THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 263, No. 3, Issue of January 25, pp. 1131-1139, Printed in U .S.A.

1988

Only One 3”Hydroxyl Group of ppp5’A2‘p5’A2’p5‘A (2-5A) Is Required for Activation of the 2-5A-dependent Endonuclease*

(Received for publication, December 1, 1986)

Paul F. Torrenee$, Danuta BrozdaS, David AlsterS, Ramamurthy Charubalas, and Wolfgang Pfleiderers From the $Laboratory of Analytical Chemistry, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892 and the §Faculty of Chemistry, University of Konstanz, Postfach 5560, 0-7750, Konstanz, Federal Republic of Germany

To investigate the relative importance of each of the ribose 3’-hydroxyl groups of 2-5A (ppp5’A2‘p5’A2’- p5’A) in determining binding to and activation of the 2-5A-dependent endonuclease (RNase L), the 3‘-hy- droxyl functionality of each adenosine moiety of 2-5A trimer triphosphate was sequentially replaced by hy- drogen. The analog in which the 5”terminal adenosine was replaced by 3‘-deoxyadenosine (uiz. ppp5’(3‘dA)- 2’pSfA2’p5’A) was bound to RNase L as well as 2-5A itself and was only 3 times less potent than 2-5A as an activator of RNase L. On the other hand, when the second adenosine unit was replaced by 3‘-deoxyadenosine (uiz. ppp5’A2’p5’(3’dA)2’p5’A), binding to RNase L was decreased by a factor of eight relative to 2-5A trimer and, even more dramatically, there was a 500-1000-fold drop in ability to activate the 2-5A-dependent endonuclease. Finally, when the 3’-hydroxyl substituent was converted to hydrogen in the 2”terminal residue of 2-5A, a significant increase in both binding and activation ability occurred. We conclude that only the 3’-hydroxyl group of the second (from the terminus) nucleotide residue of 2-5A is needed for effective activation of RNase L.

Interaction of the unique oligonucleotide ppp5‘AZfp5’- A2’p5’A(2-5A)’ ( l ) , produced in interferon-treated virus-in- fected cells, with a latent endonuclease (RNase L) can lead to RNA degradation and inhibition of translation (reviewed in Ref. 2). It appears that each individual nucleotide residue of 2-5A may assume a fundamentally different role in binding to and activation of RNase L. For instance (3), the purine N1 and/or 6-amino group of the first (5’) nucleotide unit was vital for binding to RNase L, the purine N1 and/or 6-amino residue of the second nucleotide was of minimal importance in either binding or activation, whereas the purine N1 and/ or 6-amino functionality of the third nucleotide was relatively unimportant in binding but critical for RNase L activation. Herein, we demonstrate that such discrimination also occurs at the level of the ribose phosphate backbone of 2-5A.

* A preliminary report of this work was presented at the Seventh International Round Table on Nucleosides, Nucleotides, and Their Biological Applications, Konstanz, Federal Republic of Germany, October, 1986. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are 2-5A, ppp5’A2’p5’A2‘p5’A; HPLC, high performance liquid chromatography.

MATERIALS AND METHODS’

RESULTS

Chemistry-The synthetic approach to the 3”deoxyaden- osine sequentially substituted analogs of 2-5A relied upon the application of recent developments in protecting group chem- istry, specifically the p-nitrophenylethyl and p-nitrophenyle- thoxycarbonyl groups (4, 5). These groups form stable ester bonds but can be quantitatively cleaved by a p-elimination reaction in aprotic solvents. The &elimination route has the advantage that cleavage of a C-0 rather than P-0 bond occurs during the deprotection process; thus, there is much less possibility of chain cleavage or isomerization at phosphate residues. A phosphotriester methodology was used to join appropriately protected nucleotides and/or oligonucleotides; specifically, triisopropylbenzenesulfonyl chloride and N - methylimidazole or 1,2,4-triazole and 2,5-dichlorophenyl- phosphodichloridate were used to couple a precursor with a free 5”hydroxyl to an intermediate p-nitrophenylethyl-pro- tected diester. Protecting groups were removed as follows: tert-butyldimethylsilyl, tetrabutylammonium fluoride in di- methyl formamide; p-nitrophenylethyl and p-nitrophenyle- thoxycarbonyl, diazabicyclo[5.3.0]undecene; 2,5-dichloro- phenyl, p-nitrobenzaldoximate; monomethoxytrityl, toluene- sulfonic acid. The specific synthetic details may be found in Fig. 1 and under “Materials and Methods.”

Proof of the assigned structures for the three 3’-deoxyaden- osine-substituted analog 5’-monophosphates were obtained through the method of synthesis as well as elemental analysis of a number of synthetic intermediates including 2, 3, 10, 11, 12, 13, 15, 17, 19,20, 21, 29, and 30. In addition, proton NMR (Table 1, Miniprint) showed the requisite number of adeno- sine ring protons as well as anomeric protons for each isomeric analog.

The 5’-triphosphate of each analog was generated by first converting the corresponding 5’-monophosphate to the 5’- phosphoroimidazolidate by a modification of the procedure of Mukaiyama and Hashimoto (6). After separation as sodium salts, the imidazolidates were reacted with tri-n-butylammon- ium pyrophosphate to yield the 5’-triphosphates. Confirma- tion of structure was obtained from four different approaches. (i) The HPLC retention time of each product was that pre- dicted for a 5”triphosphate (Table 2, Miniprint). (ii) Diges- tion with bacterial alkaline phosphatase gave an oligonucle- otide “core” identical to that obtained by bacterial alkaline

Portions of this paper (including “Materials and Methods” and Tables 1-4) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

1131

1132 Biological Roles of the 3' A

t

. w.1

u*:

. *",. rn p ryl* I

FIG. 1. A , synthetic scheme for the preparation of p5'- (3'&)2'p5'A2'p5'A. Abbreviations used here and in the following synthetic schemes are: MMTr, monomethoxytrityl; @, -

-Hydroxyl Groups of 2-5A

phosphatase digestion of the corresponding 5'-monophos- phate (Tables 2 and 3, Miniprint). (iii) Digestion with snake venom phosphodiesterase gave the expected cleavage products in the predicted stoichiometric ratios (Table 3, Miniprint). (iv) Extended digestion with 0.5 N KOH gave degradation products consistent with the assigned oligonucleotide struc- tures; in addition, the appropriate ratio of products was ob- tained (Table 4, Miniprint). (v) In addition under milder conditions of KOH hydrolysis (0.3 N KOH, 16 h), the 5'- triphosphate moieties remained intact. Thus, ppp5'(3'dA)- 2'p5'A2'p5'A gave ppp5'(3'dA)2'p5'A2'[3']p + A, ppp5'- A2'p5'(3'dA)2'p5'A gave ppp5'A2'[3']p + (3'dA)2'p5'A, andppp5'AZ'p5'A2'p5'(3'dA) gave ppp5'A2'[3']p + A2'[3'] p + 3'dA. Each of the products was identified by coinjection on HPLC with an authentic sample with one exception: namely, putative ppp5'(3'dA)2'p5'A2'[3']p for which no au- thentic material for comparison was available. However, its retention time on HPLC was consistent with such a highly charged product.

Biochemical Evaluation of the 2-5A Analogs-The ability of each of the synthetic analogs to bind to the 2-5A-dependent endonuclease of mouse L cells was determined using a radio- binding assay developed by Knight e t al. (7) using ppp5'A2'p5'A2'p5'A2'pEi'A3'["'P]p5'C3'p. In this method- ology, the latter 32P-labeled 2-5.4 derivative competes with the unlabeled 2-5A analog for binding to the RNase L in the nitrocellulose filter assay (7). Results are summarized in Table 1 for the 5'-triphosphates. The results can be expressed in terms of the concentrations required to inhibit binding of 50% of the applied 32P-probe, that is the ICso. Under these conditions and in agreement with earlier related experiments (3,), 2-5A was able to displace 50% of the radiolabeled probe at a concentration of about 4 X 10"O M in oligonucleotide. The synthetic oligonucleotides evaluated could be ranked in the following order of increasing binding affinity together with their respective IC5' values: ppp5'AZrp5'(3'dA)2'p5'A, 3 X lo-' M < ppp5'A2'p5'A2'p5'A = ppp5'(3'dA)2'p5'- A2'p5'A, 4 X 10"'M <ppp5'A2'~5'A2'~5'(3'dA), 8 X lo-" M.

Although not illustrated here, radiobinding assays were also performed with the 5'-monophosphate of each analog oligo- nucleotide. In this case, the following order of increasing binding affinity (and ICso values) was obtained: p5'A2'p5'- (3'dA)2'p5'A, 7 x lo-' M < p5'A2'p5'A2'p5'A = p5'A2'- p5'A2'p5'(3'dA), 1 x lo-' M < p5'(3'dA)2'~5'A2'pS'A, 5 x 10"' M. The 5'-monophosphates all were bound to RNase L with slightly less efficiency than the corresponding 5'-tri- phosphates with one exception, specifically p5'A2'p5'A2'- p5'(3'dA).

The activation of RNase L by 2-5A can lead to degradation of mRNA and subsequent inhibition of protein synthesis (2); thus, as one measure of the ability of the synthetic 2-5A analogs to activate the endonuclease, their ability to inhibit translation was determined. Nuclease-pretreated extracts of mouse LK cells were programmed with encephalomyocarditis virus RNA, and potential inhibitors were added at time 0. The concentration of oligomer needed to effect a half-maximal inhibition of translation was determined. After 120 min, tri- chloroacetic acid-insoluble radioactivity was measured. The results of such studies are summarized in Table 1. The oligo- nucleotide triphosphates could be ranked in the following

Si(CH3)2-C(CH,)3; Bz, benzoyl; NPE, p-nitrophenylethyl; NPEOC, 0 II

02N"Q CH2CH20 C -. B , synthetic scheme for the preparation of p5'A2'p5(3'&)2'p5'A. C, synthetic scheme for the preparation of pS'A2'pS'AZ'p5'(3'&).

Biological Roles of the 3‘-Hydroxyl Groups of 2-54

TABLE I Biological activities of 3‘-deoxyadenosine-substituted analogues of 2-5A

This table summarizes the ability of the various 3’-deoxyadenosine-substituted 2-5A analogues to activate RNase L as expressed by their ability to cause degradation of p~ly(U)-[~*P]pCp in the core-cellulose assay (expressed as IC, or concentration needed to effect a 50% degradation) and as measured by the analogue’s capacity to inhibit protein synthesis (expressed as IC600(mar), the concentration needed to effect half-maximal inhibition compared to 2-5A which usually caused a maximum of 80% inhibition or translation). In addition, the center column shows the ability of the analogue to bind to RNase L as determined by the radiobinding assay using ppp5’A2’p5’A2’p5’A2’p5’A2’[32CAp]p5‘Cp as probe. All of the experiments herein were done with extracts of mouse L K cells. The experiments were carried out over a period of 6 months with different preparations of mouse L cells, encephalomyocarditis virus RNA, radiolabeled probe, and/or 2-5A cellulose. 2-5A trimer was included as a control in all experiments. In a number of experiments, one analogue was compared to another. The values presented are mean values f S.E. The number of experiments carried out is indicated in parentheses.

1133

Oligomer RNA degradation (IC,)

Radiobinding (IC,)

Translational inhi- bition” (IC,,mmxd

M M M

ppp5’A2‘p5‘A2’p5’A 1.0 k 0.5 X 10”’ 6.0 & 1.7 X 10”’ 5.2 & 1.8 X 10”’

ppp5’(3’dA)2’p5‘A2’p5’A (n = 9) (n = 13) (n = 10)

3.0 f 1.0 X 10”’ 6.0 f 1.2 X 10”’ 1.8 0.3 x 10-9 (n = 7) (n = 5)

ppp5’A2’p5‘(3’dA)2‘p5’A (n = 5)

3.0 k 1.2 X lo-’ 4.0 f 2.2 X 10-9 4.0 * 1.0 x 10-7 (n = 5)

ppp5’A2’~5’A2’~5’(3’dA) (n = 3) (n = 3)

7.0 f 0.8 X lo-” 9.0 f 2 x 10-11 (n = 2) (n = 3) (n = 3)

8 x 10-12

“2-5A as well as ppp5’(3’dA)Z’p5’A2’p5‘A and ppp5’A2’~5’A2’~5’(3’dA) resulted in an 80-90% inhibition of protein synthesis as a maximum; on the other hand, ppp5’A2’p5’(3’dA)2’p5’A even at high concentrations could not effect any greater than a 50% inhibition of translation.

order of increasing potency as translational inhibitors: ppp5’A2’p5’(3’dA)2’p5’A << ppp5’(3’dA)2’p5’A2’p5pA < ppp5’A2‘p5‘A2‘p5‘A < ppp5’A2’p5’A2’p5’(3’dA). Note- worthy also was the observation (not illustrated) that ppp5’A2’p5’(3 ’ dA)2’p5 ’A, even at the highest concentration tested, could not inhibit translation to the same extent as 2- 5A itself or the other 3’-deoxyadenosine-~ubstituted analogs.

The above two methods for evaluating the biological activity of the 3’-deoxyadenosine-substituted analogs of 2-5A were carried out using crude cell-free extracts. Recently, a 2-5A- dependent RNase assay was developed (8) which involves the immobilization and partial purification of RNase L on core cellulose (A2’p5’A2’p5’A2’p5’A-~ellulose). After addition of 2-5A or analog, the breakdown of p0ly(U)-3’-[~~P]pCp could be monitored by acid-insoluble radioactivity. This “core-cel- lulose” assay thus provides a separate means of evaluating the ability of the 2-5A analogs to activate RNase L and does so with a partially purified enzyme. The results are summa- rized in Table I. Again, results could be expressed in terms of the concentration of 2-5A or analog needed to effect a 50% degradation of 32P-labeled poly(U). The oligonucleotides eval- uated could be ranked (together with their respective ICso values) in the following increasing order of potency in their ability to cause RNA degradation: ppp5‘A2‘p5‘(3’dA)2‘p5‘A, 4 X lo-’ M << ppp5‘(3’dA)2‘~5’A2‘p5’A, 2 X 10”” M < ppp5’A2’p5’A2’p5’A, 4 X lo-” M < ppp5’A2’p5’A2’- p5’(3’dA), 8 X lo”* M. This relative behavior was in good agreement with the results of the protein synthesis inhibition assay; the only difference was that the core-cellulose assay was approximately 10 times more sensitive than the protein synthesis assay.

DISCUSSION

In 1981, Doetsch et al. (9) reported that 2-5A synthetase from rabbit reticulocyte lysate was capable of elaborating a 3’-deoxyadenosine (cordycepin) analogue of 2-5A trimer with the structure ppp5’(3’dA)2’~5’(3’dA)2’~5‘(3’dA). Proof of structure for this putative cordycepin analogue of 2-5A was based largely on enzyme digests including the observation that in a single thin-layer chromatography system, the retic-

ulocyte lysate product, upon digestion with bacterial alkaline phosphatase, gave a product with the same RF as authentic “core” (3’dA)Zfp5’(3’dA)2’p5’(3’dA) prepared by Charubala and Pfleiderer (10). Doetsch et al. (9) went on to reveal that in a rabbit reticulocyte cell-free translation system, the pu- tative trimer ppp5‘(3’dA)2’~5’(3’dA)2’~5’(3’dA) was a sig- nificantly more potent inhibitor of protein synthesis than either 2-5A trimer ppp5’ A2’p5 ’ A2 ’p5’ A or tetramer ppp5’A2’p5’A2’p5’A2’p5’A. In the hands of other investi- gators, 2-5A trimer triphosphate has been found to be a poor nuclease activator in the rabbit reticulocyte lysate system (11, 12). Apparently, the tetrameric species ppp5’A2‘p5‘A2‘- p5‘A2’p5‘A is required for activation of rabbit reticulocyte RNase L at the subnanomolar concentrations effective in other systems.

Three independent groups have reported on unequivocal chemical syntheses using different synthetic routes for the 3’- deoxyadenosine analogues of 2-5A trimer and tetramer di- and triphosphates, specifically ppp5’(3’dA)2’~5’(3’dA)2’- p5’(3’dA) (13, 15), pp5’(3’dA)2’~5’(3’dA)2~5’(3’dA) (14), ppp5‘(3’dA)Z’p5’(3’dA)2’p5’(3’dA)2’p5’(3’dA) (13), and pp5‘(3’dA)2’p5‘(3’dA)Z’p5‘(3’dA)2’p5’(3’dA) (13). Struc- tures of final products as well as intermediates were estab- lished by physical and enzymic methods. A totally separate effort (10) led to the preparation of the 3‘-deoxyadenosine analogue of 2-5A trimer core. In this latter instance, however, the core oligomer was not further phosphorylated up to the mono-, di-, or triphosphate level. Nonetheless, Sawai et al. (13) were able to enzymically phosphorylate the product of Charubala and Pfleiderer’s scheme to a monophosphate iden- tical with that prepared by the route of Sawai et al. (13); conversely, Charubala and Pfleiderer’s product was shown to be identical to the bacterial alkaline phosphatase digestion product from the approach of Sawai et al. Beyond any reason- able doubt, the products of the four above synthetic chemical approaches correspond to their assigned structures.

These synthetically nrepared 3‘-deoxyadenosine analogues of 2-5A have been evaluated by a number of investigators employing a variety of assay systems. In a mouse L K cell-free system, such analogues as (p)pp5’(3’dA)2’~5’(3’dA)2’~5’-

1134 Biological Roles of the 3’-Hydroxyl Groups of 2-5A

(3‘dA) and (p)pp5’(3’dA)2’~5‘(3‘dA)2’~5’(3’dA)2’~5’- (3‘dA) were bound to the 2-5A-dependent endonuclease about 20 times less effectively than 2-5A (13). In such a mouse cell- free system, encephalomyocarditis virus RNA-programmed protein synthesis was not affected by any of the 3“deoxy- adenosine analogues listed above unless extremely high (

M ) concentrations were used. In these conditions, half- maximal protein synthesis inhibition was effected by lo-’ M 2-5A (13). If incubation time were prolonged up to 4 h from the normal 90 min used due to linearity of protein synthesis and if high concentrations of 3“deoxyadenosine analog, ppp5’(3’dA)2’~5’(3’&)2’~5’(3‘dA), were employed, then some translational inhibition could be witnessed (16). For instance, it required 2 X 10“ M ppp5’(3’dA)2’p5’(3‘dA)- 2’p5’(3’dA) to give an inhibition equivalent to 5 X 10”O M ppp5’A2‘pS‘A2’~5’A ((lS), a 400,000-fold difference in po- tency. The activity of such 3‘-deoxyadenosine analogues was further attested to by the fact that they were effective antag- onists of the inhibition of translation generated by 2-5A (13).

In a rabbit reticulocyte cell-free system using endogenous globin mRNA, ppp5’(3’dA)2’~5’(3’dA)2’~5’(3’dA), like ppp5‘A2‘p5’A2’p5‘A, was devoid of activity at least up to concentrations of M. On the other hand, under conditions where the 2-5A tetramer, ppp5’A2’~5‘A2’~5’A2’~5’A, could give 50% of maximal protein synthesis inhibition at 5 X 10”O M, it required M of the cordycepin tetramer ppp5’- (3’dA)z’p5’(3’dA)2’~5’(3’dA)2’~5’(3‘dA) to effect a similar inhibition (13).

In their study of the 5”diphosphate derivative, pp5’- (3’dA)2’~5’(3’dA)2’~5’(3’dA), Haugh et al. (14) reported that this analogue was bound to the RNase L of Daudi lymphobtastoid celIs and rabbit reticulocytes about 10 times less effectively than 2-5A trimer 5‘-diphosphate but was at least 300 times less effective in binding to the mouse L cell enzyme. Furthermore, employing a rRNA cleavage assay they found no evidence for activation of RNase L in mouse L cells or Daudi cells even when micromolar concentrations of ana- logue were used under conditions that gave rRNA cleavage at nanomolar 2-5A trimer 5”diphosphate levels. These studies were carried out with 5‘-diphosphates, and it has been dem- onstrated that such diphosphates can activate RNase L either directly (14, 17) or by conversion to the corresponding tri- phosphate (18). Subsequent studies3 (16) with the triphos- phate ppp5’(3’dA)2’p5’(3’dA)2‘p5’(3‘dA) failed to reveal any evidence of rRNA cleavage in mouse L cell, Daudi cell, or HeLa cell systems even when concentrations of analogue as high as 10“ M were employed. In those systems 2-5A caused rRNA cleavage as low as lo-’ M.

Later, taking advantage of a newly developed method that permitted immobilization and partial purification of RNase L from mouse L cells on 2’,5’-tetraadenylate (core)-cellulose, Krause et al. (19) found that concentrations of ppp5’- (3’dA)Z’p5’(3’dA)Z’p5’(3’dA) as high as 64 pM could not stimulate the degradation of p01y(U)-3’-[~*P]C, into acid- insoluble fragments whereas 2-5A did so at subnanomolar concentration.

In a separate study which used calcium phosphate or cal- cium carbonate coprecipitation to introduce the oligonucleo- tide directly into intact mouse L929 cells, the 3”deoxyaden- osine analogue was found to be approximately 100 times less potent than 2-5A as a protein synthesis inhibitor (16). It was verified in this study that the translational inhibition was in fact due to RNase L activation.

Nyilas et al. (15) studied the behavior of synthetic ppp5’(3’dA)2’~5’(3’dA)2’~5’(3’dA) in both mouse L929 cell

K. Lesiak and P. F. Torrence, unpublished observations.

extracts and in Ehrlich ascites tumor cell extracts. In the L cell sap, 3’-deoxyadenosine 2-5A analogue was bound to RNase L about 50 times less effectively than 2-5A in reason- able agreement with an earlier study (13); however, in Ehrlich ascites tumor cell extracts the analogue’s affinity for RNase L was 1000 times less than 2-5A. In a rRNA cleavage assay, the cordycepin analogue was found to be devoid of any activity in Ehrlich ascites tumor cell extracts. Using the same cleavage assay in L929 cell extracts, rRNA cleavage could be detected when a 10-fold higher concentration of ppp5’(3’dA)2’p5’- (3’dA)2’p5’(3’dA) compared to 2-5A was used however, this must be regarded as a minimum difference in activity since the intensity of the cleavage product bands was significantly less than with 2-5A (15).

In vivid contrast to the above studies, Suhadolnik et al. (20) and Lee and Suhadolnik (21) have reported that the tetramer ~pp5‘(3‘dA)~3‘dA is a more potent inhibitor of translation in rabbit reticulocyte lysate than 2-5A itself and that the 3’- deoxyadenosine 2-5A analogues, both trimer and tetramer ppp(3’dA)n3’dA, were as active as 2-5A itself as protein synthesis inhibitors in mouse L cells and human fibroblasts when calcium phosphate coprecipitation was employed to introduce the oligomers.

A completely separate matter is the biological activity of 2- 5A core and its analogues, specifically A2’p5‘A2’p5’A and (3’dA)2’~5’(3’dA)2’~5‘(3’dA). These 5’-unphosphorylated 2‘,5’-linked oligonucleotides may tend to become confused with their 5’-phosphorylated counterparts such as 2-5A itself, ppp5’A2‘p5’A2’p5‘A. In fact at this time there is no solid evidence to support any concept that core oligomers such as A2’p5‘A2’p5’A can be phosphorylated to 2-5A-related mole- cules, pnA2’p5’A2‘p5’A. The biological activities of such core molecules as A2’p5‘A2‘p5‘A and (3’dA)2’~5’(3’dA)Z’p5’- (3’dA) have been thoroughly documented (22-24), but the evidence does not support an involvement of the 2-5A system in their mechanism of action (24). In the case of the 3’- deoxyadenosine oligonucleotide, certain reports (25, 26) have ascribed its activity to degradation to the established cytotoxic nucleoside, cordycepin, or its nucleotides.

With the exception of three reports (9, 20, 21) there seems to exist a consensus among other investigators that the com- pletely 3’-deoxyadenosine-substituted 2-5A congener, ppp5’- (3’dA}2’~5’(3’dA)2‘~5‘(3‘dA) and the corresponding tetra- mer, are significantly less active both in binding to RNase L from a variety of sources and in the activation of RNase L of a number of sources. The only disparities among these latter reports relates to the extent of inactivity with estimates that the 3‘-deoxyadenosine derivatives are at least 10-fold to more than 100,000-fold less active than 2-5A. Clearly, the results are dependent on the nature of the assay system and the source of cell extract. With one possible exception (15), the RNase L of mouse cells was poorly activated by the cordycepin analogue. The lack of activation also holds for the RNase L from human cells. On the other hand, the rabbit reticulocyte enzyme seems to be more readily activated by the 3“deoxy- adenosine congener. Those variations in activity (or lack of activity) may also be related to the fact that the 3‘-deoxy- adenosine analogues are considerably more stable to degra- dation than 2-5A (13-15). This may act to effectively increase the applied concentration of oligomer relative to 2-5A, thus increasing its apparent relative activity. This could certainly be the case in an intact cell system such as that used in the study of Eppstein et al. (16). The resistance of such cordycepin 2-5A analogues to degradation also might, through inhibition of the 2-5A phosphodiesterase, serve to enhance the activity of the presence of any endogenously formed 2-5A. Finally, in

Biological Roles of the 3‘-Hydroxyl Groups of 2-5A 1135

all systems using a calcium salt coprecipitation method, the actual concentration of oligomer entering the cell may not be constant as the oligonucleotide structure is altered.

Because in most studied systems total replacement of the adenosine nucleoside residues of 2-5A with 3’-deoxyadenosine led to a significant drop in biological activity, we felt it would be of value of pinpoint the specific role of each of the 3’- hydroxyl moieties of 2-5A trimer in determining the ability of an oligonucleotide to bind to an activate RNase L. Not only might this information be valuable in understanding how RNase L operates, it also could provide valuable information for the design of 2-5A congener capable of entering intact cells and activating RNase L without the use of manipulations such as coprecipitation, hypertonic salt treatment, microin- jection, or liposomes (reviewed in Ref. 27).

The results presented herein show clearly that each of the three 3‘-hydroxyl groups of the trimeric oligonucleotide, ppp5’A2‘p5‘A2‘p5’A, assumes a significantly different role in its contribution to the biological activity of the 2-5A molecule. When the 3“hydroxyl functionality of the 5’-ter- minal adenosine residue of 2-5A trimer was replaced by hy- drogen, the resulting analog, ppp5’(3‘dA)2‘~5‘A2’~5‘A, was bound to RNase L just as effectively as the parent 2-5A itself (Table 1). Simultaneously, the ability of ppp5’(3’dA)2’p5’- A2’p5‘A to activate the 2-5A-dependent endonuclease, as judged by its ability to cause RNA degradation or to inhibit protein synthesis, was diminished only by a factor of about 3 with respect to 2-5A trimer triphosphate. Thus, the 3‘-hy- droxyl group of the 5”terminal adenosine nucleotide residue of 2-5A trimer must not play a significant role in RNase L binding and only a marginal role at best in the activation of RNase L.

A wholly contrary story emerged upon examination of the analog in which the 3‘-hydroxyl group of the second (from the 5”terminus) nucleotide residue of 2-5A was substituted by hydrogen. In this instance significant decreases in binding and activation occurred. The binding efficiency of ppp5’A2’- p5’(3’dA)2’p5’A decreased by a factor of approximately 8 relative to 2-5A trimer. Even more dramatic, however, was the loss of ability to activate RNase L as judged by either protein synthesis inhibition (300-fold decrease in activity relative to 2-5A) or by ability to cause degradation of poly(U) (1000-fold decrease in activity relative to 2-5A). From these data, it can be concluded that the 3’-hydroxyl moiety of the second (from the 5”terminus) residue of 2-5A makes some contribution to binding of 2-5A to RNase L but an even more vital contribution to the activation of RNase L.

When the 3‘-hydroxyl substituent was converted to hydro- gen in the third or 2’-terminal nucleotide residue of 2-5A, a significant increase in RNase L binding as well as activation ability occurred. According to the results of the radiobinding assay, ppp5’AZ’p5’A2’~5’(3’dA) underwent an approximate 5-fold increase in binding affinity. This was accompanied by a 3-10-fold increase in ability to activate the 2-5A-dependent endonuclease as ascertained by the translational inhibition assay or by the core-cellulose poly(U) degradation assay.

That the relative behavior of the synthetic analogs might be ascribed to differences in their degradation compared to 2- 5A was unlikely based on a study of their degradation under protein synthesis conditions (28). The results of this study revealed the following order of stability (most stable to least stable) along with the corresponding half-lives: p5’(3’dA)- 2’~5’(3’dA)Z’p5’(3’dA) ( t , >> 120 min) -p5’A2’p5’(3’dA)- 2‘p5‘A (tlh >> 120 min) > pS’A2’pS’A2’~5’(3’dA) ( t , = 140 min) > p5’(3’dA)2’p5’A2’p5’A (tlh = 90 min) 2 p5’- A2‘p5‘A2’p5‘A (tn = 70 min). Thus, the analog with the least

biological activity was among the most stable toward degra- dation. The small but significant increase in stability of p5’A2’p5’A2‘p5’(3’dA) (27) may partially explain the en- hanced biological activity of ppp5’A2’~5‘AZ‘p5’(3‘dA), al- though alternate explanations are possible.

The results of this study can be compared favorably to the results of previous investigations (13-16,29). First of all since replacement of the 3’-hydroxyl of the first nucleotide of 2-5A with hydrogen did not affect endonuclease binding and had only a minor effect on nuclease activation and since a similar substitution in the second nucleotide residue led to an %fold reduction in binding but a 500-1000-fold drop in nuclease activation potential, it might be expected that the combined substitution of hydrogen for hydroxyl in both the first and second nucleotides of 2-5A (uiz. ppp5’(3’dA)2’~5’(3’dA)- 2’p5’A) would result in an oligomer with reasonable binding affinity but poor nuclease activation properties. In fact, ppp5’(3’dA)2’~5’(3’dA)2’~5’A was found to bind to L cell RNase L 2-fold less effectively that 2-5A (29), and its behavior as a protein synthesis inhibitor produced data nearly the same as that found for ppp5‘A2’p5‘(3‘dA)2‘p5’A as detailed in Table 1. Second, an earlier study (13) showed that the 3’- deoxyadenosine-substituted analog, ppp5‘(3‘dA)2‘~5‘(3‘dA)- 2’p5‘(3’dA), was bound to the L cell 2-SA-dependent endo- nuclease about 10 times less effectively than 2-5A itself. Comparison of this previous binding data (13) with the bind- ing data obtained for the three sequence-specific 3”deoxy- adenosine analogs of the present study would suggest that a major factor for the diminished binding ability of ppp5’(3’dA)2’~5’(3’dA)2’~5‘(3’dA) was the lack of a 3’- hydroxyl group in the second nucleotide residue. The inability of ppp5’(3’dA)2’~5’(3’dA)2’~5’(3’dA) to activate the 2-5A- dependent endonuclease (13-16, 29) would be predicted on the basis of the above results. It would be reasonable to predict that the structure ppp5’(3’dA)2’~5’A2‘~5’(3’dA) would show a comparable binding affinity and RNase L activation ability to 2-5A itself. Recently, we have prepared this analogue and have found that it binds to RNase about 2-3 times less effectively than 2-5A. In addition, it is 50 times more effective as an RNase L activator as ppp5’A2’p5’(3’dA)2’p5’A and 10 times less active than 2-5A.4

Previous studies (3, 30) on 2-5A analogs in which each nucleotide was sequentially substituted either by inosine or by 7-deaza-adenosine revealed the importance of the first and third nucleotide residues of 2-5A in the binding and/or acti- vation process. On the other hand, at least with respect to these specific substitutions, the adenine ring of the second nucleotide unit played no role in the binding/activation proc- ess. The result described herein clearly implicates the vital interaction of the ribose of the second nucleotide of 2-5A with some domain in RNase L. At the same time it is also of interest that no dramatic role is played by either of the 3’- hydroxyl moieties associated with nucleotides which possess bases definitively involved in the binding or activation proc- ess.

The synthesis of sequence-specific 3“deoxyadenosine-sub- stituted analogs of 2-5A has demonstrated the following. (i) Only the 3’-hydroxyl group of the second from the 5”termi- nus nucleotide of 2-5A is needed for effective activation of RNase L. (ii) Substitution of hydrogen for the 3”hydroxyl group of the penultimate nucleotide unit of 2-5A confers substantial 2’-phosphodiesterase resistance on the molecule (28). (iii) Further synthetic strategies designed to uncover a 2-5A derivative which may effectively penetrate the intact

R. Charubala, W. Pfleiderer, D. Alster, D. Brozda, and P. F. Torwnce, unpublished observations.

1136 Biological Roles of the 3'-Hydroxyl Groups of 2-5A

cell need not deal with the 3"hydroxyl groups of all ribonu- cleoside/ribonucleotide building blocks, and this may permit a simpler synthetic access to such materials.

1.

2.

3.

4.

5.

6.

7.

8. 9.

10.

11.

12.

REFERENCES Kerr, I. M., and Brown, R. E. (1978) Proc. Natl. Acud. Sci. U. S.

Torrence, P. F. (1985) in Biological Response Modifers. New Approaches to Disease Intervention (Torrence, P. F., ed) pp. 77-105, Academic Press, Orlando, FL

Imai, J., Lesiak, K., and Torrence, P. F. (1985) J. Biol. Chern.

Pfleiderer, W., Schwarz, M., and Schirmeister, H. (1986) Chern.

Himmelsbach, F., Schulz, B. S., Trichtinger, T., Charubala, R.,

Mukaiyama, T., and Hashimoto, M. (1971) Bull. Chern. SOC. Jpn.

Knight, M., Wreschner, D. H., Silverman, R. H., and Kerr, I. M.

Silverman, R. H. (1985) Anal. Biochern. 144, 450-460 Doetsch, P., Wu, J. M., Sawada, Y., and Suhadolnik, R. J. (1981)

Charubala, R., and Pfleiderer, W. (1980) Tetrahedron Lett. 4077-

Williams, B. R. G., Golgher, R. R., Brown, R. E., Gilbert, C. S.,

Silverman, R. H., Wreschner, D. H., Gilbert, C. S., and Kerr, I.

A. 75,256-260

260,1390-1393

Scr. 26, 147-154

and Pfleiderer, W. (1984) Tetrahedron 40, 59-72

44,2284

(1981) Methals Enzymol. 79, 216-227

Nature 291,355-358

4080

and Kerr, I. M. (1979) Nature 282,582-586

M. (198i) Eur..J. Biochern..115, 79-85 13. Sawai, H., Imai, J., Lesiak, K., Johnston, M. I., and Torrence, P.

14. Haugh, M. C., Cayley, P. J., Serafinowska, H. T., Norman, P. G.,

84 Reese, C. B., and Kerr, J. M. (1983) Eur. J. Biochern. 132,77-

15. Nyilas, A., Vrang, L., Drake, A., Oberg, B., and Chattopadhyaya, J. (1986) Acta Chern. Scand. Ser. B Org. Chern. 40,678-688

F. (1983) J. Biol. Chem. 268, 1671-1677

16. Eppstein, D. A., Van Der Pas, M. A., Schryver, B. B., Sawai, H., Lesiak, K., Imai, J., and Torrence, P. F. (1985) J. Biol. Chern.

17. Martin, E. M., Birdsall, N. J. M., Brown, R. E., and Kerr, I. M.

18. Lesiak, K., and Torrence, P. F. (1985) Biochern. Biophys. Res.

19. Krause, D., Lesiak, K., Jmai, J., Sawai, H., Torrence, P. F., and Silverman, R. H. (1986) J. Biol. Chern. 261, 6836-6839

20. Suhadolnik, R. J., Devash, Y., Reichenbach, N. L., Flick, M. B., and Wu, J. M. (1983) Biochern. Biophys. Res. Cornrnun. 11 1,

21. Lee, C., and Suhadolnik, R. J. (1985) FEBS Lett. 167,205-209 22. Doetsch, P. W., Suhadolnik, R. J., Sawada, Y., Mosca, J. D.,

Flick, M. B., Reichenbach, N. L., Dang, A. Q., Wu, J. M., Charubala, R., Pfleiderer, W., and Henderson, E. E. (1981) Proc. Natl. Acud. Sci. U. S. A. 78,6699-6703

23. Henderson, E. E., Doetsch, P. W., Charubala, R., Pfleiderer, W., and Suhadolnik, R. J. (1982) Virology 122,198-201

24. Suhadolnik, R. J., Doetsch, P. W., Devash, Y., Henderson, E. E., Charubala, R., and Pfleiderer, W. (1983) Nucleosides & Nucleo- tides 2,351-366

25. Chapekar, M. S., and Glazer, R. I. (1983) Biochern. Biophys. Res. Cornrnun. 115, 137-143

26. Eppstein, D. A., Marsh, Y. V., and Schryver, B. B. (1983) Virology

27. Johnston, M. I., and Torrence, P. F. (1984) in Interferon. Volume 3. Mechanisms of Production and Action (Friedman, R. M., ed) pp. 189-298, Elsevier Scientific Publishing Co., Amsterdam

28. Alster, D., Brozda, D., Kitade, Y., Wong, A., Charubala, R., Pfleiderer, W., and Torrence, P. F. (1986) Biochern. Biophys. Res. Cornrnun. 141,555-561

29. Eppstein, D. A., Marsh, Y. V., Schryver, B. B., Larson, M. A., Barnett, J. W., Verheyden, J. P. H., and Prisbe, E. J. (1982) J.

30. Jamoulle, J. C., Lesiak, K., and Torrence, P. F. (1987) Biochern-

260,3666-3671

(1979) Eur. J . Biochern. 95,295-307

Commun. 126,917-921

205-212

131,341-354

Biol. Chem. 257, 13390-13397

i s t v 26,376-383

Biological Roles of the 3'-Hydroxyl Groups of 2-5A 1137

1138 Biological Roles of the 3'-Hydroxyl Groups of 2-5A

Biological Roles of the 3"Hydroxyl Groups of 2-5A