Synthesis of (2’-5’)(A)n from ATP CHARACTERISTICS OF THE REACTION CATALYZED BY (2’-5’)(A)n SYNTHETASE PURIFIED FROM MOUSE EHRLICH ASCITES TUMOR CELLS TREATED WITH INTERFERON*

The treatment of Ehrlich ascites tumor cells with mouse interferon increases the level of the latent enzyme (2’-5’)(A), synthetase. If activated by double- stranded RNA, this catalyzes the synthesis from ATP of a series of 2‘-5’-oligoadenylates: (2’-5’)(A), where n extends from 2 to about 15. We isolated (2’-5’)(A), synthetase in a homogeneous state. In the presence of double-stranded RNA, the purified enzyme can convert the large majority (about 97%) of the ATP into (2’- 5’)(A),, and pyrophosphate, although it does not cleave the pyrophosphate. The stoichiometry of the reaction can be formulated as: (n + I) ATP ”* (2’-5’) PPPA(PA)n + n pyrophosphate. Added pyrophosphate does not inhibit the synthesis of (2’-5’)(A),. The extent of the reverse reaction, ie. the pyrophosphorolysis of (2’-5’)(A),,, was below the level of detection under our conditions. The affinity of the enzyme for ATP is low: the reaction 10% when the of is increased from 5 m~ to 10 m ~ . The optimal concentration of double-stranded RNA increases the the enzyme. As tested

tary fractions. One of these fractions when incubated with dsRNA and ATP was shown to give rise to a small thermostable product. This in turn was found to activate a latent endonuclease, now designated as RNase L, in the other complementary fraction (17, see also Refs. 18 and 19). The small thermostable product was identified as (2'-5')(A),,2 a series of compounds originally discovered as inhibitors of protein synthesis that are formed from ATP in extracts of interferontreated cells in the presence of dsRNA (20). (2'-5')(A),, was detected in interferon-treated virus-infected cells (21). (2'-5')(A), has also been introduced into cells and was shown to cause a transient nuclease activation and an inhibition of protein synthesis and virus replication (22)(23)(24). (2'-5')(A),, introduced into lymphocytes was found to impair mitogeninduced DNA synthesis (25).
(2'-5')(A)n synthetase has been detected in Ehrlich ascites tumor cells, HeLa cells, and chick embryonic fibroblasts treated with interferon as well as in mouse lymphocytes and rabbit reticulocytes, even if not previously exposed to interferon (17,18,(26)(27)(28). The level of the enzyme was reported to increase in chicken oviducts after withdrawal of estrogens (29). The enzyme from chick embryonic fibroblasts has been partially purified by conventional procedures (16) and the enzyme from L cells by affinity chromatography on dsRNA ( 10).
We have described earlier the isolation of pure (2'-5')(A)n synthetase from EAT cells treated with interferon. As determined by polyacrylamide gel electrophoresis in the presence of SDS, the apparent molecular weight of the enzyme was around 105,000. The activity of the enzyme was enhanced by dsRNA a t least 25-fold. The size distribution of the (2'-5')(A)" synthesized extended from the dimer to the pentadecamer (31). Here we report on further characteristics of the purified enzyme, and of the reaction it catalyzes.
It should be noted that pppA(pA), and (2'-5')(A)" stand for the same compound. The formula pppA(pA), is used when the emphasis is on the 5"terminal triphosphate moiety of the compound, and (2'-5')(A), is used in other cases. We use n (to indicate chain length) in both cases to avoid unnecessary complication. In precise terms, however, (2'-5')(A)n will correspond to pppA(pA),, ~ l .

RESULTS
Kinetics of (.Y-Y)(A), Synthesis: Optimization of Reaction Condition-The curves in Fig. 1 indicate that under the conditions of the incubation the rate of (2'-5')(A),, synthesis decreases with time. This is apparently in consequence of the inactivation of the purified enzyme used at low concentrations. The inactivation cannot be overcome by adding bovine serum albumin (1 mg/ml) to the reaction mixture and it also occurs when the enzyme is incubated without ATP (not shown).
In consequence of the inactivation of the enzyme during the incubation and the need for the formation of sufficient amounts of (2'-5')(A), for precise assay, we optimized the composition of the reaction mixture for the following conditions: synthesis from 1 nm ATP with 2 p g / d of enzyme at 30°C for 2 h. For this optimization, we started with the assay conditions used in the purification of the enzyme and varied ' Portions of the paper (including "Experimental Procedures" and Miniprint is easily read with the aid of a standard magnifying glass. Figs. 1 to 7) are presented in rniniprint as prepared by the authors.
Full-size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, Md. 20014. Request Document No. 80M-809, cite authors, and include a check for $1.50 per set of photocopies. Full-size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press. connected With discontinuous lines is displaced from the scale for the data points connected with continuous lines.
the concentration of the various components of the reaction mixtures one by one. We found that the optimal concentration of KC1 is 125 mM. The replacement of KC1 by equimolar KOAc results in about an 18% decrease in (2'-5')(A),, synthesis. However, with KOAc at its optimal concentration (225 m), the yield of (2'-5')(A)n is about 25% higher than with KC1 a t its optimal concentration (not shown). The (2'-5')(A), synthesis uersus pH curve shows a peak at about pH 7.8 when determined in the presence of either KOAc or KC1 (Fig. 2). The optimal temperature of the reaction is 25°C; the yield of (2'-5')(A), is 10% lower at 30°C and 55% lower at 37°C (not shown). Nevertheless, since it is difficult to control the temperature of incubation a t 25°C (which is close to room temperature), unless otherwise specified we performed the incubations a t 30°C.
When assayed a t 1 ,m ATP concentration, the optimal Mg acetate concentration is 8 m whether tested a t 30°C (Fig.   3A) or 25°C (not shown). An increase in the concentration of ATP necessitates a similar increase in Mg'+ concentration to keep the conditions optimal (Fig. 3A). Thus, in the presence of 5 m ATP, the optimal Mg'+ concentration is 12 mM and, at 10 m ATP, the optimal Mgz+ concentration is 16 mM.
The affinity of the enzyme for ATP is low; (2'-5')(A), synthesis (as determined at the optimal Mg") increases about 3.1-fold when the concentration of ATP is increased from 1 mM to 5 m and about 3.5-fold when increased from 1 mM to 10 m (Fig. 3B). The variation of the optimal Mg'+ concentration with the concentration of ATP and the instability of the enzyme make it difficult to determine a precise K, for ATP. The data in Fig. 3B seem to indicate that the concentration of ATP at which (2'-5')(A),, synthesis is half-maximal is about 2 mM.
The optimal concentration of dsRNA (i.e. poly(1)poly(C) or reovirus dsRNA in our experiments) increases with the enzyme concentration ( Fig. 4.4); in the presence of 0.4, 2, and 10 pg/ml of enzyme, the optimal poly(I).poly(C) concentrations appear to be about 0.12,1.2, and 7 p g / d and the optimal reovirus dsRNA concentrations about 0.23, 0.67, and 1.6 pP/ ml. Thus, the optimal ratios of dsRNA concentration/eci;e concentration (both in microg-rams/ml) in the above cases are 0.30,0.62,0.70,0.57,0.48, and 0.16. Since the peaks of optimal &RNA concentrations are rather broad (see Fig. 4), only the following approximation can be proposed: (2'-5') (A), synthesis is close to optimal if the concentration of poly(1)poly(C) or to the enzyme concentration (Fig. 5); the shape of the curve of the rate versus enzyme concentration is sigmoidal. It will have to be established if this is a consequence of a cooperative interaction between enzyme subunits. Using as a basis for our calculation the amount of (2'-5')(A),, synthesized by 2 pg/ml of enzyme from 10 m ATP at 16 mM Mg'+ in 2 h (see Fig.   3A), we calculated the turnover number of (2'-5')(A)n synthetase. The number obtained is 6.9 molecules of ATP converted to (2'-5')(A),/enzyme molecule/s. This is an underestimation of the real value because the rate of the reaction decreases during the incubation in consequence of the inactivation of the enzyme (see Fig. 1).
(2'-5')(A),, Synthetase Activated by dsRNA Catalyzes the Release of PP, from ATP-(2"5')(A), synthetase catalyzes the release of PPi from [y3*P]ATP. This release depends on the presence of both enzyme and dsRNA ( Table I) reovirus dsRNA in the reaction mixture is half as much as that of the enzyme. The validity of this approximation was not tested, however, for cases in which the enzyme concentration was below 0.4 pg/ml or above 10 pg/ml. The stimulation of (2'-5')(A),, synthetase activity decreases at dsRNA concentrations above the optimal (Fig. 4). This decrease cannot be overcome by increasing the concentration of Mg'+ (not shown).
Although E. coli 16 S ribosomal RNA (36) and mouse EAT cell tRNA contain double-stranded segments (37), if tested at 1, 3, 10, 30, or 100 pg/ml, they do not substitute for poly(1). poly(C) in activating the enzyme. Moreover, at the above concentrations, these RNAs do not decrease the activity of the enzyme in the presence of 5 pg/ml of poly(I).poly(C) (not shown).
The rate of (2'-5')(A)" synthesis is not linearly proportional tare at 0 . 4 ( e ) , 2 [ A ) , OT 10 uglml (.) and the concentrations of The incubation was for 1 ( I ) , This indicates that the enzyme, even if activated, does not cleave PP, to Pi. Stoichiometry of (2"5')(A), Synthesis-To determine the molar ratio between ATP cleaved, (2'-5') (A), synthesized, and PPi released, we performed the following sets of reactions (Table 11): we incubated a mixture of [y-'"P]ATP and [a-3'PIATP (in a ratio of cpm/cpm = 1.08:l) with (2'-5')(A), synthetase with (or without) dsRNA. This resulted in the formation of (2'-5')(A), (a-and y-labeled) and PPi (y-labeled). Chromatographic fractionation and counting of the radioactivity in the fractions allowed the determination of the amount of radioactivity in the PPi released in response to the activation of the enzyme by dsRNA. This corresponded to 21% -3.2% = 17.8% of the total radioactivity in the reaction mixture.

The nonuniform length of the incubatlonr w a s made necessary by the large Varlstion in enzyme concentrations
To determine the amount of (2'-5')(A),, synthesized, we treated an incubated reaction mixture with alkaline phosphatase to remove the 5"terminal triphosphate from pppA(pA), and convert it to core (A(pA),, a-labeled). The amount of radioactivity in the core was 16% of the total radioactivity in the reaction mixture. Based on these data, we calculated the molar proportion of PPi released to AMP moieties incorporated into the core moiety of (2'-5')(A),: 17.8%:16% = 1.1. Correcting this ratio by division with the factor 1.08 (the ratio of y3'P to w3*P label) gives 1.1:1.08 = 1.03. This indicates that for every mole of AMP residues incorporated into (2'-5')(A), core, 1 mol of PPi is released.
The Equilibrium of the Reaction Catalyzed by (2"5')(A), Synthetase Favors Synthesis-With enzyme at a low concentration (e.g. 0.5 pg/ml), only a small part of the ATP added was converted to (2'-5')(A),, in our conditions (Fig. 1). This is apparently in consequence of the inactivation of the enzyme during incubation. About 90% of the ATP was converted to (2'-5')(A), when the enzyme was used at an intermediate concentration (2 pg/ml) and the reaction mixture was supplemented with Triton X-100 (0.256, v/v) to accelerate the reaction (Fig. 1). When tested at a high concentration (11 pg/ml), TABLE I Dependence ofpyrophosphate formation from A T P on (2'-5') (A), synthetase a n d dsRNA Reaction mixtures A, B, and C: the standard reaction mixtures were supplemented with 17 mM Hepes/KOH (pH 7.8), 225 mM KOAC, 8 m~ Mg(OAc)n, 1 mM [y-"P]ATP, and, if so indicated, ( 2 ' 4 ' ) (A), synthetase (2 pg/ml) or poly(I).poly(C) (5 pg/ml) or both. The incubation was for 2 h. Reaction mixture D: [y-."PP]ATP was incubated with 4 pg/ml of snake venom diesterase in 20 mM Tris-HCI (pH 8.8). 20 mM Mg(OAc)p at 30°C for 2 h to convert it to PP, (labeled) to serve as a chromatographic marker and AMP (unlabeled). Each reaction mixture was then applied to a PEI plate which was developed with 0.75 M K phosphate (pH 3.5). In this solvent, ATP and (2'-5') (A). migrate together. The labeled spots were visualized by radioautography, cut, eluted with 1 ml of 1 N HC1, and counted. The [y-"PIATP preparation used was slightly contaminated with P, and with a slow moving unidentified compound indicated as (?).

'-5') (A). synthesis
The standard reaction mixtures were supplemented with 17 mM Hepes/KOH (pH 7.8), 225 m~ KOAC, 8 m~ Mg(OAc)?, 2 pg/ml of (2'-5')(A)" synthetase, 1 mM ATP, y-"'P-and a-,"P-labeled (in a ratio of 1.08 cpm of y label/l cpm of a label), and 5 pg/ml of poly(1). poly(C) if so indicated. The incubation was for 2 h. Thereafter, the reaction mixture C was supplemented with an equal volume of a solution of 8 units/ml of bacterial alkaline phosphatase in 20 mM Tris-HCI (pH 7.5) and incubated at 30°C for 2 h (to convert pppA(pA), into A(pA).-designated as core, a-labeled). Each reaction mixture was thereafter applied to a PEI plate which was developed with 0.75 M K phosphate (pH 3.5). The labeled spots were visualized by radioautography, cut, eluted with 1 ml of 1 N HCI, and counted. The [y-"'P]ATP used was slightly contaminated with a substrate of low mobility, this is indicated as (?) in the table. Total 157,400 154,400 106,700 'I The amount of P, increased during the incubation in the complete reaction mixture in this experiment. We do not know the cause of example Table I and Table 111).

this. No such increase occurred in several other experiments (see for
' The numbers in parentheses are per cents of the total counts per min in the form of PP, or core, respectively. the enzyme can convert to (2'-5')(AL the large majority of ATP: 96.3% in one experiment and 98% in a second experiment (Table 111).
As shown earlier (Table I), PP, is formed during the reaction and is not cleaved further by the enzyme. The accumulated PPi is, however, not inhibitory: even a t a concentration as high as 7.5 mM it does not decrease the rate of conversion of 1 mM ATP to (2'-5')(A)n if tested in the presence of 16 mM M$' (Fig. 6). If assayed a t 8 mM Mg", even 2.5 mM PPi impairs the activity of the enzyme. This occurs most probably in consequence of the complexing of Mg2+ ions by PPi.
All these results indicate that the equilibrium of the reaction catalyzed by (2'-5')(A)" synthetase is strongly in favor of synthesis.
Effects of Incubation Conditions on the Size Distribution of (2"5')(A),-As reported earlier, the chain length of (2'-5')(A)n produced by the purified enzyme extends from the dimer to about the pentadecamer (31). The size distribution

Equilibrium of (2'-5') (A), synthesis
The standard reaction mixtures were supplemented with 17 mM Hepes/KOH (pH 7.5), 125 m~ KCI, 8 mM Mg(OAc)s, 1 mM [a-32P]ATP, l l p g / d of (2'-5') (A), synthetase, and, if so indicated, 5 p g / d of poly(I).poly(C) and/or 0.2% (v/v) Triton X-100. (Triton X-100 was found to accelerate the reaction approximately 2-fold.) Incubation was for 6 h. Thereafter, the reaction mixtures were heated at 95°C for 3 min to inactivate the enzyme, and were cooled and clarified by a 15-s centrifugation in the microfuge (Eppendorf). To convert (the unreacted) ATP to ADP, aliquots of the clarified solutions were supplemented with an equal volume of a second solution to give a final concentration of 0.1 p g / d of hexokinase (EC 2.7.1.1., Sigma, type VI, 75 units/mg), 10 m~ glucose, and 5 mM Mg(OAc)2 and were incubated at 30°C for 30 min. (This conversion of ATP was needed because, in the chromatographic analysis, the bulk of (2'-5') (A), (i.e. dimers to pentamers) is separated from ADP but not from ATP.) Aliquots from each reaction mixture were applied to a PEI plate which was developed with 0.75 M K phosphate (pH 3.5). P,, AMP, ADP, and ATP were used as region markers. The chromatogram was cut into regions which were eluted with 1 ml of 1 N HC1 and counted. Substances eluted from the AMP and ADP regions (in track A) were further analyzed by homochromatography (31) to separate AMP and ADP from (2'-5') (A), (some of which, less than 20%, also migrates in these regions; not shown). The homochromatogram revealed that 75% of the labeling in the "ADP + AMP" region is in (2'-5') (A),4 (mostly in the hexamer to hexadecamer range) and 25% in ADP + AMP. We also established that 4% of the total labeling in the [a-"PIATP preparation was present as ADP and AMP already at the start of the incubation (not shown). The substance designated as (?) was present as an impurity in the [L~-~'P]ATP and was not identified. There was no discrete labeled spot in the region termed track background. Calculation of the per cent of ATP converted to ( ground of (2'-5')(A)n is affected by the conditions of incubation. At 25 mM KC1 or KOAc concentration, the proportion of oligoadenylates of intermediate size (pentamer to decamer) is lower and those of larger size (undecamer and above) is higher than at 225 mM KOAc concentration. Furthermore, the proportion of longer oligoadenylates increases together with the proportion of ATP converted to (2'-5')(AL (not shown).
The homochromatographic analysis of the AMP and ADP regions of the thin layer chromatogram was performed only in this experiment. This is because: 1) in consequence of the almost complete conversion of ATP to (2'-5')(A),,, as much as 16% of the (2'-5')(A)" was hexaadenylate or larger (see the relevant section under "Results") and 2) these large oligoadenylates (but not the shorter ones) move into the AMP and ADP regions. In other experiments, much less of the ATP was converted to (2'-5')(A)" and thus little of the (2'-5')(A),, was larger than the pentaadenylate. Products of the Incubation of (2"5')(A), Synthetase with 2'-DeoxyATP and 3'-DeoxyATP-To establish if 2"deoxyATP or 3'-deoxyATP are substrates for (2'-5')(A)n synthetase, we incubated each of these compounds (unlabeled) at a 5 mM concentration separately with [a-"'P]ATP a t a low concentration (2 PM), enzyme, and dsRNA and then digested the reaction mixture fist with alkali and then with bacterial alkaline phosphatase (Fig. 7). The treatment with alkali degrades (2'-5')(A), to pppAp, 2'-AMP, and 3'-AMP (Fig. 7, track C) and further treatment with the phosphatase produces P, as the only labeled product (Fig. 7, track F ) . 2"DeoxyAMP or 3'-deoxyAMP moieties in oligoadenylates can be detected because the phosphodiester linkage connecting the 2'-or 3'-OH of such a moiety to the adjacent adenylate (or deoxyadenylate) moiety is resistant to cleavage by alkali.
(This is the case since alkaline hydrolysis of phosphodiester linkages in nucleic acids requires the participation of both 2'and 3'-hydroxyl group of the ribose moieties.) Thus, the treatment with alkali of a n oligoadenylate containing a deoxyadenylate residue (although not in the 3'-terminal pogtion) should generate oligonucleotides. For example, pppApAP would be produced if a n oligoadenylate digespd contained one deoxyadenylate moiety at its 5' terminus. (A stands for a 2'or 3'-deox adenosine moiety.) The data in Fig. 7 reveal that the ppp 2 pAp is produced upon the treatment with alkali of (2'-5')(A),, synthesized in the presence of 3'-deoxyATP but not of that produced in the presence of 2'-deoxyATP. Moreover, as expected, the treatment of the pppEfpAp with bacterial alkaline phosphatase results in the conversion to EfpA (Fig. 7, track E ) . These data indicate that in our conditions (2'-5')(A)n synthetase can link adenylate moieties to the 2'hydroxyl of 3'-deoxyATP, but not to the 3'-hydroxyl of 2'-deoxyATP. 3'-DeoxyATP is, however, a poor competitor of A T P in initiating (2'-5')(A),, chains. The numbers of chains initiated by the two nucleotides are similar although the concentration of 3"deoxyATP was 2500 times higher than that of A T P i n t h e reaction mixture (Fig. 7, track B ) . The attachment of 3'-deoxyadenylate moieties (if any) into internal positions of oligoadenylates was below the level of detection in the test.
We tested in one experiment if the enzyme can use CTP, G T P o r U T P as substrate. Using the standard reaction conditions (2.e. 2 pg/ml of enzyme, 1 mM ribonucleoside triphosphate), but a 13-n mcubation, we found a slow oligomerization of UTP tested alone and a somewhat faster oligomerization of U T P or CTP when tested in the presence of 1 m~ A T P . T h e and 2 UBI rm.3iPIATP.
Moreover the reaction mixture for track A 1.1 also SvPPlesented with 5 mU Z'-deoxy ATP. t h a t for track 8 with 5 mU )'-deoxy ATP and that for track C with 5 mU ATP. Incubation was for 1 7 hours.

DISCUSSION
In our studies on purified (2'-5')(A)n synthetase, we used poly(I).poly(C) in solution rather than bound to paper (29,31). We have chosen this although the enzyme is (at least if tested at low concentrations, i.e. about 11 pg/ml or lower) much less stable if activated by poly(I).poly(C) in solution than if activated by paper-bound poly(1) -poly(C) (not shown). We have accepted the lesser stability because the initial velocity of the enzyme is about 22 times higher in the presence of free poly(1) .poly(C) than in that of paper-bound poly(1) poly(C) (not shown).
As noted earlier, maximal or close to maximal activation of the enzyme in our conditions requires poly(1) -poly(C) or reovirus dsRNA at about half the concentration of the enzyme if expressed in terms of w/v. Taking the molecular weight of the enzyme to be 105,000 (Ref. 31; as determined by polyacrylamide gel electrophoresis in the presence of SDS) and the molecular weight of a base pair in poly(I)-poly(C) as 668, it can be calculated that this corresponds to about 79 base pairs in RNA/molecule of enzyme. This ratio is in line with the finding that polv(I).poly(C) has to be at least 65 to 80 base pairs long to activate the enzyme maximally (30).
The specific activity of the enzyme decreases if the concentration of dsRNA is increased above the optimal value. It remains to be established whether or not this reflects the need for the binding of 2 or more enzyme molecules to adjacent sites on dsRNA for maximal activity, i.e. cooperativity. (The probability of the binding to adjacent sites on dsRNA of 2 or more enzyme molecules might be diminished if the dsRNA concentration is increased above the level allowing the attachment to dsRNA of all the enzyme molecules.) The sigmoidity of the plot of (2'-5')(A)n produced uersus enzyme concentra-tion (Fig. 5) is consistent with a cooperative behavior. equation for the formation of 1 molecule of (2"5')(A)n The data presented allow the formulation of the following ( n + 1) ATP = (2"5')pppA(pA),, + n PP, The equilibrium of the reaction catalyzed by (2"5')(A)n synthetase is shifted toward synthesis: about 96 to 98% (if not more) of the ATP can be converted to (2'-5')(A)" by the enzyme. This is remarkable since the enzyme does not seem to cleave further the PPI produced in the reaction. It is curious that the extent, if any, of the reverse reaction of (2'-5')(A)n synthesis, i.e. of the pyrophosphorolysis of (2'-f~')(A)~, was below the level of detection in our conditions. Furthermore, we did not detect any exchange of ['"PIPPi into ATP when tested in the presence of (2'-5')(AIn and/or ATP and/or AMP. Elucidation of the molecular basis of these characteristics of the enzyme will require further experiments.