Fas-high cells showed much stronger Fas expression (Fig. 1b). To demonstrate that high m.w. Fas is a glycosylated form, we treated cells with tunicamycin, which inhibits N-glycosylation. Treatment with tunicamycin eliminated bands in the 48 kDa area (Fig. 1b, *), which appear to be glycosylated forms. In addition, lower m.w. forms of Fas were increased (Fig. 1b, **), indicating an increase of nonglycosylated or partially glycosylated forms.

IFN-γ induces susceptibility to Fas-mediated apoptosis in human cholangiocarcinoma cells

In a previous report, we demonstrated that Fas-high cells are sensitive to apoptosis induced by Fas antibody (CH11) but Fas-low cells are resistant.28 Subsequently, it was reported that pretreatment with certain cytokines (e.g., IFN-γ, TNF-γ, IL-1β) or combinations of some cytokines (e.g., TNF-α + IFN-γ, IL-1β + IFN-γ) prior to Fas activation increases cell death.16–20, 35, 36 To investigate whether one of these, IFN-γ, renders Fas-negative cholangiocarcinoma cells susceptible to Fas-mediated apoptosis, we treated Fas-high and Fas-low cells with 250 units/ml of IFN-γ for 18 hr prior to incubation with Fas antibody (activating antibody, CH11) to induce apoptosis. Pretreatment with IFN-γ enhanced the sensitivity to Fas-mediated apoptosis in both Fas-high and Fas-low cholangiocarcinoma cells (Fig. 2). Figure 2a is representative of 4 different experiments showing flow-cytometric analysis of annexin V-FITC/PI staining to determine apoptosis. Fas antibody alone killed only Fas-high cells, and 18 hr treatment of IFN-γ alone was not enough to induce apoptosis. However, when cells were pretreated with IFN-γ followed by stimulation with Fas antibody Fas-low cells became sensitive to Fas-mediated apoptosis and Fas-high cells became more sensitive. A summary of 4 separate experiments is shown in Figure 2b. More than 85% of Fas-high and 41% of Fas-low cells underwent apoptosis after Fas antibody treatment when pretreated with IFN-γ. In contrast, Fas antibody stimulated apoptosis by 52% and 17% without IFN-γ treatment-4050μ.

Figure 2.

IFN-γ enhances Fas-mediated apoptosis. (a) Fas-high and Fas-low cells were incubated with IFN-γ (250 units/ml) or medium alone for 18 hr, followed by additional 20 hr incubation with Fas antibody (CH11, 250 ng/ml). Apoptosis was analyzed by flow cytometry after staining with FITC-conjugated annexin V and PI. Cells in the different quadrants represent the following: a, viable cells; b, early apoptotic cells; c, late apoptotic/necrotic cells. (b) Mean percentage of apoptotic cells ± SE (n = 4). Annexin V-positive cells (cells in areas b and c) were counted as apoptotic.

IFN-γ increases expression of Fas

The inflammatory cytokine IFN-γ is able to stimulate expression of Fas in some tumor and immune cells. Upregulation of Fas by IFN-γ on HT29 human colon adenocarcinoma cells,16, 17 melanoma cells,18 squamous cell carcinoma19 and thyroid follicular cells20 has been reported; but induction levels are different in each cell type. We found that IFN-γ increases Fas expression in cholangiocarcinoma cells. Figure 3a shows the results of flow cytometry to detect surface Fas expression after IFN-γ treatment, demonstrating an increase in surface Fas in both Fas-high and Fas-low cells. After 12 hr treatment with 250 units/ml of IFN-γ, 97.0% of Fas-low cells were in the M2 region, whereas just 45.0% were in that region without treatment. Mean channel fluorescence was increased about 2.8-fold in Fas-high and -negative cells. Western blotting using anti-Fas sera showed an increase in the amount of both glycosylated (Fig. 3b, upper arrow) and nonglycosylated (Fig. 3b, lower arrows) Fas in Fas-high and Fas-low cells by IFN-γ.

Figure 3.

IFN-γ upregulates Fas expression. (a) Fas-high and Fas-low cells were treated with 250 units/ml of IFN-γ (solid line) or medium alone (dotted line) for 12 hr, then stained with PE-conjugated anti-Fas MAb and analyzed by flow cytometry. Numbers above the curves represent mean channel fluorescence. (b) Total Fas expression after IFN-γ treatment was analyzed by Western blot. Cells were incubated with or without IFN-γ (250 units/ml) for 18 hr, then lysed. Equivalent amounts of protein were separated by 12% SDS-PAGE and immunoblotted for Fas.

IFN-γ upregulates apoptosis-related molecules including several caspases, Bak and TRAIL

To investigate the effect of IFN-γ broadly on gene transcription, RNase protection assays were performed using antisense RNA probes of Fas, several caspases, the Bcl-2 family and other genes related to apoptosis (Fig. 4). Compared to Fas-high cells, Fas-low cells have less (almost undetectable) Fas mRNA, consistent with the low expression of Fas in Fas-low cells. There was no significant difference in the amount of caspases and Bcl-2 family members between Fas-high and Fas-low cells. After IFN-γ treatment, Fas mRNA was markedly increased in both Fas-high and Fas-low cells (Fig. 4a). In addition, IFN-γ upregulated gene transcription of caspases 3, 4, 7, 8 and 10a (Fig. 4b). Bak, a proapoptotic molecule, and Mcl-1, an antiapoptotic molecule, were upregulated among the Bcl-2 family proteins (Fig. 4c). Downregulation of the antiapoptotic Bcl-2 family was not detected (Fig. 4c). Interestingly, in both Fas-high and Fas-low cells, IFN-γ markedly increased TRAIL mRNA (Fig. 4a), which also may function as an apoptotic inducer by ligation with death receptors (DR4, DR5).

Figure 4.

IFN-γ upregulates expression of apoptosis-related molecules at the mRNA level. Fas-high and Fas-low cells were lysed after 4 hr treatment with or without IFN-γ (250 units/ml), and total RNA was isolated. RNA (15 μg) was hybridized with 32P-labeled antisense RNA probes generated from the multi-probe set (Pharmingen) hAPO-3 (a), hAPO-1 (b) and hAPO-2c (c). Molecules upregulated by IFN-γ treatment are indicated with arrows.

We also determined the levels of FADD and FLIP, which can directly influence Fas signaling. Figure 5 shows identical levels of FADD protein in Fas-high and Fas-low cells, and FADD levels remained unchanged after IFN-γ treatment. In both Fas-high and Fas-low cells, FLIP was expressed as long (FLIPL, approx. 55 kDa) and short (FLIPS, approx. 28 kDa) isoforms (Fig. 5). Although equal amounts of FLIP were detected in the 55 and 28 kDa areas, the 43 kDa fragment of FLIP, which is an intermediate cleavage product, was present only in Fas-low cells, suggesting different regulation of FLIP in Fas-high and Fas-low cells. After IFN-γ treatment, the expression pattern of FLIP was not changed in either Fas-high or Fas-low cells (Fig 5).

Figure 5.

IFN-γ does not change the level of FADD and FLIP. Fas-high and Fas-low cells were incubated with IFN-γ (250 units/ml) or medium alone for 18 hr. Whole-cell lysates were prepared and separated by SDS-PAGE, then immunoblotted for FADD and FLIP (FLIPL, long isoform; FLIPS, short isoform).

IFN-γ facilitates Fas-mediated caspase cleavage

To study whether upregulated caspases are activated during the apoptosis process, cleavage of caspases was detected by Western blotting using procaspase antibodies. Fas antibody alone can induce caspase cleavage, resulting in decreased amounts of proform caspases in Fas-high cells but not in Fas-low cells. However, when cells were preincubated with IFN-γ, proform caspases were completely cleaved after Fas antibody treatment in both Fas-high and Fas-low cells (Fig. 6). These results show that upregulated caspases are functionally involved in Fas-mediated apoptosis.

Figure 6.

IFN-γ enhances Fas-mediated caspase cleavage. Fas-high and Fas-low cells were incubated with IFN-γ (250 units/ml) or medium alone for 18 hr, followed by an additional 15 hr incubation with Fas antibody (CH11). Whole-cell lysates were prepared and separated by SDS-PAGE, then immunoblotted for procaspases 8, 3 and 7.

IFN-γ facilitates Fas-mediated Bax translocation and cytochrome c release

The mitochondria-operated apoptosis pathway involves translocation of the proapoptotic Bcl-2 family member Bax from the cytosol to the mitochondrial outer membrane, where it competes with the antiapoptotic Bcl-2 family to regulate cytochrome c release.37, 38 To determine the involvement of mitochondria in IFN-γ enhancement of Fas-mediated apoptosis in cholangiocarcinoma cells, cytoplasmic Bax and cytochrome c were examined by Western blot (Fig. 7). Cytosolic Bax was detected in both Fas-high and Fas-low cells, and depletion of the cytoplasmic pool by Fas antibody (CH11) alone was observed only in Fas-high cells and not in Fas-low cells. Similarly, in Fas-low cells, Fas antibody alone did not induce cytochrome c release, whereas it did in Fas-high cells. After IFN-γ treatment, there was almost complete depletion of cytoplasmic Bax and a remarkable increase of cytochrome c release by stimulation with Fas antibody. These findings indicate that the increased susceptibility to Fas-mediated apoptosis in IFN-γ-treated cells can be obtained, in part, by amplifying the mitochondria-dependent pathway.

Figure 7.

IFN-γ-enhanced Fas-mediated apoptosis involves Bax translocation and cytochrome c release. Fas-high and Fas-low cells were incubated with IFN-γ (250 units/ml) or medium alone for 18 hr, followed by an additional 17 hr incubation with Fas antibody (CH11). Cytosolic extracts were prepared and separated by SDS-PAGE, then immunoblotted for cytochrome c or Bax.

IFN-γ decreases tumorigenesis of Fas-low cells in vivo

Previously, we demonstrated that Fas-low cells formed tumors when injected s.c. into nude mice. In contrast, Fas-high cells did not produce any measurable tumors,29 suggesting the importance of lack of Fas expression in tumorigenesis and tumor progression. Because of the ability of IFN-γ to convert Fas-low to Fas-high cells, we examined the antitumorigenic effect of IFN-γ in Fas-low cell tumor xenografts. Four days after s.c. injection of Fas-low cells, we initiated IFN-γ treatment by injection into the site where cells had been injected. Three cycles of 3 daily injections with 1 injection-free day between cycles significantly reduced tumorigenesis. The mean tumor volume of IFN-γ-treated mice was 22% of that of control mice. One week after the last IFN-γ injection, tumor size in IFN-γ-treated mice remained 23% of control tumor size (Fig. 8a). Figure 8b shows representative tumors from control and IFN-γ-treated mice. These results support the concept that IFN-γ can reduce tumorigenesis of Fas-low cells via upregulation of apoptosis-related molecules including Fas and caspases.

Figure 8.

IFN-γ decreases tumorigenesis of Fas-low cells in nude mice. Fas-low cholangiocarcinoma cells (5 × 106) were suspended in 0.2 ml of PBS and injected into the flank of nude mice. Four days later, IFN-γ injections were initiated: 5 × 104 units of IFN-γ in PBS were given at each injection into the site where Fas-low cells had been injected (n = 8). PBS was injected into control mice (n = 6). (a) After 3 cycles of 3 daily injections with 1 injection-free day between cycles (2-week treatment) and an additional injection-free week (2-week treatment + injection-free week), tumor volume was measured. Bar graph represents mean tumor volume ± SE. (b) Representative photograph of tumors isolated from control and IFN-γ-treated mice.


Cholangiocarcinoma is a highly malignant tumor of the bile duct with a lower than 10% 5-year overall survival. Because of the inability to make the diagnosis early and the fact that the tumors are resistant to current chemotherapy and radiotherapy, the prognosis is poor.23 Therefore, new therapeutic modalities are needed. Information regarding the molecular mechanisms of the pathogenesis of cholangiocarcinoma is generally lacking. The human cholangiocarcinoma cell line Sk-ChA-1 expresses Fas heterogeneously, and the majority of cells (80%) fail to express Fas or only weakly express this receptor.28 This observation is consistent with the study of Shimonishi et al.,27 suggesting that downregulation or loss of Fas expression occurs during neoplastic transformation of biliary epithelial cells compared to rather homogeneous expression in normal biliary epithelial cells. These findings support the concept that cholangiocarcinoma may escape immune surveillance by alteration of Fas expression.

The stability of Fas expression on Fas-high and Fas-low cells was characterized by incubating cells for 2–6 weeks. The Fas expression profile was stable during culture,29 indicating that these subpopulations can be a good model for studying the molecular mechanisms of Fas-mediated apoptosis. We also found heterogeneous Fas expression in another cholangiocarcinoma cell line, OZ, and in MCF-7 breast cancer cells (data not shown).

There are several possible mechanisms to downregulate cell-surface Fas expression, including inhibition of gene transcription, alteration of posttranscriptional modification and protein processing. Fas can be localized to the Golgi and, upon appropriate stimulation, is rapidly translocated to the plasma membrane.39 Our results using flow cytometry, Western blot (Fig. 1) and RNase protection assay (Fig. 4) indicate that the transcriptional regulatory mechanism may be involved in the downregulation of Fas expression in cholangiocarcinoma. DNA methylation has been shown to suppress transcription of the Fas gene in some cell types.33, 34 However, Butler et al.40 detected no methylation in 47 specimens of colorectal carcinoma, which frequently shows reduced or no Fas expression. We can exclude the role of hypermethylation in reduced Fas expression in Fas-low cholangiocarcinoma cells since treatment of 5-aza-2′-deoxycytidine, an inhibitor of methylation, does not restore Fas expression in Fas-low cells.

The transcriptional machinery controlling Fas expression is largely unknown. However, several transcription factors have been reported to regulate Fas transcription. p53, a tumor-suppressor protein,41 has been proposed to be a trans-activator of Fas. DNA sequence studies have identified p53-binding sites in the promoter and first intron of the Fas gene.42, 43 Transcription factor NFκB also has been reported to bind Fas promoter and to upregulate Fas in activated Jurkat cells and adenoviral hepatitis.44, 45 The mechanism of inhibition of Fas transcription in Fas-low cholangiocarcinoma is under investigation in our laboratory. Our 2 subpopulations with markedly different levels of Fas expression are an excellent system for understanding the mechanism of Fas transcription.

In this study, we show that treatment of Fas-low cells with IFN-γ upregulated Fas, resulting in enhanced Fas-mediated apoptosis of Fas-low cells, which, prior to treatment, were completely resistant to Fas-mediated apoptosis. These data indicate that there is an intact Fas gene for a functional Fas receptor in Fas-low cells. Downstream caspases 3, 4, 7, 8 and 10a were also upregulated by IFN-γ (Fig. 4) and activated by cleavage during Fas-mediated apoptosis, which was confirmed by Western blot (for caspases 3, 7, 8; Fig. 6). IFN-γ-mediated upregulation of caspases has been reported by several groups. However, in most reports, increased expression of caspase-1 mRNA has been a distinctive feature in IFN-γ-treated cells.21, 22, 46 In addition, upregulation of caspases 4, 7 and 10a by IFN-γ has not been observed. Lin et al.46 reported that IFN-γ induces caspase-1, but not caspase-4, gene expression in HT-29 colon carcinoma cells. In this study, we clearly demonstrate the broad effect of IFN-γ on gene expression of the caspase family and upregulation of caspases 4, 7 and 10a in human cholangiocarcinoma cells. In contrast, caspase-1 mRNA was not significantly increased (Fig. 4b). These results suggest that caspase family members are differentially regulated by IFN-γ in different cell types.

The amount of FADD, FLIP and most of the Bcl-2 family proteins remained unchanged after IFN-γ treatment (Figs. 4, 5), which suggests that Fas upregulation may be a primary mechanism for induction of Fas sensitivity in Fas-low cells after IFN-γ treatment. In addition, upregulation of caspases and Bak facilitates Fas-mediated apoptotic signaling but is not the primary mechanism of the induction of Fas sensitivity because the basal level of these molecules is identical in both Fas-high and Fas-low cells.

However, Fas-negative cells were still less susceptible than Fas-positive cells to Fas-mediated apoptosis after IFN-γ treatment (Fig. 2). It is possible that an inhibitor(s) of the Fas-mediated apoptotic pathway, such as FLIP and inhibitor of apoptosis, may be involved. FLIP functions as an antiapoptotic molecule blocking apoptosis induced by death receptors.47, 48 When present at elevated levels, FLIP is recruited to the death-inducing signaling complex and cleaved into a p43 intermediate product, which prevents the recruitment and activation of caspase-8.48, 49 In our study, we found more p43 cleavage product of FLIP in Fas-low cells (Fig. 5) and IFN-γ did not change the level of FLIP, which may explain the lower sensitivity to Fas-mediated apoptosis of Fas-low cells even after IFN-γ treatment.

To investigate the effect of IFN-γ on tumorigenesis in vivo, Fas-low cells were injected into nude mice with or without administration of IFN-γ. Compared to control, the IFN-γ-treated group showed reduced tumor size (Fig. 8). There are multiple mechanisms by which IFN-γ could decrease tumorigenesis. IFN-γ has an antiangiogenic effect.50 Several studies have demonstrated the reduced activation of integrin αVβ3, an adhesion receptor that plays a key role in tumor angiogenesis, and decreased expression of vascular platelet endothelial cell adhesion molecule-1 (PECAM-1).51, 52 These effects may occur after IFN-γ treatment, resulting in inhibition of angiogenesis and tumor growth. Also, after IFN-γ treatment, Fas molecules will be upregulated and Fas-expressing tumor cells may be the target of NK cells, which express FasL in a constitutive manner.53, 54 Although IFN-γ treatment fails to upregulate FasL in cholangiocarcinoma cells in vitro (Fig. 4), immunostaining of tumor sections reveals increased FasL expression (data not shown). Thus, FasL-expressing tumor cells or tumor-infiltrating cells with FasL (e.g., NK cells) may kill Fas-expressing tumor cells.

TNF-α-mediated apoptosis may also contribute to the killing of tumor cells. This concept is supported by our observation that IFN-γ treatment sensitizes both Fas-high and Fas-low cholangiocarcinoma cells to TNF-α-mediated apoptosis in vitro (data not shown). Since Fas-high cholangiocarcinoma cells, which are sensitive to Fas-mediated apoptosis and nontumorigenic, are also resistant to TNF-α-mediated apoptosis (data not shown), it appears that the loss of sensitivity to TNF-α is not primarily responsible for cholangiocarcinoma tumorigenesis, though it may have a killing effect on tumors when combined with IFN-γ. It is also possible that upregulated TRAIL induces autocrine apoptosis by ligation with DR4 and DR5. The role of TRAIL-mediated apoptosis is currently under investigation in our laboratory.

In conclusion, we show that IFN-γ upregulates various apoptosis-related molecules, which enhance apoptosis initiated by the Fas receptor and inhibit growth of cholangiocarcinoma in vivo. IFN-γ may be a new therapeutic modality for human cholangiocarcinoma.


We thank Dr. N. Jhala (Department of Pathology, University of Alabama at Birmingham) for helpful discussion. This work was supported in part by NIH grants CA72823 and CA72823-S and AI4909 by a Veterans Affairs Merit Award (all to J.M.M.).