Preparation and biological properties of a highly active poly(G) X poly(C) inducer of interferon

Experimental conditions necessary for the full expression of interferon-inducing activity by a complex of polyguanylic acid and polycytidylic acid included: (i) a sufficiently high molecular size of each homopolymer; (ii) annealing conditions which insured complete denaturation of polyguanylic acid self-structure; and (iii) the specific biological assay employed to assay interferon-inducing potency. The complex of polyguanylic acid and polycytidylic acid possessed several properties that suggested it may be an atypical polynucleotide interferon inducer. For instance, it was inactive in primary rabbit kidney cell cultures, usually exquisitely sensitive to polynucleotide interferon inducers, unless it was incubated on the cell cultures for prolonged times or in the presence of DEAE-dextran. Polyguanylic acid X polycytidylic acid could induce interferon in rabbits and mice but gave a more protracted response than did poly(I) X poly(C). Finally, poly(G) X poly(C) was, without any modification, resistant to degradation by serum nucleases.

Synthesis and biological activity of tubercidin analogues of ppp5'A2'p(5'A2'p)n5'A

A series of tubercidin (7-deazaadenosine) analogues of 2-5A of the general formula p5'(c7A)2'p[5'(c7A)-2'p]n5'(c7A) (n = 0-5) were prepared by lead ion catalyzed polymerization of the 5'-phosphoroimidazolidate of tubercidin. Through the corresponding imidazolidates, these oligonucleotide 5'-monophosphates were converted to the 5'-triphosphates. All reported structures were corroborated by enzyme digestion and 1H or 31P nuclear magnetic resonance. When evaluated for its ability to bind to the 2-5 A-dependent endonuclease of mouse L cells, the tubercidin analogue of trimeric 2-5A, namely, ppp5'(c7A)2'p5'(c7A)2'p5'(c7A), and the corresponding tetramer were bound as effectively as 2-5A itself; nonetheless, it and the corresponding tetramer, ppp5'-(c7A)2'p5'(c7A)2'p5'(c7A)2'p5'(c7A), failed to stimulate the 2-5A-dependent endonuclease as judged by its inability to inhibit translation in extracts of mouse L cells programmed with encephalomyocarditis virus RNA and to give rise to ribosomal RNA cleavage in the same cell system under conditions where 2-5A showed activity at 10(-9) M. The trimer, ppp5'(c7A)2'p5'(c7A)2'p5'(c7A), was an antagonist of 2-5A action in the L cell extract. In the lysed rabbit reticulocyte system, both the trimeric and tetrameric tubercidin 2-5A analogues were bound to the 2-5A-dependent endonuclease as well as 2-5A, but in this case, the tetramer triphosphate, ppp5'(c7A)2'p5'(c7A)2'p5'(c7A)2'p5'(c7A), was just as potent an inhibitor of translation as 2-5A tetramer triphosphate. Moreover, this inhibition was prevented by the established 2-5A antagonist p5'A2'p5'A2'p5'A.(ABSTRACT TRUNCATED AT 250 WORDS)

Oligonucleotide structural parameters that influence binding of 5'-O-triphosphoadenylyl-(2'----5')-adenylyl-(2'----5')-adenosine to the 5'-O-triphosphoadenylyl-(2'----5')-adenylyl-(2'----5')-adenosine dependent endoribonuclease: chain length, phosphorylation state, and heterocyclic base

A number of 2',5'-linked oligoadenylates and their analogues were prepared and evaluated for their ability to interact with the 5'-O- triphosphoadenylyl -(2'----5')-adenylyl-(2'----5')-adenosine (2-5A) dependent endoribonuclease of mouse L cells. The oligonucleotides were assayed for their ability to antagonize the action of 2-5A, to displace a radiolabeled probe from the 2-5A-dependent nuclease, or to inhibit translation in a cell-free system. These experiments demonstrated the following: (1) Three AMP residues in a 5'-phosphorylated oligonucleotide were needed for maximum interaction with the endonuclease, and higher oligomers (greater than or equal to 4 AMP residues) did not show significantly higher binding. (2) The third (2'-terminal) adenosine residue was required for optimal binding activity. (3) 5'-Phosphorylation of the oligonucleotide was necessary for maximum binding to the endonuclease, but the first (from the 5' terminus) internucleotide phosphate of higher unphosphorylated or core oligomers, such as A2'p5'A2'p5'A2'p5'A, may partly replace the requirement for a 5'-monophosphate moiety; in agreement with this, the 5'-methyl ester of 5'pA2'p5'A2'p5'A, i.e., Me-p5'A2'p5'A2'p5'A, was bound to the endonuclease as well as or better than the higher core oligomers but approximately 100 times more effectively than the trimer core, A2'p5'A2'p5'A. (4) Base-modified analogues, such as p5'C2'p5'C2'p5'C, p5'U2'p5'U2'p5'U, or p5'I2'p5'I2'p5'I, were at least 2000 times less effectively bound to the endonuclease than p5'A2'p5'A2'p5'A. (5) The triphosphate ppp5 'I2'p5'I2'p5'I was 10 000 times less active than 2-5A as an inhibitor of translation. These latter two points implied the critical role of the adenine N1-nitrogen and/or exocyclic amino group in the binding of 2-5A to the endonuclease.

Synthesis and biological activity of 5'-capped derivatives of 5'-triphosphoadenylyl(2'----5')adenylyl(2'----5')adenosine

The oligonucleotides A5'pp5'A2'p5'A2'p5'A and A5'ppp5'A2'p5'A2'p5'A were prepared by reaction of AMP or ADP, respectively, with the 5'-(phosphoimidazolidate) of A2'p5'A2'p5'A. A5'pppp5'A2'(p5'A)n (n = 1-3) were synthesized by reaction of p5'A2'(p5'A)n (n = 1-3) with adenosine 5'-trimetaphosphate. All structures were confirmed by enzyme digestion and 1H and 31P nuclear magnetic resonance (NMR). The products A5'pppp5'A2'p5'A and A5'pppp5'A2'p5'A2'p5'A were found to be identical with two of the products of the 2-5A synthetase catalyzed reaction of Ap4A with ATP, thus confirming the structural assignments made by earlier investigators. In extracts of mouse L cells programmed with encephalomyocarditis virus RNA, A5'pppp5'A2'p5'A2'p5'A2'p5'A and A5'pppp5'A2'p5'A2'p5'A were equipotent with 2-5A itself as inhibitors of translation. The oligomers A5'ppp5'A2'p5'A2'p5'A and A2'pppp5'A2'p5'A were about 100 times less active than 2-5A, and A5'pp5'A2'p5'A2'p5'A was without translational inhibitory activity. When affinity for the 2-5A-dependent endonuclease was determined (by displacement of 2-5A[32P]pCp from endonuclease), all of the analogues, as well as 2-5A itself, had similar affinities for the endonuclease except for A5'pppp5'A2'p5'A, which was bound approximately 100 times less effectively. Under conditions of the radiobinding assay, A5'pppp5'A2'p5'A2'p5'A was degraded (t1/2 = 2 h) to ATP, ADP, AMP, ppp5'A2'p5'A2'p5'A, and p5'A2'p5'A2'p5'A.(ABSTRACT TRUNCATED AT 250 WORDS)

A sensitive immunoenzymometric assay for 2',5'-oligoadenylate. Detection of elevated 2',5'-oligoadenylate synthetase in human peripheral mononuclear cells

A competition immunoenzymometric assay for 2',5'-oligoadenylate was developed and employed to measure the interferon-inducible enzyme 2',5'-oligoadenylate synthetase in cell extracts. Microtiter plates coated with a novel conjugate of p5'A2'p5'A2'p5'A and N-(2-aminoethyl)-carbamylmethylated-Ficoll (AECM-Ficoll) bound rabbit polyclonal or mouse monoclonal antibody directed against 2',5'-oligoadenylate. Binding was inhibited by soluble 2',5'-oligoadenylate. Estimates of 2',5'-oligoadenylate concentrations based on inhibition of antibody binding compared favorably with those obtained using a protein synthesis inhibition assay. Low concentrations of 2',5'-oligoadenylate synthesized in vitro by extracts of human peripheral mononuclear cells were conveniently estimated using less than or equal to 10(6) cells. Virtually identical results were obtained when either total extract or synthetase bound to poly(I) . poly(C)-agarose was used for the in vitro incubation. When peripheral mononuclear cells were incubated in vitro with interferon, the normally low levels of 2',5'-oligo(A) synthetase rose dramatically. The assay was employed to measure synthetase levels in peripheral mononuclear cells isolated from patients with systemic lupus erythematosus. Some of these patients were found to have elevated levels of 2',5'-oligoadenylate synthetase.

Biological activities of phosphodiester linkage isomers of 2-5A

To determine the relative importance of the 2',5'-phosphodiester bond of 2-5A in its binding to and activation of the 2-5A-dependent ribonuclease (RNase L, RNase F), a number of phosphodiester linkage isomers of 2-5A were prepared. These isomers were obtained either by lead ion-catalyzed polymerization of adenosine 5'-phosphorimidazolidate or by T4 polynucleotide kinase-catalyzed 5'-phosphorylation of adenylyl(3' leads to 5')adenylyl(3' leads to 5')adenosine followed by reaction of the corresponding phosphorimidazolidates with tri(n-butylammonium)pyrophosphate. The following 2-5A isomers thus were prepared: ppp5'A2'p5'A3'p5'A, ppp5'A3'p5'A2'p5'A, ppp5'A3'p5'A3'p5'A("3-5A"), ppp5'A2'p5'A3'p5'A2'p5'A,and ppp5'A3'p5'A2'p5'-A2'p5'A. The ability of these isomeric 2-5As to interact with the 2-5A-dependent endonuclease was ascertained by three different criteria: (i) ability to prevent the protein synthesis inhibitory effects of 2-5A, (ii) activity as an inhibitor of translation in encephalomyocarditis RNA-programmed L cell extracts, and (iii) ability to prevent binding of the radiolabeled probe, ppp5'A2'p5'A2'p5'A2'p5'A3'[32P]p5'Cp, to the endonuclease of L cell extracts. In certain experiments, degradation of oligonucleotide was minimized or eliminated by altering assay conditions, providing alternate phosphodiesterase substrates, or by using purified endoribonuclease of Ehrlich ascites cells. By all criteria, replacement of 2',5'-bond by a 3',5'-bond led to a substantial decrease in biological activity. Generally, replacement of just one 2',5'-phosphodiester bond with a 3',5'-linkage led to at least a one order of magnitude loss of activity. In accord with this trend, ppp5'A3'p5'A3'p5'A(3-5A) was greater than 10,000 less active than 2-5A in binding to the endonuclease or as an inhibitor of protein synthesis.

Assay of 2',5'-oligoadenylate phosphodiesterase activity in mouse L-cell extracts

A rapid and convenient assay for adenylyl(2' leads to 5')adenosine(A2'p5'A) or adenylyl(3' leads to 5')adenosine(A3'p5'A) phosphodiesterase activities is described. The dinucleotides A3'p5'A and A2'p5'A were labeled to a high specific activity by means of a catalytic-exchange procedure. Degradation studies of each of these labeled dinucleotides showed an asymmetrical distribution of label between the two adenine bases. Enzymatic degradation of [3H]A3'p5'A or [3H]A2'p5'A could be quantitated by first digesting the reaction products with bacterial alkaline phosphatase and then adding a slurry of DEAE-Sephadex. Under conditions described, adenosine did not adsorb to the resin, whereas dinucleotides as well as AMP did adsorb. As a consequence, when liquid scintillation fluid was added to the DEAE-Sephadex reaction mixture slurry, the radioactivity of the dinucleotides and AMP was severely quenched. This permitted a direct estimation of the amount of adenosine liberated during the phosphodiesterase degradation and subsequent alkaline phosphatase digestion. This method was applied to the measurement of A2'p5'A degrading activities in extracts of mouse L cells. Extracts from control mouse L cells were as active in degrading A2'p5'A as extracts from interferon pretreated cells.

Cordycepin analogues of 2-5A and its derivatives. Chemical synthesis and biological activity

Cordycepin (3'-deoxyadenosine) analogues of 2-5A were prepared by dicyclohexylcarbodiimide-induced polymerization of cordycepin 5'-monophosphate. A series of oligomers of the general formula p5'(3'dA)-2'[p5'(3'dA)]n (n = 1-5) was obtained. Cordycepin trimer 5'-monophosphate (n = 2) and cordycepin tetramer 5'-monophosphate (n = 3) were converted to the corresponding 5'-di- and triphosphates via the phosphorimidazolide method. Confirmation of assigned structures was provided through enzyme digestions, 1H and 31P NMR, and comparison with material which was synthesized by a completely independent route. Neither the cordycepin trimer triphosphate, ppp5'(3'dA)-2'p5'(3'dA)2'p5'(3'dA), nor the cordycepin tetramer triphosphate, ppp5'(3'dA)2'p5'(3'dA)2'p5'(3'dA)2'p5'-(3'dA), were inhibitors of translation in cell-free extracts of mouse L-cells programmed with encephalomyocarditis virus. In rabbit reticulocyte lysates, the cordycepin trimer triphosphate was devoid of activity, whereas the cordycepin tetramer triphosphate had approximately 1/100th the activity of 2-5A tetramer triphosphate, ppp5'A2'p5'A2'p5'A2'p5'A. Even though the cordycepin analogues were unable to activate the 2-5A-dependent endoribonuclease, they were able to bind to the endonuclease, thereby antagonizing the translational inhibitory effects of 2-5A. Thus, replacement of the 3'-hydroxyl groups of all the ribose moieties of 2-5A results in an analogue which can bind to the 2-5A-dependent endoribonuclease but is incapable of activating the enzyme. The 3'-hydroxyl groups of 2-5A, therefore, are critical for activation of the 2-5A-dependent endonuclease.

Double-stranded RNA-dependent phosphorylation of protein P1 and eukaryotic initiation factor 2 alpha does not correlate with protein synthesis inhibition in a cell-free system from interferon-treated mouse L cells

The double-stranded RNAs (I)n X (C)n and (A)n X (dUfl)n (dUfl is 2'-fluoro-2'-deoxyuridylic acid) have been compared as inhibitors of translation in cell-free systems from interferon-treated mouse L cells and from rabbit reticulocytes. In the interferon-treated mouse L-cell system, both double-stranded RNAs stimulated kinase activity, leading to phosphorylation of protein P1 and eukaryotic initiation factor 2 alpha (eIF-2 alpha), but only (1)n X (C)n activated the (2'-5')-oligoadenylate synthetase. Moreover, in this system, (I)n X (C)n, but not (A)n X (dUfl)n, inhibited translation. Both (A)n X (dUfl)n and (I)n X (C)n also activated the rabbit reticulocyte kinase to phosphorylate protein P1 and eIF-2 alpha, but, in contrast to mouse L-cell systems, both (A)n X (dUfl)n and (I)n X (C)n were potent inhibitors of translation in reticulocyte lysates. These results indicate that protein P1 and eIF-2 alpha phosphorylation are not sufficient to cause inhibition of protein synthesis in interferon-treated mouse L-cell extracts. They further suggest that protein synthesis inhibition by (I)n X (C)n in extracts of interferon-treated L cells correlates better with activation of (2'-5')-oligoadenylate synthetase than with activation of the protein P1 and eIF-2 alpha kinase.

5-Substituted 2'-deoxyuridines: correlation between inhibition of tumor cell growth and inhibition of thymidine kinase and thymidylate synthetase

A large variety of 5-substituted 2'deoxyuridines (dUrds) and 2'-deoxyuridylates (dUMPs) have been evaluated for their inhibitory effects on the thymidine (dThd) kinase or thymidylate (dTMP) synthetase isolated from mouse leukemia L1210 cells. The most potent inhibitors of dThd kinase were 5-chloro-, 5-bromo- and 5-iodo-dUrd. Their Ki/Km values ranged from 0.57 to 0.82. All dUrd analogs tested showed competitive kinetics with respect to dThd. However, there was little, if any, correlation between the inhibitory effects of the compounds on L1210 cell growth and their inhibitory activities against dThd kinase (r = 0.16). The most potent inhibitors of dTMP synthetase were (in order of decreasing activity): 5-nitro-dUMP greater than 5-formyl-dUMP greater than 5-fluoro-dUMP greater than 5-oxime of 5-formyl-dUMP greater than 5-azidomethyl-dUMP greater than (E)-5-(2-bromovinyl)-dUMP. The ki/Km values for these compounds ranged from 0.001 to 0.665. All dUMP analogs tested showed competitive kinetics with respect to dUMP (if not preincubated with the enzyme at 37 degrees). There was a strong correlation (r = 0.833) between the inhibitory effects of these compounds on L1210 cell growth and their inhibitory activities against dTMP synthetase. Thus, the suppressive action of 5-substituted dUrd derivatives on tumor cell growth would involve prior conversion of the nucleoside analogs to the corresponding 5'-monophosphates followed by an inhibition of dTMP synthetase.

Chemical modification potentiates the biological activities of 2-5A and its congeners

Chemical modification of p5'A2'(p5'A2')np5'A by a periodate oxidation/Schiff base formation/borohydride reduction cycle gave a series of 2-5A analogues in which the ribose of the 2'-terminal nucleotide was transformed to an N-substituted morpholine (azahexapyranose). 2',5'-Oligoriboadenylate 5'-monophosphates bearing this modification were 5-10 times more potent as antagonists of the action of 2-5A or poly(I).poly(C) than was unmodified p5'A2'p5'A2'p5'A. Application of this modification to the tetramer triphosphate ppp5'A2'p5'A2'p5'A2'p5'A resulted in an analogue with 10 times the activity of ppp5'A2'p5'A2'p5'A (2-5A trimer triphosphate) as an inhibitor of protein synthesis or activator of the 2-5A-dependent endoribonuclease. This activator of the 2-5A-dependent endoribonuclease. This new 2-5A analogue, the most potent 2-5A derivative reported to date, inhibited translation in extracts of mouse L-cells programmed with encephalomyocarditis virus RNA at a concentration of 10(-10) M (concentration for half-maximal inhibition). All such N-substituted morpholine modified 2',5'-oligoriboadenylates were found to be extremely resistant to degradation by L-cell extracts under conditions where unmodified 2-5A or its derivatives were quickly destroyed. These data demonstrate the necessity for an intact terminal ribose ring for the action of the 2-5A phosphodiesterase. Thus, extensive chemical modification of 2' terminus of 2-5A may be possible without adversely affecting its biological activity while endowing it with other favorable properties such as resistance to degradation.

Role of thymidine kinase in the inhibitory activity of 5-substituted-2'-deoxyuridines on the growth of human and murine tumor cell lines

Twenty-four 5-substituted 2'-deoxyuridines have been evaluated for their inhibitory effects on the growth of three human lymphoblast cell lines (Namalva, RAji and TK- (thymidine kinase deficient) Raji) and these inhibitory effects were compared to those for two murine leukemia cell lines (L1210/0 and L1210/BdUrd). The latter was selected from the parental L1210/0 cell line by its ability to grow at high concentrations of 5-bromo-dUrd and could also be considered as TK-. There was a close correlation between the inhibitory effects of the deoxyuridine analogs on Namalva, Raji and L1210 cells: the correlation coefficient (r) for log ID50 (median inhibitory dose) for L1210 cell growth, on the one hand, and log ID50 for Namalva or Raji cell growth, on the other hand, was 0.902 and 0.929, respectively. There was also a strong correlation (r = 0.936) between the log ID50 values for the two human lymphoblast cell lines. However, there was no significant correlation (r less than 0.40) either between the log ID50 for the TK- Raji cells and the parental TK+ Raji cells, or between the log ID50 for the TK- L1210/BdUrd cells and the parental TK+ L1210/0 cells. We may conclude therefore, that (i) the murine leukemia L1210 cell system is predictive for the growth-inhibitory effects of 5-substituted 2'-deoxyuridines on human lymphoblast cell lines, and (ii) the antitumor cell activity of the 5-substituted 2'-deoxyuridines is, to a large extent, dependent on the thymidine kinase activity of the tumor cells.

Antiviral, antimetabolic, and cytotoxic activities of 5-substituted 2'-deoxycytidines

Various 5-substituted 2'-deoxycytidines, including 5-bromo-dCyd, 5-iodo-dCyd, 5-nitro-dCyd, 5-ethynyl-dCyd, 5-propyl-dCyd, (E)-5-(2-bromovinyl)-dCyd, and (E)-5-(2-iodovinyl)-dCyd, were evaluated for their antiviral and antimetabolic properties in primary rabbit kidney (PRK) cell cultures and for their inhibitory effects on murine L1210 cell proliferation. All dCyd analogues proved to be selective inhibitors of herpes simplex virus (HSV) replication: 5-bromo-dCyd, 5-iodo-dCyd, 5-nitro-dCyd, and 5-ethynyl-dCyd were more selective in their anti-HSV activity than were the corresponding 5-substituted 2'-deoxyuridines, whereas 5-propyl-dCyd, (E)-5-(2-bromovinyl)-dCyd, and (E)-5-(2-iodovinyl)-dCyd were as selective as their dUrd counterparts. The dCyd analogues were also less cytotoxic (for both PRK and L1210 cells), as could be monitored by inhibition of either cell proliferation or host-cell DNA synthesis (incorporation of radiolabeled precursors). Of all 5-substituted 2'-deoxycytidines tested, the (E)-5-(2-halogenovinyl) derivatives emerged as the most potent and most selective inhibitors of HSV (Type 1) replication.

5'-O-Monophosphoryladenylyl(2' goes to 5')adenylyl(2' goes to 5')adenosine is an antagonist of the action of 5'-O-triphosphoryladenylyl-(2' goes to 5')adenylyl(2' goes to 5')adenosine and double-stranded RNA

5'-O-Monophosphoryladenylyl(2' goes to 5')adenylyl-(2' goes to 5')adenosine [(2' goes to 5')(pA)3] antagonizes the protein synthesis inhibitory effects of 5'-O-triphosphoryladenylyl-(2' goes to 5')adenylyl(2' goes to 5')adenosine by preventing activation of the (2' goes to 5')oligo(a)-activated ribonuclease which degrades mRNA. The oligoribonucleotide (2' goes to 5')(pA)3 also antagonizes the translational inhibitory capacity of poly(I).poly(C) in extracts of interferon-treated L cells, suggesting that (2' goes to 5')oligo(A) is the primary mediator of the protein synthesis inhibitory effects of poly(I).poly(C) in this cell-free system.

Interferon induction by poly(inosinic acid).poly(cytidylic acid) segmented by spin-labels

Poly(inosinic acid).poly(cytidylic acid) [(I)n.(C)n] duplexes of which the (C)n strand was modified to various degrees chemically or enzymatically with nitroxide radicals (spin-labels) were evaluated for interferon-inducing activity. Upon annealing of the chemically modified (C)n, (1C,Cx)n (X = 1000 or 16), with (I)n, the interferon-inducing activity was similar to that of (I)n.(C)n in PRK cell cultures. However, to overcome hydrolysis of the spin-label linkage in (1C,Cx)n, and enzymatic approach was taken to synthesize (1S(4)U,Cx)n copolymers with x = 100, 38, 16, and 8. The (1s(4)U,Cx)n copolymers were chemically stable, and upon annealing with (I)n the correlation time of the nitroxide moiety in (I)n. (1s(4)U,Cx)n was determined. A comparison of this correlation time with that measured for (RUGT,U100)n.(A)n, which contains the nitroxide moiety in position 5 of the U moiety, suggests that the 1s(4)U residue is in a nonintrahelical conformation and partitions the duplex into double-helical segments of varying size. The interferon-inducing activity of (I)n. (1s(4)U,Cx)n was evaluated in primary rabbit kidney, human skin fibroblast (strain VGS), and mouse L-929 cell cultures as well as in rabbits. The 1s(4)U residue did not cause a significant change in the interferon induction as compared to (I)n.(C)n in most systems tested unless x less than 16. These findings indicate that double-helical segments of approximately 16 base pairs partitioned by nonintrahelical 1s(4)U residues suffice to trigger the interferon response in all systems studied.

Synthesis and antiviral activity of 5-[(cyanomethylene)oxy]-2'-deoxyuridine

To study the influence of substitution of CN for C identical to CH in the anti-herpes virus nucleoside 5-(propynyloxy)-2'-deoxyuridine (1), 5-[(cyanomethylene)oxy]-2'-deoxyuridine (2) was prepared. When the potassium salt of 5-hydroxy-2'-deoxyuridine was reacted with iodoacetonitrile in dry DMF, the bisalkylated product 3-(cyanomethyl)-5-[(cyanomethylene)oxy]-2'-deoxyuridine (3) was the major product with a lesser amount of 3-(cyanomethyl)-5-hydroxy-2'-deoxyuridine (5) and only a trace amount of the desired product (2). In contrast, when 5-hydroxy-2'-deoxyuridine was alkylated in water in the presence of 1 equiv of KOH, compound 2 was the major product. In cultures of primary rabbit kidney (PRK) cells, compound 2 showed an anti-herpes virus activity that was comparable to that of 1 and ara-A. Compound 2 did not inhibit incorporation of [Me-3H]dThd or [1',2'-3H]dUrd into DNA of PRK cells; however, its anti-herpes virus activity was completely prevented upon the addition of either dThd or dUrd.