Alteration of dihydropyrimidine dehydrogenase expression by IFN-α affects the antiproliferative effects of 5-fluorouracil in human hepatocellular carcinoma cells

5-Fluorouracil (5-FU) is widely used in the treatment of various gastrointestinal cancers and other types of tumor. It is converted to the active metabolite 5-fluoro-2′-deoxyuridine-5′-monophosphate (FdUMP) and inhibits thymidylate synthase (TS) activity competitively through the formation of a ternary complex of FdUMP, TS, and 5,10-methylenetetrahydrofolate. Cancer cells with high levels of FdUMP and low levels of TS are thus known to be sensitive to 5-FU (1).

Dihydropyrimidine dehydrogenase (DPD) is a rate-limiting enzyme involved in the degradation of pyrimidine bases and pyrimidine-based antimetabolites, such as 5-FU, and so diminishes the antitumor activity of 5-FU. This catabolism occurs mainly in the liver. DPD activity shows wide variation in both cancer patients and the healthy population (2).

The human DPD gene (DPYD) is located on chromosome 1p22. It is a single copy 950-kb gene comprising 23 exons (3), in which 39 mutations and polymorphisms have been identified (4–6). Abnormalities of DPYD that decrease DPD activity are observed in 3% to 5% of the total population (7), and several patients with congenital DPD deficiency were reported as suffering from severe toxicity after the administration of 5-FU (8). DPD activity in tumor cells is critical to the antitumor effects of 5-FU (9), and its inhibition is expected to enhance these effects.

Hepatocellular carcinoma (HCC) is the fifth most common malignancy in the world. The most effective treatment for patients with HCC is the surgical resection of hepatic lesions. Local therapeutic approaches, such as transcatheter arterial embolization (10), percutaneous transhepatic ethanol injection (11), microwave coagulation (12), and radiofrequency ablation (13), are also effective. However, these therapies are not sufficient for patients with advanced HCC, for whom surgery is often not suitable and whose 5-year survival rate is extremely low (14). For patients with advanced HCC, the clinical response of almost every anticancer drug is insufficient, and several combination chemotherapies have been tried. Combined chemotherapy with 5-FU and IFN-α has been used previously for patients with advanced HCC, with improved therapeutic effects (15–17). However, these reports of positive effects are contradicted by a previously observed lack of antitumor activity accompanied by increased toxicity (18). To improve the therapeutic index of 5-FU and IFN-α combination therapy, it is therefore important to establish the cause of this occasional decrease in efficacy and increase in side effects.

Plausible mechanisms, such as an increase in FdUMP, TS inhibition rate, and thymidine phosphorylase activity, a decrease in TS levels, and altered 5-FU pharmacokinetics, have been suggested to explain the improved therapeutic effects of IFN-α and 5-FU (19–23). Takaoka et al. (24) showed that transcription of the p53 gene is induced by IFN-α/IFN-β, accompanied by an increase in p53 protein level, and that the apoptotic response by IFN-β combined with 5-FU was enhanced. Milano et al. (25) reported that IFN-α inhibits DPD activity in human tumor cells, suggesting that inhibition of DPD activity could be involved in 5-FU–induced antiproliferative activity. Consensus IFN was shown to enhance the antiproliferative effect of 5-FU against hepatoma cells through down-regulation of DPD expression (23). By contrast, increased expression of DPD protein by IFN-γ was reportedly observed at a concentration equivalent to that in the sera of patients (26).

According to the presence or absence of synergism by the combination of 5-FU and IFN-α, we classified six human HCC cell lines into two groups: the S-group containing three cell lines, which showed a synergistic effect, and the A-group containing the remaining three cell lines, which showed additive effects (27). The expression levels of type I IFN receptor subunits were specifically up-regulated by 5-FU in all three cell lines of the S-group but not in those of the A-group (27). In this study, we asked whether DPD could limit the antiproliferative effect of 5-FU against HCC cells in culture when 5-FU was applied in combination with IFN-α. This work also shows that inhibiting DPD activity in HCC cells with high DPD levels improves the efficacy of 5-FU combined with IFN-α, following up-regulation by IFN-α.

5-FU was purchased from Kyowa Hakko Kogyo Co. Ltd. (5-FU Injection 250 Kyowa) and natural human IFN-α was purchased from Otsuka Pharmaceutical Co. Ltd. (OIF). 5-Chloro-2,4-dihydroxypyridine (CDHP) was a gift from Taiho Pharmaceutical Co. Ltd.

HCC cell lines, KYN-2, KYN-3, and HAK-1A (28, 29), were grown in DMEM (Nissui Seiyaku Co.) with 10% fetal bovine serum (FetalClone III, Hyclone) in a humidified atmosphere of 5% CO₂ at 37°C. We confirmed the expression of type I IFN receptor subunits 1 and 2 in these three HCC cell lines (27).

Cells were seeded into 96-well plates at 1,000 cells/100 μL/well and incubated overnight. On the following day, 100-μL aliquots containing IFN-α and/or 5-FU with or without CDHP were added to each well and cells were cultured for a further 5 days. The number of viable cells was estimated by assaying the activity of cellular succinate dehydrogenases using WST-8 reagent (Cell Counting Kit-8, Dojindo; ref. 30). We confirmed that untreated groups of KYN-2, KYN-3, and HAK-1A cells grew exponentially for 6 days under these experimental conditions (data not shown).

The combined effects of 5-FU and IFN-α were quantified using a combination index (CI) method developed by Chou and Talalay (31). This method involves plotting dose-effect curves, for each agent and their combination, using the median-effect equation: fa / fu = (D / Dm)^m, where D is the dose of the drug, Dm is the dose required for a 50% effect (equivalent to IC₅₀), fa and fu are the affected and unaffected fractions, respectively (fa = 1 − fu), and m is the exponent signifying the sigmoidicity of the dose-effect curve.

In this study, relative concentrations (RC) of IFN-α and 5-FU, determined as (concentration) / (IC₅₀ value), were used for analysis. The computer software Xlfit version 2.0.6 (ID Business Solutions Ltd.) was used to calculate the values of Dm and m. The CI used for the analysis of the drug combinations was determined by the isobologram equation for mutually nonexclusive drugs that have different modes of action: CI = (D)₁ / (Dx)₁ + (D)₂ / (Dx)₂ + (D)₁(D)₂ / (Dx)₁(Dx)₂, where (D)₁ and (D)₂ are RCs of drugs 1 and 2 and x is the percentage of inhibition. Combination indices CI < 1, CI = 1, and CI > 1 indicate synergism, additive effects, and antagonism, respectively.

Total RNA was extracted using Isogen (Nippon Gene Co., Ltd.) and reverse transcribed using a reverse-transcription system (Promega Corp.) according to the manufacturer's instructions. Quantitative real-time reverse transcription-PCR was done with an ABI Prism 7300 (PE Applied Biosystems). The primers used were as follows: TS 5′-GAATCACATCGAGCCACTGAAA-3′ (forward primer), 5′-CAGCCCAACCCCTAAAGACTGA-3′ (reverse primer), and 5′-(FAM)TTCAGCTTCAGCGAGAACCCAGA(TAMRA)-3′ (probe) and DPD 5′-AATGATTCGAAGAGCTTTTGAAGC-3′ (forward primer), 5′-GTTCCCCGGATGATTCTGG-3′ (reverse primer), and 5′-(FAM)TGCCCTCACCAAAACTTTCTCTCTTGATAAGGA(TAMRA)-3′ (probe). Primers and Taqman probes for glyceraldehyde-3-phosphate dehydrogenase were prepared by Assay-on-Demand Gene Expression Products (PE Applied Biosystems).

HCC cells were cultured for 48 h with 0, 20, 100, or 500 IU/mL IFN-α. Total protein was extracted using a protein extraction reagent (M-PER, Pierce) supplemented with protease inhibitors (Halt Protease Inhibitor Cocktail kit, Pierce). Cell lysates were loaded into 7.5% SDS-polyacrylamide gels. After electrophoresis, the separated proteins were electrotransblotted onto polyvinylidene difluoride membranes (Immobilon-P membrane, Millipore). After blocking, membranes were probed with antihuman DPD polyclonal rabbit antibody and antihuman TS monoclonal mouse antibody (gifts from Taiho Pharmaceutical). The proteins were visualized using horseradish peroxidase–conjugated antibodies (Pierce) followed by enhanced chemiluminescence (Pierce). The intensity of luminescence was quantified using an image analysis system (LAS-1000, Fuji Film).

DPD activity was measured using [6-¹⁴C]5-FU as a substrate (32, 33). Cells were homogenized and centrifuged at 105,000 × g for 60 min at 4°C, and the supernatant was used for assay. A reaction mixture containing 10 mmol/L potassium phosphate (pH 8.0), 0.5 mmol/L EDTA, 0.5 mmol/L 2-mercaptoethanol, 2 mmol/L DTT, 5 mmol/L MgCl₂, 20 μmol/L [6-¹⁴C]5-FU, 0.1 mmol/L NADPH, and 25 μL of cell extract in a total volume of 50 μL was incubated at 37°C for 30 min. After chemical hydrolyzation and neutralization using KOH and HClO₄, a 5-μL aliquot was applied to a TLC plate (2.5 × 20 cm, silica gel 60 F₂₅₄ plate, Merck) and developed with a mixture of ethanol and 1 mol/L ammonium acetate (5:1, v/v) and diethylether, acetone, chloroform, and water (50:50:40:1, v/v). DPD activity was determined as the sum of the products converted from 5-FU (i.e., dihydrouracil, 2-fluoro-β-ureidopropionic acid, and 2-fluoro-β-alanine) that were visualized and quantified with an imaging analyzer (BAS-2000, Fujix).

The sensitivities of three HCC lines, HAK-1A, KYN-2, and KYN-3, to separately administered 5-FU and IFN-α were determined as IC₅₀ values. KYN-3 was the most resistant to 5-FU of the three HCC lines. The IC₅₀ values of IFN-α in HAK-1A, KYN-2, and KYN-3 cells were 720, 510, and 24 IU/mL, respectively. KYN-3 cellular sensitivity to IFN-α was therefore approximately 30-fold and 20-fold higher than in HAK-1A and KYN-2 cell lines, respectively. The IC₅₀ value of IFN-α in HepG2 cells established from human hepatoblastoma and widely used in experiments was found to be over 10,000 IU/mL (data not shown).

Sensitivity was then compared with mRNA levels of DPD and TS, and DPD activity (as conversion rate from [6-¹⁴C]5-FU to its metabolites) in the three cell lines, which were shown to be broadly comparable (Table 1). DPD mRNA levels relative to those of KYN-3 cells, taken as 100%, and those of TS relative to HAK-1A, taken as 100%, are shown in Table 1. The cellular level of TS mRNA in KYN-2 cells was much lower than in the other two lines. Basal DPD activity in KYN-3 cells was approximately 5.5-fold and 9.1-fold higher than in HAK-1A and KYN-2 cell lines, respectively.

Dose-response curves of 5-FU alone and in combination with various concentrations of IFN-α are shown in Fig. 1A to C. Because sensitivities to IFN-α and 5-FU alone were different for each HCC cell line, RCs to the IC₅₀ value were used. As the concentration of combined IFN-α was increased, the dose-response curves of 5-FU were shifted down in an IFN-α concentration-dependent manner in all three HCC cell lines. For instance, following cotreatment with 312 IU/mL (RC = 0.45) IFN-α,the dose-response curve of HAK-1A to 5-FU was significantly shifted down and the 5-FU IC₅₀ value of 2.3 μmol/L was reduced to 0.28 μmol/L. However, there were clear differences between the three HCC cell lines. For KYN-3 cells, after treatment with the equivalent RC, 0.42 (8.0 IU/mL) IFN-α,the dose-response curve of 5-FU was significantly shifted down, but the 5-FU IC₅₀ value of 9.6 μmol/L was only reduced to 5.5 μmol/L.

An enhancement factor (EF) was defined to evaluate synergism between 5-FU and IFN-α, based on a 50% antiproliferative effect, as EF = 1 / [(RC of IFN-α) + (RC of 5-FU)]. When EF is 1, this combined effect is additive; values >1 or <1 imply synergistic or antagonistic effects, respectively. EF values of HAK-1A and KYN-2 were >1 at almost all combined doses, but EF values of KYN-3 at all tested combined doses were closer to 1. The combined effect of 5-FU and IFN-α on HAK-1A and KYN-2 cells was thus judged to be synergistic and that on KYN-3 to be additive, consistent with our previous study (27).

The synergism of IFN-α and 5-FU at various fractional efficacies was then quantified using the CI methods of Chou and Talalay (31). The antiproliferative effects of 5-FU on HAK-1A and KYN-2 cells were synergistically enhanced by IFN-α, indicated by CI values of <1 at most fractional effects (Fig. 2A and B). By contrast, the CI value in KYN-3 cells was >1 for fractional effects of 0.7 and under (Fig. 2C). The combined effect of IFN-α and 5-FU against KYN-3 thus seems to be antagonistic, except at very high fractional effect levels. These data show that the degree of synergism between IFN-α and 5-FU against KYN-3 cells (with a higher CI value) is weaker than against HAK-1A and KYN-2 cells over a large range of fractional effects.

Cellular mRNA levels of DPD and TS were examined following treatment with or without 500 IU/mL IFN-α for 24 h (Fig. 3A and B). IFN-α treatment resulted in an ∼7.5-fold increase in DPD mRNA levels in KYN-3 cells while reducing DPD mRNA levels in HAK-1A and KYN-2 cells to approximately one quarter and one third of that in IFN-α untreated cells, respectively (Fig. 3A). TS mRNA levels in all three HCC cell lines decreased after treatment with IFN-α to ∼60% of those in untreated cells (Fig. 3B).

Cellular protein levels of DPD and TS were also determined by Western blot analysis. The intensities of blotted bands were normalized by that of β-actin, and fold increase was measured as the relative intensity to that of untreated cells, taken as 1.0. Protein extracts of 10 μg/lane were loaded for KYN-3 cells, and 50 μg/lane were loaded for both HAK-1A and KYN-2 cells (Fig. 4). DPD protein levels were roughly comparable with mRNA levels and DPD activity in all three cell lines. Expression of DPD protein, molecular weight of 110,000, was down-regulated in HAK-1A and KYN-2 cells when treated with over 100 IU/mL IFN-α for 48 h. Conversely, that in KYN-3 cells was up-regulated after treatment with over 20 IU/mL IFN-α for 48 h in a concentration-dependent manner (Fig. 4). Expression of TS protein, molecular weight of 36,000, was down-regulated to a similar extent in a concentration-dependent manner in all three HCC cell lines when treated with IFN-α.

The combined antiproliferative effects of 5-FU and IFN-α can be modulated by treatment with a DPD competitive inhibitor, CDHP. We first examined the effect of CDHP on DPD activity in three HCC cell lines. KYN-3 cell extracts showed relatively high DPD activity of ∼15 pmol/min/mg protein (consistent with Table 1 findings). By contrast, DPD activity in HAK-1A and KYN-2 cells was very low with a marginal limit of evaluation for enzyme activity. DPD activity in KYN-3 cells was reduced to 40%, 7%, and 5% of basal activity after treatment with 0.7, 7.0, and 70 μmol/L CDHP, respectively (Fig. 5). There seemed to be almost no marked inhibition of DPD activity by CDHP in the other two cell lines, possibly due to their low enzyme activities.

We next examined whether cotreatment of CDHP could modulate the antiproliferative effect of 5-FU alone and in combination with IFN-α in three HCC cell lines. As seen in Fig. 6A to C, the dose-response curves of 5-FU alone in HAK-1A, KYN-2, and KYN-3 cells and those of 5-FU combined with IFN-α in HAK-1A and KYN-2 cells were not shifted by cotreatment with CDHP. None of these differences were significant at all data points. By contrast, there seemed a marked and significant shift of the 5-FU and IFN-α combined dose-response curve in KYN-3 cells to CDHP: the 5-FU IC₅₀ value of 4.6 μmol/L was reduced to 1.1 μmol/L by cotreatment with CDHP in the presence of IFN-α (Fig. 6C).

Consistent with the findings of our recent study (27), the combination of 5-FU and IFN-α showed a synergistic antiproliferation effect on two HCC cell lines, HAK-1A and KYN-2, and an additive or antagonistic antiproliferation effect on KYN-3 cells (Figs. 1 and 2). Expression of type 1 IFN receptor was specifically up-regulated by exposure to 5-FU in both HAK-1A and KYN-2, but not KYN-3 cells, suggesting that the modulation of IFN receptor expression by 5-FU could play a pivotal role in therapeutic efficacy (27). In this study, we further examined the molecular events underlying the antiproliferative effects of 5-FU and IFN-α. One of the major mechanisms of antiproliferative activity of 5-FU is the inhibition of TS activity with formation of the ternary complex of FdUMP, TS, and 5,10-methylenetetrahydrofolate. However, we observed a marked IFN-α−induced decrease in TS expression at similar levels in all three cell lines, suggesting that modulation of TS expression itself might not be directly involved in the absence or presence of synergism by the combination of 5-FU and IFN-α.

DPD is a key enzyme involved in 5-FU inactivation, which modulates FdUMP levels and controls formation of the ternary complex. Several clinical studies have shown that the intratumoral expression level of DPD is closely associated with clinical responses to 5-FU in patients with colorectal cancer (1), gastric cancer (34), and non–small cell lung cancer (35). In our present study, expression of DPD protein and mRNA levels in KYN-3 cells were specifically increased 5-fold or more over the basal level after exposure to IFN-α (Figs. 3 and 4). By contrast, down-regulation of DPD by IFN-α was observed in both HAK-1A and KYN-2.

In these two HCC cell lines, we previously proposed that 5-FU–induced up-regulation of the IFN receptor was the main mechanism underlying the synergistic antiproliferative effect of 5-FU and IFN-α (27). Moreover, down-regulation of DPD by IFN-α in these two cell lines might be involved in the synergistic effect. By contrast, up-regulation of DPD by IFN-α might account for the antagonism between IFN-α and 5-FU in KYN-3 cells.

Shestopal et al. (36) reported that 5′ flanking region of DPYD gene lacks the canonical TATA and CCAAT boxes, however, contains several cis-acting regulatory elements including binding sites for activator protein-2, nuclear factor-κB, Sp1, and Egr families. About the regulatory mechanism for IFN-α modulation of DPYD expression, we analyzed the sequence of 1.2 kb of 5′ flanking region of DPYD and found that this region contains two putative consensus binding sites for signal transducer and activator of transcription families (data not shown), suggesting that gene expression of DPYD is highly susceptible to IFN-α, which strongly activates signal transducer and activator of transcription 1 and signal transducer and activator of transcription 2 through the Janus-activated kinase-signal transducer and activator of transcription pathway. However, it remains unclear why DPYD expression is differentially controlled by IFN-α in HCC cell lines. Further elucidation of this differential regulatory mechanism at the molecular basis is required.

Our present study showed that cotreatment with the DPD inhibitor CDHP further synergistically enhanced the antiproliferative effect of 5-FU and IFN-α in KYN-3 cells only and not in HAK-1A and KYN-2 cells (Fig. 6). The antiproliferative effect of 5-FU alone was only slightly altered by CDHP cotreatment in KYN-3 cells, if any, and not at all in the other two cell lines (Fig. 6). This IC₅₀ reduction, however, was not statistically significant. Basal DPD activity and expression levels were much higher in KYN-3 cells than in HAK-1A and KYN-2 cells. Moreover, cellular DPD levels were specifically up-regulated >5-fold in KYN-3 cells by IFN-α, but this was not observed in HAK-1A and KYN-2 cells. It was presumed that 15 pmol/min/mg protein of basal DPD activity in KYN-3 cells was up-regulated to 80 to 110 pmol/min/mg protein. Taken together, this suggests that a relatively high DPD activity might be more susceptible to inhibition by CDHP, resulting in an apparent synergistic effect of CDHP on the antiproliferative effect by 5-FU and IFN-α.

Certain levels of DPD in cancer cells could be sensitive to CDHP-induced inhibition. A relevant study by Takechi et al. (37) showed that the antiproliferative activity of 5-FU could be markedly enhanced by cotreatment with 69 μmol/L CDHP in two human tumor cell lines with relatively high DPD activities, approximately 101 and 153 pmol/min/mg protein, respectively, but not those with low enzyme activity, 33 pmol/min/mg protein. However, this plausible mechanism why CDHP did induce synergism of 5-FU and IFN-α against only KYN-3 cells requires further study to validate these findings.

CDHP has been applied as a modulator in the newly developed antimetabolite TS-1 (Taiho Pharmaceutical). TS-1 consists of tegafur, CDHP, and potassium oxonate in a molar ratio of 1:0.4:1 (38). Potassium oxonate is a competitive inhibitor for orotate phosphoribosyltransferase that activates 5-FU. Potassium oxonate is mainly distributed in the gastrointestinal tract after p.o. administration and prevents gastrointestinal toxicity induced by 5-FU without reducing 5-FU activity in tumor (38, 39). In Japan, TS-1 has been used to treat patients with gastric, head and neck, and pancreatic cancers and shows potent therapeutic efficacy against gastric tumors, with a response rate of 46.5% (40–42). Nakamura et al. (43) applied a new combination regimen of TS-1 and IFN-α to advanced HCC patients with portal vein thrombus and multiple pulmonary metastases and observed some improvement in therapeutic efficacy. The combination of TS-1 and IFN-α could therefore be effective against patients with advanced HCC. However, the side effects of these combination therapies should be seriously considered before their implementation. Further studies are required to determine how DPD could be differentially controlled between normal cells including normal hepatic cells and malignant hepatic cells in patients with HCC when DPD inhibitory drugs are introduced.

We thank T. Kobunai, H. Tsujimoto, J. Chikamoto, Y. Fukui, Dr. Y. Basaki, and Dr. M. Kiniwa for technical support and fruitful discussion and N. Shinbaru for editorial help.