Identification of 2′-Phosphodiesterase, Which Plays a Role in the 2-5A System Regulated by Interferon*

The 2-5A system is one of the major pathways for antiviral and antitumor functions that can be induced by interferons (IFNs). The 2-5A system is modulated by 5′-triphosphorylated, 2′,5′-phosphodiester-linked oligoadenylates (2-5A), which are synthesized by 2′,5′-oligoadenylate synthetases (2′,5′-OASs), inactivated by 5′-phosphatase and completely degraded by 2′-phosphodiesterase (2′-PDE). Generated 2-5A activates 2-5A-dependent endoribonuclease, RNase L, which induces RNA degradation in cells and finally apoptosis. Although 2′,5′-OASs and RNase L have been molecularly cloned and studied well, the identification of 2′-PDE has remained elusive. Here, we describe the first identification of 2′-PDE, the third key enzyme of the 2-5A system. We found a putative 2′-PDE band on SDS-PAGE by successive six-step chromatographies from ammonium sulfate precipitates of bovine liver and identified a partial amino acid sequence of the human 2′-PDE by mass spectrometry. Based on the full-length sequence of the human 2′-PDE obtained by in silico expressed sequence tag assembly, the gene was cloned by reverse transcription-PCR. The recombinant human 2′-PDE expressed in mammalian cells certainly cleaved the 2′,5′-phosphodiester bond of 2-5A trimer and 2-5A analogs. Because no sequences with high homology to this human 2′-PDE were found, the human 2′-PDE was considered to be a unique enzyme without isoform. Suppression of 2′-PDE by a small interfering RNA and a 2′-PDE inhibitor resulted in significant reduction of viral replication, whereas overexpression of 2′-PDE protected cells from IFN-induced antiproliferative activity. These observations identify 2′-PDE as a key regulator of the 2-5A system and as a potential novel target for antiviral and antitumor treatments.

Virus infection to cells induces the production and secretion of interferons (IFNs), 1 which play a major role as the first line host defense against pathogens. IFNs are multifunctional cytokines with important roles in antiviral activity, cell growth, cell differentiation, and immunomodulation (1,2). The 2-5A system is well known as one of the major pathways induced by IFNs, in which unusual oligoadenylates, referred to as 2-5A, modulate RNA degradation in cells (3). 2-5A comprises 5Јtriphosphorylated oligoadenylates containing adenosines linked with a 2Ј-5Ј-phosphodiester bond (pppA(2ЈpA) n , where n Ն 2), which is resistant to most known RNases (4,5). The 2-5A system is composed of three kinds of enzymes: 2Ј,5Јoligoadenylate synthetases (2Ј,5Ј-OASs) activated by doublestranded RNA (dsRNA), which produce 2-5A from ATP (6); 2Ј-phosphodiesterase (2Ј-PDE), which degrades 2-5A to AMP and ATP (7); and RNase L activated by 2-5A, which degrades RNA, resulting in inhibition of protein synthesis sometimes leading cells to apoptosis (8 -11). Because RNase L is expressed constitutively in a wide variety of cells, the amount of 2-5A is believed to be a primary factor that controls RNase L activity in the 2-5A system (3,12,13). 5Ј-Phosphatase is indicated to have a role in the 2-5A system because the 5Ј-unphosphorylated 2-5A core has significantly reduced ability to activate RNase L (9,14). Since complete inactivation of 2-5A is only accomplished by cleavage of the 2Ј-5Ј bond (9), which is resistant to most known RNases (4,5), cleavage of the 2Ј-5Ј bond by 2Ј-PDE appears to play an important role in inactivation of 2-5A. Viral infection of cells induces the secretion of IFNs which up-regulate 2Ј,5Ј-OASs severalfold to several hundredfold (3,6). Because viral infection frequently results in production of dsRNA, the amount of 2-5A in the infected cells is increased preferentially. As a result of RNA degradation by activated RNase L, the viral replication in the infected cells is suppressed by inhibition of protein synthesis and/or apoptosis (3). dsRNAdependent protein kinase (PKR). Mx protein, and other systems are also known as other mechanisms of antiviral action of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB115695.
Among enzymes in the 2-5A system, 2Ј,5Ј-OASs and RNase L have been molecularly cloned and studied well (3,6,8). Although characterization of 2Ј-PDE such as substrate specificity using a purified enzyme has been investigated (27,28), molecular characterization of 2Ј-PDE has remained to be elucidated. To understand the molecular mechanism and biological role of the 2-5A system more precisely, identification of 2Ј-PDE is indispensable. In this study, we partially purified 2Ј-PDE from bovine liver and molecularly cloned the human homologous protein. Overexpression of human 2Ј-PDE suppressed antiproliferative activity of IFNs and dsRNA. Moreover, suppression of 2Ј-PDE by a small interfering RNA (siRNA) and a 2Ј-PDE inhibitor exhibited antiviral activity. Our findings suggest 2Ј-PDE as a potential novel target for therapeutics against replication of viruses and development of tumors.

EXPERIMENTAL PROCEDURES
Purification and Identification of 2Ј-PDE from Bovine Liver-All purification steps were conducted at 4°C. A portion of bovine liver (200 g, Tokyo Shibaura Zouki) was homogenized in 700 ml of 10 mM MOPS, pH 7.4, containing 10 mM MgCl 2 , 40 mM NaCl, 0.2 mM dithiothreitol (DTT), and 0.1 mM phenylmethylsulfonyl fluoride. The homogenate was centrifuged at 15,000 ϫ g for 20 min, and the supernatant was centrifuged further at 100,000 ϫ g for 60 min. The 30 -50% saturated ammonium sulfate precipitates of the supernatant were collected, dissolved in 200 ml of HEPES buffer (20 mM HEPES, pH 7.0, containing 5 mM MgCl 2 , 1 mM DTT, and protease inhibitor mixture (complete, Roche Applied Science)), and dialyzed against 10 liters of HEPES buffer for 3 h followed by a change of buffer to 10 liters of HEPES buffer containing 2 M NaCl. The dialyzed sample was filtered and loaded onto a hydrophobic interaction column (HiPrep 16/10 Phenyl FF (High sub), Amersham Biosciences) and eluted with a linear gradient of 2-0 M NaCl in HEPES buffer. The active fractions were then serially purified on an affinity column (HiTrap Blue HP 5 ml, Amersham Biosciences) with a linear gradient of 0 -1 M NaCl, an anion exchange column (Resource Q 1 ml, Amersham Biosciences) with a linear gradient of 0 -1 M NaCl and an anion exchange column (Mono Q PC 1.6/5, Amersham Biosciences) in HEPES/CHAPS buffer (HEPES buffer containing 0.05% CHAPS). Between each step, the active fractions were dialyzed against the HEPES/ CHAPS buffer. A portion of the active fraction from the Mono Q column was loaded onto a gel filtration column (Superdex 75 PC 3.2/30, Amersham Biosciences) equilibrated with HEPES/CHAPS buffer containing 150 mM NaCl. A portion of the active fractions from the Mono Q column and the gel filtration column was loaded onto SDS-polyacrylamide gel under reduced conditions, and the gel was stained with SYPRO Ruby (Molecular Probes) and scanned by Molecular Imager FX (Bio-Rad). The 2Ј-PDE activity was measured by NH 3 assay as described below, and the protein concentration was determined by the modified Bradford method (Coomassie Plus Protein Assay, Pierce) with bovine serum albumin as a standard.
The band on the SDS-polyacrylamide gel was excised, and the protein in the gel was reduced, alkylated, and digested as described previously (29). The resultant peptides were subjected to reverse-phase liquid chromatography with tandem mass spectrometry as described previously (30) with slight modification. The tandem mass spectra were searched against the GenBank non-redundant protein data base using the Mascot program (Matrix Sciences).
Molecular Cloning of Human 2Ј-PDE-To obtain the full-length sequence of human 2Ј-PDE, we performed a BLAST search against the human EST data base using the corresponding nucleotide sequence of BAB85079.1 (AK074423.1, GenBank) as a query. We found a series of EST fragments with obvious identity. By assembling these EST sequences (AK074423.1, BM894197.1, H06019.1, and H42112.1, all in GenBank), we obtained a virtual longest nucleotide sequence (1,939 bp, AB115695, DDBJ/EMBL/GenBank), which revealed an open reading frame predicted to encode 609 amino acids with the C-terminal extension of BAB85079.1.
Construction of Classification Tree-The human proteins containing the PF03372 domain (31) in the Swiss Protein data base and the human hypothetical proteins with 20 -30% identity to human 2Ј-PDE in Gen-Bank data bases were retrieved. Partial sequences corresponding to the PF03372 domain were extracted from these proteins and subjected to multiple alignment using ClustalW (32). A dendrogram was subsequently drawn using the Treeview program.
Northern blot analysis was performed for various cell lines. Total RNA from the cells was prepared using TRIzol reagent (Invitrogen), and 15 g of the RNA was loaded and separated on 1% agarose gel containing 6% formaldehyde and transferred onto nylon membranes (Hybond-N ϩ , Amersham Biosciences). About a 1-kbp fragment of the human 2Ј-PDE cDNA or human ␤-actin cDNA was 32 P labeled using Rediprime II Random Prime Labeling Kit (Amersham Biosciences), and then the membrane was hybridized with these probes in ExpressHyb buffer (Clontech). The expression levels of each mRNA were determined using a bioimage analyzer (Fujix BAS2000, Fujifilm).

Expression of Human 2Ј-PDE in Mammalian Cells-
The open reading frame of the full-length human 2Ј-PDE was subcloned into the expression vector pEF6/V5-His TOPO (Invitrogen), which encodes human 2Ј-PDE with the C-terminal V5 epitope and His 6 (V5-His) tags. 20 g of the DNA was transfected into semiconfluent COS-1 cells (1 ϫ 10 6 cells) using 30 l of FuGENE 6 (Roche Applied Science) in Dulbecco's modified Eagle medium with 10% fetal bovine serum (FBS). As a negative control, an equal amount of pEF6/V5-His-TOPO/LacZ (Invitrogen) and FuGENE 6, or only FuGENE 6, was transfected into the COS-1 cells. The transfected cells were cultured for 72 h, washed twice with phosphate-buffered saline, and lysed with CelLytic-M (Sigma) containing 1 mM DTT and protease inhibitor mixture at 25°C for 5 min. The lysate was centrifuged, and the supernatant was stored at Ϫ80°C until use. For experiments to determine 2Ј-PDE activity in the high performance liquid chromatography (HPLC) assay or analysis, recombinant human 2Ј-PDE with V5-His tags was partially purified using a HiTrap Chelating HP column (Amersham Biosciences) and used as an enzyme source.
In the NH 3 assay, 10 l of sample was added to 40 l of HEPES buffer containing 1 mg/ml bovine serum albumin and 1.5 mM A2ЈpA (Seikagaku), and the mixture was incubated at 37°C for 1 h. The produced adenosine was converted to inosine and NH 3 at 37°C for 20 min by adding 10 l of 0.5 M succinic acid, pH 6.0, containing 3.5 g/ml adenosine deaminase (Sigma) and 2.5 M KCl. The NH 3 produced was assayed with an ammonia test kit (Wako Pure Chemicals). One unit/ml 2Ј-PDE activity was defined as the concentration required for 1 mM adenosine production. 3Ј-PDE activity was determined by using 1.5 mM A3ЈpA (Seikagaku) in place of A2ЈpA.
In the HPLC assay, 80 l of 0.75 mM A2ЈpA in the assay buffer (20 mM HEPES, pH 7.0, containing 1 mg/ml bovine serum albumin, 1 mM DTT, and 5 mM MgCl 2 ), 10 l of partially purified human 2Ј-PDE, and 10 l of inhibitor dissolved in dimethyl sulfoxide were mixed and incubated at 37°C for 1 h. The samples were mixed with 200 l of methanol to inactivate the enzyme, and 10 l of the filtrate (0.45 m) was subjected to a C 18 column (Inertsil ODS-2, 5 m, 4.6 ϫ 150 mm, GL Science) and analyzed with isocratic elution in 50 mM ammonium acetate, pH 5.0, containing 5% acetonitrile. Absorbance at 260 nm was monitored, and 2Ј-PDE activity was determined by measuring the decreased peak area of A2ЈpA.
HPLC Analysis of Degradation of 2-5A Trimer by Recombinant 2Ј-PDE-2-5A trimer, pppA2ЈpA2ЈpA, was enzymatically synthesized by chicken 2Ј,5Ј-OAS (33)(34)(35), purified using an anion exchange column (TSKgel DEAE-2SW, 4.6 ϫ 250 mm, Tosoh), and confirmed by mass spectrometry. 90 l of 50 M pppA2ЈpA2ЈpA in the assay buffer (20 mM HEPES, pH 7.0, containing 1 mg/ml bovine serum albumin, 1 mM DTT, and 5 mM MgCl 2 ) and 10 l of partially purified human 2Ј-PDE were mixed and incubated at 37°C for specified times. As a negative control, pppA2ЈpA2ЈpA was incubated for 7 h with 10 l of H 2 O instead of the enzyme. The samples were incubated at 95°C for 5 min to inactivate the enzyme and centrifuged briefly, and 50 l of the supernatant was subjected to a C 18 column (XTerra MS C 18 , 5 m, 4.6 ϫ 150 mm, Waters) with a guard column (3.9 ϫ 20 mm) of the same material equilibrated with 50 mM triethylamine acetate, pH 7.0. After injection, the column was washed with the same buffer for 5 min, and the nucleotides were eluted with a linear gradient of 0 -10% acetonitrile over 20 min. Absorbance at 260 nm was monitored, and each eluted peak was fractionated and analyzed by mass spectrometry. Peaks corresponding to AMP and ATP were further confirmed using standards.
Protection from IFN-or dsRNA-induced Cell Death in Human Prostate Cancer Cells Expressing 2Ј-PDE-Human prostate cancer PC-3 cells were grown in RPMI 1640 medium supplemented with 10% heatinactivated FBS, 100 g/ml streptomycin, and 100 units/ml penicillin (all from Invitrogen) in humidified 5% CO 2 at 37°C. To establish stable clones expressing 2Ј-PDE, a human 2Ј-PDE cDNA fragment was inserted into the mammalian expression vector pEF-DEST51 containing the blasticidin-resistant determinant to express 2Ј-PDE with C-terminal V5-His tags. The resulting construct was introduced into PC-3 cells using LipofectAMINE Plus reagent (Invitrogen), and then the blasticidin-resistant transformants were selected in RPMI 1640 medium containing 10% FBS and 10 g/ml blasticidin S. Stable transformant colonies were isolated and confirmed by Western blot analyses. Whole cell extracts were subjected to 10% SDS-PAGE and then electrotransferred onto a polyvinylidene difluoride membrane (Bio-Rad). After having been blocked with Block Ace (Dainippon Pharmaceutical), the membrane was incubated first with indicated antibodies to a V5 epitope tag (Invitrogen) or actin (Oncogene), and then with peroxidase-conjugated secondary antibody followed by an ECL Plus Western blotting detection system (Amersham Biosciences).
Transfection of siRNA, Quantitative Real Time PCR, and Virus Infection Assay-HeLa cells (1 ϫ 10 5 cells) were grown in semiconfluent monolayers on a 24-well plate (collagen type I-coated, Asahi Techno Glass) in minimal essential medium (MEM) supplemented with 10% heat-inactivated FBS in humidified 5% CO 2 at 37°C. Transfection of siRNA (20 pmol) into HeLa cells was accomplished using Lipo-fectAMINE 2000 (Invitrogen), and the cells were cultured for 24 h. siRNAs for the human 2Ј-PDE (5Ј-GUACAAGGUGGAGCGCAAC-dTdT-3Ј and 5Ј-GUUGCGCUCCACCUUGUACdTdT-3Ј) and unrelated protein (mouse carboxyl esterase 3, 5Ј-GGUGCUCUCAGAGCUCUUC-dTdG-3Ј and 5Ј-GAAGAGCUCUGAGAGCACCdTdG-3Ј) were obtained from Dharmacon. After replacing the medium with 250 l of MEM containing 2.5% FBS, the cells were cultured for 24 h with or without 300 units/well IFN-␣ (EMD Biosciences). After changing the medium to fresh MEM containing 2.5% FBS, the cells were infected or noninfected with vaccinia virus (WR strain, multiplicity of infection ϭ 0.001), cultured for 24 h, and subjected to analysis of the 2Ј-PDE transcript and viral replication.
For quantification of 2Ј-PDE mRNA in HeLa cells, total RNA was prepared using RNeasy Mini Kit (Qiagen), and cDNA was prepared from 1 g of total RNA using oligo(dT) and Superscript II RNase H reverse transcriptase in a final volume of 20 l. Real time PCR amplifications were performed from 5 l of cDNA diluted 1:125 by using gene-specific primer sets as follows: 2Ј-PDE forward, 5Ј-GTCATCAAT-GGCAGCATTCCAGAG-3Ј; 2Ј-PDE reverse, 5Ј-CTATTTCCATTTTA-AATCACATACAAGTGC-3Ј; ␤-actin forward, 5Ј-CATTGCTCCTCCT-GAGCGCAA-3Ј; ␤-actin reverse, 5Ј-CTGCGCAAGTTAGGTTTTGTC-3Ј. Each primer set was used at a concentration of 300 nM in a final volume of 50 l using the QuantiTect SYBR Green PCR Kit (Qiagen). All PCRs were performed on a sequence detection system (ABI Prism 7900HT, Applied Biosystems), and quantification of a given gene was calculated after normalization to ␤-actin. The specificity of the PCR amplification was verified by melt curve analysis of the final products directly in the sequence detector and by agarose gel electrophoresis.
To determine viral replication, the cells were washed with phosphate-buffered saline, fixed with absolute ethanol at 25°C for 10 min, and dried. The procedure for visualization of infected cells by peroxidase-antiperoxidase staining was described previously (36) and used with some modifications. The fixed cells were treated successively with rabbit anti-vaccinia virus antibody (1:500, Virostat, Portland, ME), goat anti-rabbit immunoglobulin (1:1,000, DakoCytomation), and peroxidase-rabbit antiperoxidase complex (1:1,000, DakoCytomation) for 60 min each. The peroxidase reaction was conducted for 5 min using a liquid DAB Substrate-Chromogen System (DakoCytomation). The plaque numbers were counted visually, and the antiviral effects were normalized using the plaque numbers of virus infection only as 100%.
Effect of 2Ј-PDE Inhibitor on Virus Infection Assay-The effect of a 2Ј-PDE inhibitor on the virus infection assay was determined in a manner similar to that of siRNA. Semiconfluent HeLa cells on a 24-well plate grown in MEM with 2.5% FBS were treated with or without 300 units/well IFN-␣ in MEM with 2.5% FBS for 24 h. The medium was changed with fresh MEM containing 2.5% FBS and an inhibitor, and the cells were infected with vaccinia virus (WR strain, multiplicity of infection ϭ 0.001). After 24 h of culture, the degree of viral infection was determined using peroxidase-antiperoxidase staining as described above.
Statistical Analysis-Comparison of two groups was analyzed by the two-tailed unpaired Student t test and the dose effects were evaluated by the Dunnett test.

RESULTS
Purification and Identification of 2Ј-PDE-We developed two assays for detecting 2Ј-PDE activity, an NH 3 assay and an HPLC assay. In the NH 3 assay, the product of the 2Ј-PDE enzymatic reaction, adenosine, was converted to NH 3 and inosine by adenosine deaminase, and the resultant NH 3 was quantified using a commercially available kit. The linear correlation efficient between the adenosine concentration and final absorbance at 630 nm was more than 0.99. One unit/ml 2Ј-PDE activity was defined as 1 mM adenosine production. Typical background absorbance without substrate was sufficiently low (A 630 ϭ 0.06) compared with absorbance of 1 mM adenosine used as a standard (A 630 ϭ 0.17). In the case of liver homogenates or samples after ammonium sulfate precipitation, residual ammonia brought high background absorbance, therefore we dialyzed these samples to reduce the background absorbance before the NH 3 assay. The HPLC assay was based on the assay of Johnston and Hearl (28) with slight modification. The NH 3 assay had high throughput capacity using a 96-well plate, whereas the HPLC assay had better sensitivity and tolerance to dimethyl sulfoxide. Although 2-5A is defined as an oligomer more than a dimer of triphosphorylated 2Ј,5Ј-linked oligoadenylates and the A2ЈpA dimer is known not to activate RNase L (9), we used A2ЈpA as a substrate in both the NH 3 and HPLC assays for the following reasons: (i) 2Ј-PDE was reported to degrade A2ЈpA (7, 27, 28); (ii) A2ЈpA was commercially available; (iii) 5Ј-unphosphorylated oligoadenylates are more stable than 5Ј-phosphorylated oligoadenylates; and (iv) a high throughput assay can be applied using adenosine production (NH 3 assay). We primarily used the NH 3 assay because of its high throughput capacity.
To elucidate the molecular mechanism and biological role of the 2-5A system, we purified 2Ј-PDE activity from bovine liver by successive six-step purification and obtained more than 28,000-fold concentrated 2Ј-PDE (Table I). By comparing the bands of the fifth (data not shown) and final (Fig. 1) purification steps on SDS-polyacrylamide gel, a band of ϳ65 kDa was supposed to correlate with the 2Ј-PDE activity. Fortunately, we could identify the human hypothetical 59-kDa protein (BAB85079.1, GenBank) corresponding to the bovine 2Ј-PDE candidate protein by mass spectrometry. However, the pre-dicted molecular mass of 59 kDa deduced from its amino acid sequence (535 amino acids) was smaller than that of the 65-kDa protein observed on SDS-polyacrylamide gel, suggesting that BAB85079.1 is a partial clone of human 2Ј-PDE.
Molecular Cloning and Characterization of Human 2Ј-PDE-To obtain the full-length sequence of human 2Ј-PDE, we performed in silico EST assembly and obtained a virtual full size nucleotide sequence (AB115695, DDBJ/EMBL/GenBank), which revealed an open reading frame predicted to encode 609 amino acids with the C-terminal extension of BAB85079.1 (Fig.  2A). The predicted molecular mass of this protein was 67 kDa, which is well consistent with the observed molecular mass of our purified protein on SDS-polyacrylamide gel. We succeeded in isolating cDNA clones with DNA sequences identical to the assembled EST sequence by RT-PCR of human liver poly(A) ϩ RNA. Thus, we designated this assembled EST sequence as human 2Ј-PDE.
The BLAST and BLAT searches against GenBank data bases revealed that human and mouse 2Ј-PDE genes consist of three exons and are mapped onto 3q21.2 and 14A3, respectively. Human and mouse 2Ј-PDE proteins share 88% identity ( Fig.  2A). Notably, the proteins have only one potential functional domain of the endonuclease/exonuclease/phosphatase family (Pfam: PF03372) with a high score (E-value: 5.3 ϫ 10 Ϫ29 for human 2Ј-PDE; 1.0 ϫ 10 Ϫ26 for mouse 2Ј-PDE). BLAST searches with the full-length human 2Ј-PDE or PF03372 domain revealed no proteins or cDNAs sharing more than 50% identity to human 2Ј-PDE (data not shown). To examine the genetic distance between human 2Ј-PDE and other human proteins with the PF03372 domain, we constructed a classification tree (Fig. 2B). Human 2Ј-PDE belongs to an unknown functional subgroup that contains hypothetical proteins with 20 -30% identity to human 2Ј-PDE. Because RNases do not have the PF03372 domain, human 2Ј-PDE is supposed to be more distant from RNases than DNases. RT-PCR analysis indicated that the mRNA encoding 2Ј-PDE was widely distributed in human tissues (Fig. 3A). Human 2Ј-PDE mRNA was detected at ϳ4 kb by Northern blot analysis in all tissues examined (data not shown) and in human PC-3, LNCaP, HEK293, and HepG2 cell lines (Fig. 3B).
To determine whether isolated human 2Ј-PDE cDNA encodes protein with 2Ј-PDE activity, the expression vector for 2Ј-PDE was transfected into COS-1 cells, and 2Ј-PDE activity was measured using A2ЈpA as a substrate. Significant 2Ј-PDE activity was observed in the lysate of the transfected cells expressing human 2Ј-PDE as compared with the lysate of cells expressing LacZ or untransfected cells (Fig. 3C). These results indicate that the cloned gene certainly encoded human 2Ј-PDE. The 2Ј-PDE was found to cleave the 3Ј-5Ј phosphodiester bond of A3ЈpA (Fig. 3D). By comparing the increased 2Ј-and 3Ј-PDE activities over the activities of the negative controls, human 2Ј-PDE showed no marked preference for degradation of the 2Ј-5Ј over the 3Ј-5Ј bond. We observed the same preference by partially purified bovine and recombinant human 2Ј-PDE (data not shown).
To examine whether our cloned 2Ј-PDE certainly cleaves the 2Ј-5Ј bond of authentic triphosphorylated 2-5A, we investigated degradation of 2-5A trimer, pppA2ЈpA2ЈpA, by human recombinant 2Ј-PDE with V5-His tags. Because 2-5A was not commercially available and difficult to synthesize chemically, we synthesized 2-5A enzymatically using chicken 2Ј,5Ј-OAS (33)(34)(35), which had been generously donated to us. Our human 2Ј-PDE indeed degraded pppA2ЈpA2ЈpA to produce the final product of AMP and ATP (Fig. 4). Interestingly, we observed only a small increase in the intermediate pppA2ЈpA during the reaction, which indicated that 2Ј-PDE has a low dissociation   2. Amino acid sequences of 2-PDE. A, the human (AB115695, DDBJ/EMBL/GenBank) and mouse (BAC40606) amino acid sequences of 2Ј-PDE are aligned (first and second sequences, respectively). Identical (*) and homologous (⅐) residues between the human and mouse sequences are indicated. The underlining indicates the identified sequences by liquid chromatography/tandem mass spectrometry analysis which would be conserved between human and bovine enzymes. The box shows the predicted endonuclease/exonuclease/phosphatase family domain (Pfam, rate constant against pppA2ЈpA and tends to cleave the 2Ј-5Ј bond successively without dissociating from the substrate after the first cleavage. We also observed that our 2Ј-PDE degraded pA2ЈpA2ЈpA (data not shown). Taken together, our cloned 2Ј-PDE certainly degrades authentic 2-5A and 2-5A analogs.
Reduced Cytotoxicity of IFNs and dsRNA by 2Ј-PDE Overexpression-IFNs are reported to have an antiproliferative effect on various cell lines including the human prostate cancer cell line, PC-3 (37). Furthermore, recent studies have shown an association between RNase L and prostate cancer (21)(22)(23)(24)(25)(26). To study the potential role of 2Ј-PDE in the 2-5A system that is implicated in IFN action, we established stable transfectants of PC-3 cells expressing human 2Ј-PDE. Western blot analysis confirmed the expression of the gene in two clones, both at a high level (clone 6) and low level (clone 10) (Fig. 5A). The two proximate bands observed in the Western blot analysis might have been caused by a difference in the translation start site at the first and the second methionine in the open reading frame. The high level clone (clone 6) showed more than a 50-fold increase of the 2Ј-PDE transcript in quantitative RT-PCR and more than a 20-fold increase of 2Ј-PDE enzymatic activity in the HPLC assay than the parental PC-3 (data not shown). Both transfectants expressing higher and lower levels of 2Ј-PDE showed protection from loss of cell viability induced by IFNs or dsRNA (Fig. 5, B-D). When the expression of 2Ј-PDE was increased, the antiproliferative activity of the IFNs and dsRNA was decreased. These observations provide new evidence for a principal role of 2Ј-PDE as a negative regulator of the 2-5A system, especially in antitumor action by IFNs.
Reduced Viral Replication by 2Ј-PDE Suppression or Inhibition-To investigate the role of 2Ј-PDE in the antiviral mechanism, we used siRNAs for human 2Ј-PDE to suppress the expression of 2Ј-PDE in HeLa cells. HeLa cells are known to possess relatively high levels of 2Ј,5Ј-OAS activity, whose activity is increased additively by IFN treatment (38). Transfection of 2Ј-PDE siRNA suppressed the 2Ј-PDE expression level in HeLa cells to approximately half of that in cells untreated or transfected with unrelated siRNA regardless of vaccinia virus infection (Fig. 6A). We attempted to determine the decrease in levels of 2Ј-PDE enzymatic activity by siRNA; however, our PF03372). B, classification tree of human 2Ј-PDE. An unrooted tree was constructed from sequences of 18 human genes which have the PF03372 domain, by ClustalW and Treeview. The scale bar indicates maximum likelihood branch length of Ͼ0.1 inferred substitutions/site. Proteins from the SWISS-PROT data base are indicated by entry names, and other proteins with 20 -30% identity to human 2Ј-PDE are indicated by GenBank accession number. Three major subgroups are shown in the classification tree: the inositol phosphatase family (green), the DNase family (red), and proteins of unknown function (blue). HPLC assay could not detect 2Ј-PDE activity even in the untransfected cells (data not shown), which would be the result of insufficient sensitivity of our assay and the low expression levels of 2Ј-PDE in HeLa cells. Suppression of the 2Ј-PDE expression by the specific siRNA showed significant reduction of vaccinia virus replication in HeLa cells in the absence of IFNs (Fig. 6B).
We next examined whether a 2Ј-PDE inhibitor, A-74528a (Fig. 6C), could suppress viral replication. A-74528a was isolated from Streptomyces sp. SANK 61196 as an inhibitor for partially purified bovine 2Ј-PDE. 2 As described above, although we used primarily the NH 3 assay, we used the HPLC assay for measuring inhibitory activity of A-74528a because A-74528a was not dissolved in aqueous buffer but in dimethyl sulfoxide. A-74528a inhibited enzymatic activity of human 2Ј-PDE with an IC 50 value of 34 g/ml (Fig. 6D) as well as that of bovine 2Ј-PDE (data not shown). As expected, A-74528a showed dose- dependent reduction of viral replication irrespective of IFN-␣ pretreatment (Fig. 6E). Significantly, A-74528a showed no cytotoxicity in an XTT (sodium 3Ј-[1-[(phenylamino)-carbonyl]-3,4-tetrazolium]-bis(4-methoxy-6-nitro)benzene-sulfonic acid hydrate) assay in the absence of viral infection (data not shown).
Although vaccinia virus was reported to be sensitive to IFN treatment concomitant with accumulation of 2-5A, the vaccinia virus is more resistant than other viruses such as the encephalomyocarditis virus (39,40), which is thought to be caused by the E3L gene of vaccinia virus, which encodes a dsRNA-binding protein that inhibits both dsRNA-dependent protein kinase and 2Ј,5Ј-OAS (41,42). Therefore, our vaccinia virus assay would be less sensitive than an assay using other viruses to detect influence on viral replication. Nevertheless, the siRNA for 2Ј-PDE and the 2Ј-PDE inhibitor showed statistically significant reduction of viral replication. These results demonstrated that 2Ј-PDE counteracted the antiviral pathway, most likely by inhibiting the 2-5A system. Our results also suggest that 2Ј-PDE inhibitors like A-74528a are likely to have therapeutic value in the treatment of viral diseases. DISCUSSION Although the 2-5A system is well known as an antiviral and antitumor mechanism, remarkably 2Ј-PDE had not been elucidated molecularly. In this study, we have for the first time identified the molecular structure of 2Ј-PDE, which appears to be a key enzyme in the 2-5A system. 2Ј-PDE was reported to be 2Ј-5Ј-exonuclease by Schmidt et al. (27). Unfortunately, we observed just a small increase in pppA2ЈpA during degradation of pppA2ЈpA2ЈpA by 2Ј-PDE and could not clearly demonstrate the direction of cleavage (Fig. 4). However, preliminary results using 2-5A tetramer revealed 2Ј to 5Ј direction of cleavage (data not shown), which indicates that our 2Ј-PDE was identical to the enzyme reported before (27). On the other hand, our cloned 2Ј-PDE appears to have characteristics different from the enzymes reported as purified 2Ј-PDE in previous studies (27,28). The molecular mass of human 2Ј-PDE identified in this study is 67 kDa, which is similar to the 65-kDa bovine spleen enzyme reported in the study of Johnston and Hearl (28). However, their 2Ј-PDE cleaved preferably a 3Ј,5Ј-phosphodiester bond at more than 100-fold efficiency (28) and this is very different from our 2Ј-PDE, which showed no marked preference for cleavage of the 2Ј-5Ј bond over the 3Ј-5Ј bond (Fig. 3). We also observed no marked preference of partially purified bovine 2Ј-PDE. However, this difference in substrate preference between Johnston and Hearl's (28) and ours would be the result of a difference between the bovine and human enzymes because our purification was partial. Therefore, the possibility cannot be excluded that our cloned 2Ј-PDE is a human homolog of Johnston and Hearl's (28). With regard to this substrate specificity, 2Ј-PDE from mouse L cells reported by Schmidt et al. (27) is very similar to our 2Ј-PDE, although the molecular mass of their 2Ј-PDE was 35 kDa on SDS-polyacrylamide gel. Because the mouse sequence homologous to human 2Ј-PDE in the data base showed a calculated molecular mass of 67 kDa, their 2Ј-PDE might be a processed form or a splice variant of 2Ј-PDE. Further studies are needed to clarify whether our 2Ј-PDE is different from that of these studies.
Preliminary experiments using partially purified human 2Ј-PDE showed a K m of ϳ170 M for A2ЈpA and 180 M for A3ЈpA (data not shown), and these values are in a range similar to those (2-20 M) reported by Johnston and Hearl (28). Because RNase L can be activated by nanomolar concentration of 2-5A (9), this K m seems to be high. As discussed by Johnston and Hearl (28), dephosphorylation of 2-5A by 5Ј-phosphatase may primarily mediate rapid inactivation of 2-5A, and complete degradation of 2-5A by 2Ј-PDE may be a slow process. Our observations that overexpression of 2Ј-PDE in PC-3 cells attenuated IFN action and that inhibition or suppression of 2Ј-PDE reduced viral replication indicated that 2Ј-PDE plays an important role in the regulation of the 2-5A system. The dominant pathway of inactivation of 2-5A remains elusive. More detailed enzymatic characterization of 2Ј-PDE and, furthermore, 2Ј-PDE knockout animals, would reveal the contribution of 2Ј-PDE in the metabolism of 2-5A.
The identified human 2Ј-PDE would be encoded by a single gene and highly conserved between human and mouse (88% identity) according to BLAST and BLAT searches against Gen-Bank™ data bases including the human genome sequence ( Fig.  2A and data not shown). In the 2-5A system, RNase L is also encoded by a single gene as is clearly revealed in RNase L Ϫ/Ϫ mice (10), and human and mouse RNase L has moderate identity (65%) between the overlapping regions (8). In contrast, 2Ј,5Ј-OAS has several forms, which are encoded by three distinct genes, two of which have two splice variants (6). The homology of the 2Ј,5Ј-OAS domains between each of the forms has a range of 41-60% identity. Human 2Ј-PDE is distinct from common phosphodiesterases including DNases and RNases, and only genes of unknown function were found to have slight homology to human 2Ј-PDE (Fig. 2B). These hypothetical genes might be related to metabolism of other ribonucleotide analogs, such as diadenosine oligophosphates (43).
The activation of the 2-5A system, that is, promotion of RNA breakdown by RNase L, has been thought to have antiviral and antitumor effects (3). There are three approaches to achieve this activation: promotion of 2-5A synthesis by 2Ј,5Ј-OASs; inhibition of 2-5A degradation by 2Ј-PDE; and direct activation of RNase L (Fig. 7). Among them, partly because of the difficulty in regula-tion of the gene expression of 2Ј,5Ј-OASs and the absence of molecular identification of 2Ј-PDE, 2-5A analogs to activate RNase L have been especially studied (44 -46). However, most 2-5A analogs are difficult to use as test compounds practically because they cannot penetrate into cells because of their anionic nature. Moreover, constitutive expression of RNase L might lead to concern about possible adverse effects of such 2-5A analogs in normal cells. In contrast to an RNase L activator, a 2Ј-PDE inhibitor might have potential to act as an enhancer of IFN action. 2-5A was reported to be present in healthy human plasma and serum (47). As 2Ј-PDE is widely and constitutively expressed in various human tissues (Fig. 3A), a physiological role of 2Ј-PDE would be to protect normal cells from loss of cell viability by preventing unexpected increase in 2-5A to activate the 2-5A system. This protective function of 2Ј-PDE, however, might reduce the efficiency of the host defense mechanism mediated by the 2-5A system. Indeed, overexpression of 2Ј-PDE inhibited IFN-and dsRNA-induced cell death (Fig. 5), whereas suppression of 2Ј-PDE expression by siRNA or inhibition of catalytic activity by a 2Ј-PDE inhibitor showed significant reduction in viral replication (Fig. 6). These results strongly suggest that 2Ј-PDE is an important negative regulator of the 2-5A system. Suppression of 2Ј-PDE activity at the time of viral infection or exposure to IFNs might increase the 2-5A level by changing the balance between synthesis and degradation of 2-5A, leading to activation of RNase L in the 2-5A system. Therefore, 2Ј-PDE would be a novel molecular target for new antiviral and antitumor therapeutic agents.
We partially purified 2Ј-PDE from bovine liver at more than 28,000-fold concentration of activity (Table I). However, the band corresponding to 2Ј-PDE activity was a minor band far from homogeneity (Fig. 1A). Protein identification by mass spectrometry permits (i) cross-species identification where amino acid sequences are highly conserved among species (48); (ii) identification from a complex mixture as far as the band is distinct from other bands on SDS-polyacrylamide gel; and (iii) identification with high sensitivity, which reduces the required amount of a protein. Incomplete purification and reduction of the final product mean fewer purification steps and less starting material. Indeed, we successively identified our protein in such an incomplete purification procedure, which took only 6 days. This rapid micropurification and well developed identification technology enabled us to identify 2Ј-PDE, a forgotten enzyme for more than 15 years. We believe this technology can be applied to other abandoned proteins with difficulty of purification.
In conclusion, we have purified and molecularly identified 2Ј-PDE, which might be a key enzyme in the 2-5A system. Molecular cloning of 2Ј-PDE can provide a more definitive evaluation of the 2-5A system in the physiological context. In addition, 2Ј-PDE inhibitors would be useful as novel research tools and, importantly, may have potential as antiviral and/or antitumor drugs.