Adenosine kinase is a key determinant for the anti-HCV activity of ribavirin

Authors


  • See Editorial on Page 1203

Address reprint requests to: Nobuyuki Kato, Ph.D., Department of Tumor Virology, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, 2-5-1 Shikata-cho, Okayama 700-8558, Japan. E-mail: nkato@md.okayama-u.ac.jp; fax: (81)86-235-7392.

Abstract

Ribavirin (RBV) is often used in conjunction with interferon-based therapy for patients with chronic hepatitis C. There is a drastic difference in the anti–hepatitis C virus (HCV) activity of RBV between the HuH-7-derived assay system, OR6, possessing the RBV-resistant phenotype (50% effective concentration [EC50]: >100 µM) and the recently discovered Li23-derived assay system, ORL8, possessing the RBV-sensitive phenotype (EC50: 8 µM; clinically achievable concentration). This is because the anti-HCV activity of RBV was mediated by the inhibition of inosine monophosphate dehydrogenase in RBV-sensitive ORL8 cells harboring HCV RNA. By means of comparative analyses using RBV-resistant OR6 cells and RBV-sensitive ORL8 cells, we tried to identify host factor(s) determining the anti-HCV activity of RBV. We found that the expression of adenosine kinase (ADK) in ORL8 cells was significantly higher than that in RBV-resistant OR6 cells harboring HCV RNA. Ectopic ADK expression in OR6 cells converted them from an RBV-resistant to an RBV-sensitive phenotype, and inhibition of ADK abolished the activity of RBV. We showed that the differential ADK expression between ORL8 and OR6 cells was not the result of genetic polymorphisms in the ADK gene promoter region and was not mediated by a microRNA control mechanism. We found that the 5' untranslated region (UTR) of ADK messenger RNA in ORL8 cells was longer than that in OR6 cells, and that only a long 5' UTR possessed internal ribosome entry site (IRES) activity. Finally, we demonstrated that the long 5' UTR functioned as an IRES in primary human hepatocytes. Conclusion: These results indicate that ADK acts as a determinant for the activity of RBV and provide new insight into the molecular mechanism underlying differential drug sensitivity. (Hepatology 2013;58:1236–1244)

Abbreviations
Abs

antibodies

ADK

adenosine kinase

5azaC

5-azacytidine

CC50

50% cytotoxic concentration

cDNA

complementary DNA

CHC

chronic hepatitis C

EC50

50% effective concentration

GTP

guanosine triphosphate

HCV

hepatitis C virus

HPLC

high-performance liquid chromatography

IMPDH

inosine monophosphate dehydrogenase

IMP

inosine-5'-monophosphate

IRES

internal ribosome entry site

kb

kilobase

mRNA

messenger RNA

NS

nonstructural protein

nt

nucleotide

ORF

open reading frame

4-PBA

4-phenylbutyric acid

Peg-IFN

pegylated-interferon

PHHs

primary human hepatocytes

RACE

rapid amplification of cDNA ends

RBV

ribavirin

RL

renilla luciferase

RMP

RBV 5'-monophosphate

RT-PCR

reverse transcription-polymerase chain reaction

siRNA, small interfering RNA

miRNAs, microRNAs

SNP

single-nucleotide polymorphism

SVR

sustained virologic response

UTR

untranslated region.

Hepatitis C virus (HCV) is an enveloped RNA virus, the genome of which consists of a positive-stranded 9.6-kilobase (kb) RNA encoding 10 structural and nonstructural (NS) proteins.[1] The combination of pegylated-interferon (Peg-IFN) and ribavirin (RBV) was the standard treatment for patients with chronic hepatitis C (CHC) until last year, when a new triple-agent combination therapy using an inhibitor of HCV NS3-4A protease (i.e., either telaprevir or boceprevir), in combination with Peg-IFN and RBV, was started.[2] The sustained virologic response (SVR) rate of genotype 1 using this new therapy is expected to increase from 55% to more than 70%.[3] However, there has also been an increase in side effects by RBV in the triple therapy, including several severe side effects, such as skin rash by telaprevir, ageusia by boceprevir, and advanced anemia by telaprevir/boceprevir.[3, 4]

The main hurdle to resolving the side-effect profile is that the anti-HCV mechanism of RBV is not well understood, although several possible mechanisms have been proposed.[5, 6] To date, there has been no cell-culture system enabling analysis of the anti-HCV mechanism of RBV at clinically achievable concentrations (5-14 µM), because the human hepatoma cell line, HuH-7, which has been the only cell line available for robust HCV replication, is not sensitive to RBV.[5, 7, 8] Indeed, we also observed that the 50% effective concentration (EC50) of RBV against HCV RNA replication in our developed HuH-7-derived assay system (OR6), in which the genome-length HCV RNA (O strain of genotype 1b) encoding renilla luciferase (RL) replicates efficiently, was more than 100 µM, and 50% cytotoxic concentration (CC50) was also more than 100 µM.[9, 10]

On the other hand, we recently found that a new human hepatoma cell line, Li23, whose gene expression profile was distinct from that of HuH-7, enabling efficient HCV RNA replication and persistent HCV production, was sensitive to RBV.[10-12] Indeed, the EC50 value of RBV against HCV RNA replication in our developed Li23-derived assay system (ORL8), which is comparable to the OR6 assay system, was 8.7 µM, and the CC50 value was more than 100 µM.[10] It was noteworthy that this EC50 value was equivalent to the clinically achievable concentrations of RBV. Therefore, this finding led us to analyze the anti-HCV mechanism of RBV, and, consequently, we found that the anti-HCV activity of RBV was mediated by the inhibition of inosine monophosphate dehydrogenase (IMPDH), and that IMPDH was required for HCV RNA replication.[10]

From these findings, we anticipated that the comparative analysis of RBV-sensitive ORL8 cells and RBV-resistant OR6 cells would lead to the identification of host factor(s) determining the anti-HCV activity of RBV. Here, we report the finding that adenosine kinase (ADK) is an essential determinant of the anti-HCV activity of RBV.

Materials and Methods

Cell Cultures

HuH-7- and Li23-derived cells and PH5CH8 cells were maintained as described previously.[11] HT17 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Primary human hepatocytes (PHHs; PhoenixBio, Higashihiroshima, Japan) were also maintained in the medium for the Li23-derived cells.

Reagents

RBV was kindly provided by Yamasa (Chiba, Japan).

Inosine-5'-monophosphate (IMP) and nucleoside triphosphates (cytidine triphosphate, uridine triphosphate, adenosine triphosphate, and guanosine triphosphate [GTP]) were also purchased from Yamasa. ABT-702 was purchased from Calbiochem (San Diego, CA). 5-azacytidine (5azaC) and 4-phenylbutyric acid (4-PBA) were purchased from Sigma-Aldrich (St. Louis, MO).

Western Blotting Analysis

Preparation of cell lysates, sodium dodecyl sulfate polyacrylamide gel electrophoresis, and immunoblotting analysis were performed as previously described.[13] Polyclonal-ADK (ab54818; Abcam, Cambridge, MA), monoclonal-ADK (F-5; Santa Cruz Biotechnology, Santa Cruz, CA), and β-actin (AC-15; Sigma-Aldrich) antibodies (Abs) were used.

Reverse-Transcription Polymerase Chain Reaction

Reverse-transcription polymerase chain reaction (RT-PCR) was performed to detect ADK messenger RNA (mRNA), as described previously,[14] using the primer sets (ADKF and ADKR; ADK-5'-untranslated region [UTR]-187nts and ADK-5'-UTR checkR) listed in Supporting Table 1.

Quantitative RT-PCR

Quantitative RT-PCR analysis for ADK mRNA was performed using a real-time LightCycler PCR (Roche Diagnostics, Indianapolis, IN), as described previously,[11] with the primer sets (ADKF and ADKR; ADK-5'UTR-384nts and ADK-5'UTR checkR; ADK-5'UTR-318nts and ADK-5'UTR checkR; ADK-5'UTR-187nts and ADK-5'UTR checkR; ADK-5'UTR-125nts and ADK-5'UTR checkR) listed in Supporting Table 1.

RL Assay

RL assay was performed as described previously.[9] Experiments were performed at least in triplicate.

High-Performance Liquid Chromatography Analysis

Quantitative high-performance liquid chromatography (HPLC) analysis was performed using the extract from the OR6 or ORL8 cells treated with 50 µM of RBV for 8 hours. HPLC analysis was performed as described previously.[15]

RNA Interference

Small interfering RNA (siRNA) duplexes targeting the coding regions of human ADK (catalog no.: M-009687-01; Dharmacon, Inc., Lafayette, CO) were chemically synthesized. A nontargeting siRNA duplex (catalog no.: D-001206-13; Dharmacon) was also used as a control. ORL8 cells were transfected with the indicated siRNA duplexes using Oligofectamine (Invitrogen, Carlsbad, CA).[10]

Ectopic Expression of ADK

The methods of plasmid construction for ectopic expression of ADK and retroviral infection using the constructed plasmids are described in the Supporting Materials.

Plasmid Construction and Internal Ribosome Entry Site Activity Assay

The method of plasmid construction for internal ribosome entry site (IRES) activity assay is described in the Supporting Materials. The dual luciferase reporter assay for IRES activity was performed by the method described previously.[14]

Statistical Analysis

Data are presented as means ± standard deviation. The Student unpaired t test was performed for statistical analysis between the two groups, and the difference was considered significant at P < 0.05.

Results

High Expression Level of ADK in ORL8 Cells

To identify the host factor responsible for the difference in RBV responses between Li23-derived ORL8 and HuH-7-derived OR6 cells, we first recompared the previous data from complementary DNA (cDNA) microarrays using Li23 and HuH-7 cells. Although we assigned 17 genes that showed dramatic differences in expression between Li23 and HuH-7 cells,[12] none of these genes were considered to be involved in the response to RBV. Expression of IMPDH1 (NM_000883) and IMPDH2 (NM_000884), which were involved in the anti-HCV mechanism of RBV,[8, 10] was also at a similar level between Li23 and HuH-7 cells or between cured ORL8 (ORL8c) and cured OR6 (OR6c) cells (Supporting Table 2).

We next attempted to verify the process of GTP reduction that is expected to occur after RBV is incorporated into cells. To this end, we performed a quantitative HPLC analysis using the extract from the OR6 or ORL8 cells treated with 50 µM of RBV for 8 hours, which is the working time of RBV against HCV RNA replication.[10] Amounts of IMP and GTP were calculated from the peak area obtained by HPLC analysis. Volume of cells was calculated from the mean diameter of cells, and we found 106 cells to be equivalent to 1.1 mm3. We assumed that the extracted nucleotides (nts) were uniformly distributed in the cell aqueous volume. As expected, the level of intracellular GTP in ORL8 cells showed a significant (60%) reduction, whereas that in OR6 cells showed only a 27% reduction (Fig. 1A and Supporting Fig. 1A-D). These results support our previous finding that the inhibitory effect of RBV on HCV RNA replication in ORL8 cells is stronger than that in OR6 cells.[10]

Figure 1.

Expression level of ADK in ORL8 cells was higher than that in OR6 cells. (A) Signal of GTP obtained by HPLC analysis using nts extracted from RBV-treated cells was quantified. (B) Signal of IMP obtained by HPLC analysis using nts extracted from RBV-treated cells was quantified. (C) Expression level of ADK mRNA in ORL8 cells was compared with that in OR6 cells by quantitative RT-PCR analysis. (D) Level of ADK in ORL8 cells was compared with that in OR6 cells by western blotting analysis. (E) ORL8 cell extract was diluted and then western blotting analysis was performed for the detection of ADK. Two isoforms of ADK (40 and 38 kDa) were loaded as molecular markers. Experiments (A-C) were performed in triplicate. *P < 0.05; NS, not significant.

In addition, we noticed an unexpected phenomenon: A substantial accumulation of IMP occurred as the result of IMPDH inhibition in ORL8 cells, but not in OR6 cells. The IMP level in ORL8 cells became approximately 30 times higher than that in OR6 cells (Fig. 1B and Supporting Fig. 1A-D). However, no additive effect of inosine (up to 100 µM) on HCV RNA replication in ORL8 cells was observed (Supporting Fig. 2).

It has been reported that RBV is metabolized in vivo through RBV 5'-monophosphate (RMP), a competitive inhibitor of IMPDH, by ADK.[16] Based on our findings, we expected that ADK activity might be able to control the anti-HCV activity of RBV. Indeed, microarray analysis revealed that the actual expression levels of ADK were 764 and 2,840 in OR6c and ORL8c cells, respectively. Quantitative RT-PCR analysis also showed that the mRNA level of ADK in ORL8 cells was 4.5 times higher than that in OR6 cells (Fig. 1C). Furthermore, we found that the protein level of ADK in ORL8 cells was much higher than that in OR6 cells (Fig. 1D).

On the other hand, it is known that ADK has two major isoforms: ADK-long (NM_006721) localized in the nucleus and ADK-short (NM_001123) localized in the cytoplasm.[17] ADK-long differs in the 5' UTR and initiates translation at an alternative start codon, compared to ADK-short. ADK-long is 17 amino acids longer than ADK-short. We prepared ORL8 cells stably overexpressing ORL8-derived ADK-long or ADK-short using a retroviral gene transfer system and examined its mobility in western blotting analysis. Fortunately, two isoforms were discriminable as 40 (ADK-long) and 38 kDa (ADK-short) (Fig. 1E). Using these isoforms as molecular markers, we performed semiquantitative western blotting analysis by the sample dilution method. The results revealed that the expression level of ADK-short in ORL8 cells was approximately 16 times higher than that in OR6 cells, and that ADK-long was little expressed in both cells (Fig. 1E). From these results, we assumed that the differences in ADK expression were involved in the dramatic differences in RBV sensitivity between the two cell lines. To address this assumption, we focused on the ADK-short in the following study; hereafter, ADK-short is designated as ADK.

ADK Is a Host Factor Determining the Anti-HCV Activity of RBV

To evaluate the hypothesis that ADK controls the anti-HCV activity of RBV, we first examined the effect of ABT-702, an ADK inhibitor, on the anti-HCV activity of RBV. The results revealed that ABT-702 cancelled the activity of RBV in ORL8 cells in a dose-dependent manner (Fig. 2A). Furthermore, we demonstrated that the activity of RBV was cancelled in ADK-knockdown ORL8 cells (Fig. 2B). These results suggest that the inhibition of ADK in ORL8 cells converts them from an RBV-sensitive phenotype to an RBV-resistant phenotype.

Figure 2.

ADK is a determining host factor for the anti-HCV activity of RBV. (A) ORL8 cells were cotreated with RBV (50 µM) and ABT-702 (nM) for 72 hours, after which an RL assay was performed. Relative luciferase activity (RLU) (%) calculated at each time point, when the level of luciferase activity in nontreated cells was assigned to be 100%, is shown. (B) ORL8 cells were transfected with 8 nM of siRNA targeting ADK. After 72 hours, expression levels of ADK were monitored by western blotting analysis (lower panel). ADK-knockdown ORL8 cells were treated with 12.5 µM of RBV for 72 hours, after which an RL assay was performed, as described in (A, upper panel). (C) Expression level of ADK in OR6-ADK cells was monitored by western blotting analysis. (D) OR6-ADK cells were treated with RBV for 72 hours and then an RL assay was performed as described in (A). (E) OR6-ADK cells were cotreated with RBV (5 µM) and ABT-702 (nM) for 72 hours and then an RL assay was performed, as described in (A). Experiments (A, B, D, and E) were performed in triplicate. *P < 0.05.

To directly demonstrate the involvement of ADK, we first prepared OR6 cells stably expressing ADK (OR6-ADK) (Fig. 2C). We were able to demonstrate that the OR6-ADK cells were dramatically converted from an RBV-resistant phenotype with an EC50 value of more than 100 µM to an RBV-sensitive phenotype with an EC50 value of 2.6 µM (Fig. 2D). We next examined whether or not the GTP reduction or IMP accumulation observed in ORL8 cells treated with RBV (Fig. 1A,B) occurs in OR6-ADK cells. The results revealed that the GTP reduction and IMP accumulation in RBV-treated OR6-ADK cells were more pronounced than in RBV-treated ORL8 cells (Supporting Fig. 3A,B). Because OR6 is a clonal cell line harboring genome-length HCV RNA, we used a polyclonal cell line (sOR) harboring HCV replicon RNA[9] to prepare sOR-ADK cells stably expressing ADK (Supporting Fig. 3C) and examined their sensitivity to RBV. sOR-ADK cells were also dramatically converted from an RBV-resistant phenotype with an EC50 value of more than 100 µM to an RBV-sensitive phenotype with an EC50 value of 6.0 µM (Supporting Fig. 3D). In addition, ORL8-ADK cells stably overexpressing ADK also showed EC50 values ranging from 13.2 to 1.2 µM (Supporting Fig. 3E). Furthermore, we demonstrated that the anti-HCV activity detected in OR6-ADK cells was also cancelled by ABT-702 treatment in a dose-dependent manner (Fig. 2E). Considering these results together, we conclude that ADK is a key determinant for the anti-HCV activity of RBV.

The Suppression of ADK Expression in OR6 Cells Was Not the Result of Genetic Variations or Epigenetic Alterations in the ADK Gene Promoter

To clarify the mechanism underlying the difference in ADK expression between OR6 and ORL8 cells, we first examined the nt sequences of up to several kb upstream from the transcription start point estimated from NM_001123 (31-OCT-2010) using the data of AL731576. Several possible transcription elements, such as the GC box (−12 and −187 of ADK gene), p53 response element (−252 and −585), and heat shock element (−559, −971, −1486, and −1797) were detected in up to approximately 2 kb upstream from the estimated transcription start point, but not in more 2 kb. Accordingly, we amplified approximately 2 kb including the 5' UTR (187 nts estimated by NM_001123 [31-OCT-2010]) by PCR using DNA prepared from ORL8 or OR6 cells, and each PCR product was inserted into pGL4.10-luc2 for the sequence analysis and reporter analysis of gene promoter activity. Sequence analysis confirmed that the sequences of the inserts were the same as the sequence data of the ADK gene (AL731576), except in the case of a single-nucleotide polymorphism (SNP) [rs10824095; C for ORL8 cells and T for OR6 cells] located 20 bases upstream from the initiation codon. Luciferase reporter assay using ORL8c cells revealed that the promoter activity of OR6 origin was almost equal to that of ORL8 origin (Supporting Fig. 4A), indicating that the detected SNP was not involved in the level of promoter activity.

We next evaluated the epigenetic effects on ADK expression level. The results revealed that the expression level of ADK mRNA in OR6 cells was not enhanced in the cells treated with 5azaC and/or 4-PBA for 48 hours (Supporting Fig. 4B). Moreover, the protein level of ADK was not increased in the OR6 cells treated with 5azaC for 6 days (Supporting Fig. 4C). Taken together, these results suggest that the low level of ADK mRNA in OR6 cells was not the result of genetic polymorphisms or epigenetic alternations in the ADK gene promoter region.

The Differential ADK Expression Between OR6 and ORL8 Cells Was Not Mediated by a microRNA Control Mechanism

To explain the above-described gap between the 4.5-fold difference in the mRNA level and the 16-fold difference in the protein level (Fig. 1C,E), we hypothesized that the 3' UTR of ADK mRNA was different in the length or nt sequences between OR6 and ORL8 cells, and that such differences affected the control mechanism by microRNA (miRNA). To test this hypothesis, we first performed 3' rapid amplification of cDNA ends (RACE) analysis on ADK mRNA using total RNA prepared from OR6 or ORL8 cells. Sequence analysis using more than 45 cDNA clones obtained from each cell line was carried out. 3' UTRs of four different lengths were detected in both OR6 and ORL8 cells, because four potential poly(A) additional signals were present in the downstream ADK open reading frame (ORF) (Supporting Fig. 5). The results revealed no qualitative difference of 3' UTR species between OR6 and ORL8 cells (Supporting Fig. 5).

Because the 3' UTR of ADK mRNA contained the seed sequences of miR-182, miR-203, mir-125a-3p, and miR-106b (Supporting Fig. 5), we assumed that the difference in expression levels of these miRNAs causes the different protein levels of ADK. To examine this possibility, we performed an miRNA microarray analysis between OR6 and ORL8 cells. This analysis revealed very low expression levels (measured values of less than 7) of miR-182, miR203, and miR-125a-3p in both cell lines. Although only miR-106b was moderately expressed (measured value of approximately 300) in OR6 and ORL8 cells, the values obtained from both cell lines were almost the same. From these results, these miRNAs may not participate in the translational regulation of ADK mRNAs in OR6 and ORL8 cells.

The 5' UTR of ADK mRNA in ORL8 Cells Was Longer Than That in OR6 Cells

To explain the dramatic difference in ADK expression between ORL8 and OR6 cells, we next focused on the 5' UTR. To date, two different lengths of 5' UTR (384 nts in accession number NM_001123[25-MAR-2011] and 187 nts in accession numbers NM_001123[31-OCT-2010] and HSU_50196) have been deposited in GeneBank. Because the 384 nts form has been considered to be a unique species in testis tissue, we performed 5' RACE analysis to determine the length of the 5' UTR of ADK mRNA in ORL8 or OR6 cells. Sequence analysis was carried out using more than 20 cDNA clones obtained from each cell line. Consequently, we obtained 319 and 125 nts as the major 5' UTR species in ORL8 and OR6 cells, respectively. We confirmed these results by RT-PCR analysis using four different primer sets for the 5' UTR (Fig. 3A). The amount of 384 nts species in ORL8 cells was estimated to be less than one thirtieth the amount of the 319-nts species (Fig. 3A). These results indicate that the length of 5' UTR in ORL8 cells is longer than that in OR6 cells.

Figure 3.

Level of ADK mRNA possessing long 5' UTR was correlated with the expression level of ADK. (A) Total RNAs prepared from ORL8 and OR6 cells were subjected to RT-PCR using the primer sets (Supporting Table 1) for various lengths of 5' UTR of ADK mRNA. (B) Expression levels of ADK were compared by western blotting analysis. The two molecular markers of ADK shown in Fig. 1E were loaded on every two lanes. (C) Amounts of 5' UTR species of ADK mRNAs were compared by quantitative RT-PCR analysis using the primer sets described in (A). Experiments were performed in triplicate.

From these results, we considered the possibility that the length of the 5' UTR is associated with the protein level of ADK. To test this possibility, we first compared the expression levels of ADK in various human hepatoma cell lines and human immortalized hepatocyte lines. Low expression level of ADK was observed in HT17 and Hep3B cells as well as OR6 cells, although the other cell lines, including ORL8, HuH-6, HepG2, HLE, and PH5CH8 cells, showed high expression level of ADK (Fig. 3B and Supporting Fig. 6). We next performed quantitative RT-PCR analysis on the 5' UTR using total RNAs from OR6, ORL8, HT17, and PH5CH8 cells. Consequently, we found that the 319 nts species of the 5' UTR was abundant in PH5CH8 cells, but not in HT17 cells (Fig. 3C), indicating good correlation between the amount of 319 nts species and the amount of ADK protein (Fig. 3B,C). These results suggest that the 319 nts species of 5' UTR is involved in the high protein level of ADK.

The Long-Form 5' UTR of ADK mRNA Possessed IRES Activity

From the results of 5' UTR analysis, we assumed that the 319 nts species of the 5' UTR possesses IRES activity because it is GC rich (72%) and highly structured (estimated ΔG = −110.7 kcal/mol), and because it contains an upstream ORF for 70 amino acids. To test this assumption, we used a bicistronic dual luciferase reporter assay system for the detection of IRES activity (Fig. 4A). As a positive control, we constructed a pGL4-based reporter plasmid containing HCV IRES (377 nts; 341 nts in the 5' UTR plus the first 36 nts in the Core-encoding region). We next replaced the HCV IRES structure in this plasmid with several different lengths (forward or reverse direction) of the 5' UTR derived from ORL8 cells. ORL8c cells were transfected with these plasmids, and at 48 hours after transfection, dual luciferase assays were performed. Consequently, we found that the forward 319 nts, but not the forward 125 nts, of 5' UTR clearly showed IRES activity at the same level as HCV IRES (Fig. 4B). The 187 nts species also showed weak IRES activity (Fig. 4B). None of the 5' UTR species with reverse direction and none of the HCV IRES with reverse direction showed any IRES activities (Fig. 4B). Furthermore, similar results were obtained in the genome-length HCV RNA-replicating OL8 cells and their cured cells (OL8c) (Supporting Fig. 7A,B), suggesting that IRES activity does not depend on cell strains or HCV RNA replication. In addition, we did not observe any effects of an SNP (rs10824095), which was located 20 bases upstream from the initiation codon, on the IRES activities of OR6 and ORL8 cell-derived 5' UTRs (319 nts) (Supporting Fig. 8).

Figure 4.

Long-form 5' UTR of ADK mRNA possessed IRES activity. (A) Partial structure of the plasmid used as a dicistronic dual reporter assay system. (B) ORL8c cells were transfected with the plasmid as shown in (A). After 48 hours, a dual luciferase assay was performed. The ratio of the RL activity to firefly luciferase activity was calculated. The relative value calculated at each sample, when the ratio in the control vector-transfected cells (−) was assigned to be 1, is presented. F and R indicate the forward and reverse direction of insert in the reporter plasmid, respectively. (C) Deletion mutant analysis of the 5' UTR in IRES assay. IRES assay was performed using ORL8c cells transfected with the reporter plasmid containing the deleted forms of the 5' UTR, as described in (B). (D) ADK expression derived from the long-form 5' UTR transcript was more productive than that from the short-form 5' UTR transcript. OR6c cells were transfected with the plasmid, in which the XbaI fragment of the plasmid used for HCV IRES activity assay was replaced by the ADK ORF possessing the 5' UTR of 319 or 125 nts. After 48 hours, western blotting analysis was performed. Firefly luciferase activities were measured to check equal transfection efficiency. Experiments (B and C) were performed in triplicate. *P < 0.05; NS, not significant.

To identify the entry site of the 40S ribosome in the IRES region, we prepared three deletion mutants (deleted upstream 30, 60, and 90 nts from the initiation codon) of the 5' UTR and measured their IRES activities in ORL8c cells. The results revealed that the deletion up to 60 nts from the initiation codon did not decrease IRES activity, but the 90 nts deletion abolished IRES activity (Fig. 4C). Similar results were also obtained in OL8 and OL8c cells (Supporting Fig. 7C,D). These results suggest that the entry site of the 40S ribosome is between 60 and 90 nts upstream from the initiation codon, and that the region from 319 to 61 nts upstream from the initiation codon is necessary for the IRES activity. It is noteworthy that this region forms a stable secondary structure (estimated ΔG = −108.4 kcal/mol) (Supporting Fig. 7E). Furthermore, we demonstrated that ADK expression derived from the long-form 5' UTR transcript was more productive than the expression from the short-form 5' UTR transcript in OR6c cells (Fig. 4D).

The Long-Form 5' UTR of ADK mRNA Functioned as an IRES in Primary Human Hepatocytes

To obtain a final conclusion, we examined whether the novel mechanism in ADK translation plays a role in PHHs. We first examined ADK expression level in PHHs, and the results revealed that ADK protein level was higher in PHHs than in ORL8 cells (Fig. 5A). We next performed RT-PCR analysis using the primer sets used in Fig. 3A to examine the amounts of 319 and 125 nts forms of the 5' UTR. The results showed that the 319 nts species was the major 5' UTR species in PHHs, but not in HuH-7 cells, which are the parent of OR6 cells (Fig. 5B), indicating a good correlation between the amount of 319 nts species and the amount of ADK protein in PHHs. Finally, we demonstrated that the 319 nts form, but not the 125 nts form, of 5' UTR clearly showed IRES activity in PHHs (Fig. 5C).

Figure 5.

Long-form 5' UTR of ADK mRNA functioned as an IRES in PHHs. (A) ADK expression level in PHHs was compared with those in OR6 and ORL8 cells by western blotting analysis. (B) Total RNAs prepared from PHHs and HuH-7 cells were subjected to RT-PCR analysis using the primer sets described in Fig. 3A. (C) PHHs were transfected with the plasmid shown in Fig. 4A. After 48 hours, a dual luciferase assay was performed as described in Fig. 4B. Experiments (B and C) were performed in triplicate. *P < 0.05; NS, not significant.

Considering all these results together, we conclude that not only ORL8 cells, but also PHHs express the long-form 5' UTR of ADK mRNA possessing IRES activity and then produce high levels of ADK, which works as an RBV kinase.

Discussion

In this study, we identified, for the first time, a host factor ADK whose expression level could control the anti-HCV activity of RBV. Furthermore, we found that the expression level of ADK was associated with the amount of ADK mRNA possessing long 5' UTR exhibiting IRES activity. This finding suggests that the RBV sensitivity on HCV RNA replication is regulated by the IRES-dependent translation of ADK mRNA. If ADK expression levels or activity differ between patients with CHC, it may be a useful therapeutic target.

It has recently been reported that a functional SNP (rs1127354; major C and minor A) in inosine triphosphatase was the most significant SNP associated with RBV-induced anemia.[18] In this context, we hypothesized that this SNP is associated with the expression level of ADK. To test this hypothesis, we examined the status of rs1127354 in ORL8 and PH5CH8 cells showing high expression levels of ADK and in OR6 and Hep3B cells showing low expression levels of ADK. The results revealed that all cell lines showed the major C of the SNP, suggesting that rs1127354 is not associated with the expression level of ADK.

The most striking highlight in this study is the IRES activity found in ADK mRNA. It has recently been reported that cellular IRES-mediated translation is activated by many physiological and pathological stress conditions in eukaryotic cells.[19] To achieve efficient IRES-dependent translation, some triggers will be needed. However, HCV RNA replication was not such a trigger, in the present study, because a similar level of IRES activity was observed in both OL8c cured cells and genome-length HCV RNA-replicating OL8 cells (Supporting Fig. 7A-D). The addition of adenosine did not act as a trigger for IRES (Supporting Fig. 9). Another possible explanation for the high level of ADK in ORL8 cells would be the involvement of one or more miRNA(s) in stabilizing the IRES-containing ADK mRNA, as reported in HCV RNA.[20] To test this possibility, we performed comparative miRNA microarray analysis using ORL8, PH5CH8, OR6, and HT17 cells. The results revealed that nts 1-8 of miR-424, whose expression levels in ORL8 and PH5CH8 cells were several times higher than those in OR6 and HT17 cells, showed base pairs in the nt 61-68 upstream initiation codon of ADK mRNA. It was noticed that this region in ADK mRNA overlaps the region (nt 60-90 upstream initiation codon of ADK mRNA) identified as the entry site of the 40S ribosome. However, a preliminary experiment showed that overexpression of miR-424 in ORL8 or OR6 cells did not enhance the translation of ADK (Supporting Fig. 10), suggesting that miR-424 is not associated with the high level of ADK in ORL8 cells. The possibility remains that other miRNA(s) participate in the up-regulation of ADK.

At this time, we have identified ADK as a host factor that controls the anti-HCV activity of RBV and clarified the molecular mechanism underlying regulation with ADK. Furthermore, we demonstrated that such a novel mechanism plays a role in PHHs. From our finding, we suggest that ADK expression is artfully regulated both at the transcription and translation stage. Although we identified ADK, which participates in nucleotidic metabolism, as an enzyme functionally controlled by the specific expression of an IRES-containing mRNA, there may be other gene products controlled by a similar mechanism.

Acknowledgments

The authors thank Naoko Kawahara, Takashi Nakamura, and Keiko Takeshita for their technical assistance.

Ancillary