Elucidation of Basic Mechanistic and Kinetic Properties of Influenza Endonuclease Using Chemically Synthesized RNAs*

Influenza virus utilizes a unique mechanism for initi- ating the transcription of viral mRNA. The viral transcriptase ribonucleoprotein complex hydrolyzes host cell transcripts containing the cap 1 structure (m 7 GpppG(2 (cid:42) -OMe)-) to generate a capped primer for viral mRNA transcription. Basic aspects of this viral endonuclease reaction are elucidated in this study through the use of synthetic, radiolabeled RNA sub- strates and substrate analogs containing the cap 1 structure. Unlike most ribonucleases, this viral endonuclease is shown to catalyze the hydrolysis of the scissile phosphodiester, resulting in 5 (cid:42) -phosphate- and 3 (cid:42) -hydroxyl-containing fragments. Nevertheless, the 2 (cid:42) -OH adjacent to the released ribosyl 3 (cid:42) -OH is shown to be important for catalysis. In addition, while the endonuclease steady-state turnover rate is measured to be 2 h (cid:50) 1 , phosphodiester bond hydrolysis is not rate-limiting. The di- rect generation of a free 3 (cid:42) -OH and the subsequent slow release of this product are consistent with the viral need for efficient use of the capped primer in subsequent reactions of the influenza transcriptase complex. slightly faster mobility for the triphosphorylated oligoribonucleotides. The homogeneity of the oligoribonucleotides was analyzed by reverse phase C-18 HPLC (DuPont 8800 instrument coupled to a DuPont 8800 UV detector) employing a linear gradient of acetonitrile (1.4–14% in 20 min, flow rate (cid:53) 1.5 ml/min) in 50 m M aqueous triethylammonium acetate buffer (pH 7.0). Coinjection of control oligoribonucleotides with their corresponding oligoribonucleotide 5 (cid:57) -triphosphates revealed two peaks. The 31 P NMR of the gel purified fully deprotected oligoribonucleotide III showed the characteristic signals of a triphosphate ( (cid:103) P, (cid:50) 8.84 ( d , J (cid:103) P (cid:98) P (cid:53) 18.9 Hz); (cid:97) P, -10.75 ( d , J (cid:97) P (cid:98) P (cid:53) 18.2 Hz); (cid:98) P, (cid:50) 21.99 ( dd , J (cid:98) P (cid:103) P (cid:97) P (cid:53) 18.5 Hz)) in addition to the phosphodiester signals, (cid:50) 0.05 to (cid:50) 0.22 (m). 31 P NMR spectra were recorded at 145.79 mHz on a Bruker WH 360 spectrometer with 1 H decoupling and 85% H 3 PO 4 as the external standard. Samples contained 50% D 2 O and 10 m M EDTA, pH 8. Preparation of Cap 1 RNAs— High specific activity RNA was prepared by adding 16 units of guanylyltransferase to a 100- (cid:109) l reaction containing 50 m M Tris-HCl, pH 8, 1.25 m M MgCl 2 , 6 m M KCl, 2.5 m M dithiothreitol, 5.0 (cid:109) M triphosphorylated oligoribonucleotide Endonuclease Typical biphasic reaction

Influenza virus utilizes a unique mechanism for initiating the transcription of viral mRNA. The viral transcriptase ribonucleoprotein complex hydrolyzes host cell transcripts containing the cap 1 structure (m 7 GpppG(2-OMe)-) to generate a capped primer for viral mRNA transcription. Basic aspects of this viral endonuclease reaction are elucidated in this study through the use of synthetic, radiolabeled RNA substrates and substrate analogs containing the cap 1 structure. Unlike most ribonucleases, this viral endonuclease is shown to catalyze the hydrolysis of the scissile phosphodiester, resulting in 5-phosphate-and 3-hydroxylcontaining fragments. Nevertheless, the 2-OH adjacent to the released ribosyl 3-OH is shown to be important for catalysis. In addition, while the endonuclease steady-state turnover rate is measured to be 2 h ؊1 , phosphodiester bond hydrolysis is not rate-limiting. The direct generation of a free 3-OH and the subsequent slow release of this product are consistent with the viral need for efficient use of the capped primer in subsequent reactions of the influenza transcriptase complex.
Most viral and cellular mRNA molecules contain a methylated cap structure at the 5Ј-end (for reviews see Refs. 1 and 2). The presence of a cap is important for mRNA maturation, initiation of translation, and protection against degradation by RNases present in the cell. The general structure of a capped RNA can be designated as m 7 G(5Ј)ppp(5Ј)Puo-(Puo, the penultimate base, is typically a purine (1)). The penultimate base of cap 1-containing mRNAs is 2Ј-O-methylated and can be designated m 7 G(5Ј)ppp(5Ј)Puo(2Ј-OMe).
The first step in the synthesis of influenza virus mRNA is the binding of host cell nuclear mRNA having a cap 1 structure (3)(4)(5) by the PB2 protein of the viral transcriptase complex (6 -8). Subsequent cleavage of the capped RNA 10 -15 bases from the penultimate nucleoside generates capped oligoribonucleotides that serve as primers in mRNA synthesis by the viral transcriptase (9 -12).
Heretofore, the RNA endonuclease reaction of the influenza transcriptase complex has not been well defined, either in terms of the chemical structure of substrates and products or in the quantification of reaction kinetics. The difficulty has been the unavailability of homogeneous preparations of short substrates containing the cap 1 structure. Initial characterizations of the endonuclease have employed mRNAs isolated from nat-ural sources (10) or prepared by in vitro ''run-off'' transcription, so that the preparations were heterogeneous in length or in cap structure. In the present study, the basic mechanism of the endonuclease reaction is defined through the use of chemically synthesized, short capped RNAs.
Synthesis and 5Ј-Triphosphorylation of Oligoribonucleotides-Oligoribonucleotides were synthesized on an Applied Biosystems 380B DNA/RNA synthesizer (1-mol scale) using standard phosphoramidite chemistry. The method of Ludwig and Eckstein (13) for the solution phase synthesis of the nucleoside 5Ј-O-(1-thiotriphosphates) was modified for the conversion of the CPG-bound oligoribonucleotides into the corresponding oligoribonucleotide triphosphates. 2 After oligonucleotide synthesis the column was dried in a high vacuum oven for 15 min. A 0.2-mol aliquot was removed, deprotected, and purified as described below. The remaining CPG-bound RNA was transferred into a 2-ml vial and dried for 2 h at 35°C. The CPG was covered with pyridine (50 l) and dioxane (150 l) under argon. A 0.5 M solution of 2-chloro-4H-1,3,2benzodioxaphosphorin-4-one in dioxane (20 l, 10 mol) was added to the suspension followed by (after 15 min) 150 l (75 mol) of 0.5 M bis(tri-n-butylammonium) pyrophosphate in N,N-dimethylformamide and tri-n-butylamine (50 l). The solution was removed after 15 min, and 500 l (47 mol) of 1% iodine in tetrahydrofuran/pyridine/water (80:10:10, v/v/v) was added. After 20 min, the CPG was washed three times each with tetrahydrofuran (2 ml) and ethanol (2 ml). The oligoribonucleotides were deprotected and polyacrylamide gel-purified as described (16,17).
Purification of the Oligoribonucleotides-Crude products of the oligoribonucleotides and their corresponding 5Ј-triphosphates were purified on the same gel for comparative analysis, revealing, in general, a * 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.
‡ To whom correspondence should be addressed. Tel.: 215-652-5250; Fax: 215-652-2525; E-mail: david_olsen@merck.com.  (Table I), 100 M S-adenosyl-L-methionine, 0.35 mCi of [␣-32 P]GTP, 3.5 M GTP, and 120 units of RNasin. The reaction was incubated for 5 h at 37°C before being chloroform/phenol-extracted (18). Control reactions using the nontriphosphorylated RNAs were not capped. The capped 19-base RNAs were purified by polyacrylamide gel electrophoresis using 20% gels containing 8 M urea followed by Elutrap electroelution (Schleicher and Schuell) and ethanol precipitation. Alternatively, the capped RNAs were purified by reverse phase high performance liquid chromatography using a Spectra Physics 8800 instrument coupled with a SP8450 UV/VIS detector and a Vydac C18 column. The RNA was eluted with a gradient as described above. Low specific activity RNA was prepared using 50 M GTP and 10 Ci of [␣-33 P]GTP.
Enzyme Kinetics-Enzymic reactions were carried out at 25°C in the presence of 100 mM Tris-HCl, pH 7.8, 50 mM KCl, 5 mM dithiothreitol, 0.25 mM MgCl 2 (buffer A); 2, 1.6, or 1.2 g of ribonucleoprotein/ml (50, 40, and 30 pM endonuclease, respectively); and 400 pM 32 P-capped I. Reactions were initiated by the addition of I or MgCl 2 . Aliquots (4.3 l) were quenched after 0.5-90 min by mixing with 3 l of glycerol-tolerant gel buffer (U.S. Biochemical Corp.) stop mix (GTGB-SM; 30 mM Tris-HCl, pH 9, 30 mM Taurine and 0.5 mM EDTA, 90% formamide, 0.1% (w/v) bromphenol blue, 0.1% (w/v) xylene cyanol FF, and 90 g/ml tRNA). Alternatively, the reaction was stopped by filtering aliquots through a Multiscreen-NC nitrocellulose membrane and diluting with 4 l of GTGB-SM. Reaction products were separated from the substrate by running the samples on a 20% acrylamide denaturing sequencing gel and quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Biphasic time course data were fit to the equation, where p(t) is the amount of product formed at time t, a is the burst amplitude, k 1 is the first order rate constant, and k 2 is the steady-state turnover rate constant. The concentration of endonuclease active sites was determined from the burst amplitude, correcting for the difference between the burst and steady-state rates by multiplying the burst amplitude by (k 1 ϩ k 2 )/k 1 . IC 50 Determination for Capped 10-Base Oligoribonucleotide-Reactions were performed at 25°C in buffer A containing 3.3 g protein/ml, 0 -2.0 nM capped IV and were initiated after a 30-min preincubation by the addition of 10 l of 1.2 nM 32 P-capped I. Aliquots (4.5 l) of the reaction mixture were quenched after 0.5, 1.0, 1.5, and 2.0 min by mixing with 3 l of GTGB-SM. The amount of cleavage of the substrate was quantified using a PhosphorImager. Capped 32 P-labeled IV was unchanged in gel mobility by incubation with purified ribonucleoprotein. Inhibitor saturation data were curve-fit using a nonlinear leastsquares algorithm to the Hill equation (19).
Structure of the 3Ј-Uncapped RNA Cleavage Product-Capped I and 5Ј-AUUUUC-3Ј were 3Ј-end labeled using the enzyme poly(A) polymerase and [␣-32 P]cordycepin-TP according to the manufacturer's instructions. The other control oligoribonucleotide, 5Ј-AUUUUC(3Ј-dA)-3Ј, was 5Ј-radiolabeled using [␥-32 P]ATP and polynucleotide kinase. Capped I was incubated with the endonuclease to yield the capped labeled 13base and 6-base cordycepin-labeled products, which were analyzed on a 20% acrylamide sequencing gel and compared with the 5Ј-phosphorylated and nonphosphorylated 6-base cordycepin-labeled control RNAs.
Equilibrium Binding-Ribonucleoprotein (2 g of protein/ml) was incubated at 25°C for 30 min with 33 P-capped IV in 50 l of buffer A. Uncapped IV (50 nM) was included in all samples to displace nonspecific binding of the labeled capped IV. Enzyme-bound oligoribonucleotides were captured by nitrocellulose filtration in a 96-well manifold and were washed three times with 200 l of buffer. Enzyme-bound capped IV was quantified using a PhosphorImager. Fig. 1 demonstrates the specificity of the influenza endonuclease toward the cleavage of capped RNA substrate I. In the absence of added ribonucleoprotein, capped I was stable against hydrolysis. In the presence of the enzyme there was a time-dependent decrease in the amount of capped 19-base RNA with a concomitant increase in the amount of a specifically cleaved product. The position of hydrolysis was previously determined to be between nucleotides A 13 and A 14 (20). When GTP was supplied to the reaction, one or more GMP residues was added by the endonuclease-associated transcriptase activity to the cleavage product, as directed by the viral RNA template. This indicates that the product is specifically derived from influenza endonuclease and was not due to nonspecific hydrolysis of the RNA (Fig. 1). In order to determine the fate of the phosphate at the site of cleavage, doubly labeled (5Ј-capped and 3Ј-cordycepin) capped I was incubated in the presence of the endonuclease. As shown in Fig. 2, the 6-base cordycepin-labeled product from the cleavage reaction co-migrated with the control 5Ј-32 P-labeled 6-base cordycepin synthetic oligoribonucleotide, 5Ј-32 P-AUUUUC(3Ј-dA)-3Ј, and not with the nonphosphorylated RNA, 5Ј-HO-AUUUUC-32 P-(3Ј-dA)-3Ј, indicating cleavage of the 3Ј-O-P bond.

Substrate Requirement and Product Determination-
The ability of the endonuclease to cleave modified capped RNAs, where the 2Ј-OH at the preferred cleavage site was replaced by a hydrogen or a fluorine atom, was investigated. Oligoribonucleotides I and II, having either an unmodified adenosine or a 2Ј-deoxyadenosine located at A 13 , respectively, were incubated with ribonucleoprotein in the presence or absence of one of the four nucleoside triphosphates. The results are shown in Fig. 3. First, there was no detectable cleavage at A 13 for the 2Ј-dA containing RNA, and the site of cleavage was shifted from 3Ј to A 13 to the 3Ј side of A 14 . In addition, the hydrolysis rate was approximately 10 times slower than that for capped I, and similar results were obtained for the 2Ј-fluorocontaining oligoribonucleotide III (data not shown). The data presented in Fig. 3 also show that the specific cleavage products for both substrates (capped I and II) are extended when incubated in the presence of GTP and not with any of the other three nucleoside triphosphates. Endonuclease Reaction Kinetics-Typical biphasic reaction progress curves for the cleavage of I to the 13-base product are shown in Fig. 4A. Similar curves were obtained regardless of whether the reaction was initiated with capped I or with 0.25 mM Mg 2ϩ , the concentration of Mg 2ϩ that produces the maximal rate of reaction in this system (data not shown). Chung et al. (23) have shown that the endonuclease was able to bind but was unable to cleave capped RNA when the divalent metal ion was absent from the reaction. The amplitude of the burst phase was approximately proportional to the amount of endonuclease added. This observation suggests that the early phase of the reaction signifies a reaction that occurs prior to the establishment of steady-state turnover, which was likely due to a slow capped product off-rate. To confirm this interpretation, the same reaction was performed wherein the endonuclease-bound product (E⅐cap-13-mer) was removed from the substrate (cap-19-mer) and free cleaved product (cap-13-mer) by filtering the reaction mixture at various times through a nitrocellulose membrane. The results are shown in Fig. 4B (open symbols) and compared with those obtained from the same reaction but without the removal of enzyme-bound product (closed symbols). With the removal of the E⅐cap-13-mer, the burst phase of the reaction was absent and only the steady-state phase was observed. The rate was identical to the steady-state rate obtained for the unfiltered reaction. Using the burst amplitude to determine the endonuclease active site concentration we calculate that 0.5 mol of substrate was converted to product in the burst per mol of enzyme (concentration based on our equilibrium binding data below). In addition, burst and steady-state rates of 21 Ϯ 3 h Ϫ1 and 2.1 Ϯ 0.1 h Ϫ1 , respectively, were determined for the endonuclease reaction under these conditions. Tight Binding of Small Capped RNA IV-The product analog, capped IV, contained a 3Ј-deoxyadenosine and was therefore not a polymerization substrate. It was found to be a potent inhibitor of the endonuclease. An IC 50 value of 83 Ϯ 5 pM was determined, and the data are shown in Fig. 5A. In contrast, uncapped IV was not inhibitory when tested at concentrations up to 50 nM.
Equilibrium binding experiments were performed by separation of free from bound capped IV using a nitrocellulose filter retention assay. The equilibrium titration results are shown in Fig. 5B. The data were consistent with a single class of noninteracting binding sites. Nonlinear least-squares fitting returned values of K d ϭ 170 pM and a saturation amplitude of 52 pM. Only 10% of the bound 32 P-capped IV was competed by up to 150 nM uncapped IV. In contrast, 95% was displaced by competition with unlabeled capped IV. Thus, the binding to the endonuclease was dependent on the presence of a 5Ј-cap and reflected direct binding to the endonuclease active site. The effective endonuclease active site concentration was derived from the value of the saturation amplitude. FIG. 1. Sequencing gel (8% polyacrylamide) electrophoretic analysis of the products from an endonuclease cleavage reaction of 32 P-labeled RNA-capped I. Incubation was from 3 to 60 min in the absence and presence of 40 M GTP. The major product results from specific hydrolysis at position 13. When GTP was included, the cleavage product was extended by the endonuclease-associated transcriptase activity by one or more GMP residues. A small amount of nonspecific cleavage was obtained that was not extended in the presence of GTP.  6 and 12). The A 13 product (a) resulting from capped I and the A 14 product (b) from capped II were only extended by the endonuclease-associated transcriptase activity when GTP was present. The presence of the 2Ј-deoxyadenosine nucleoside in oligoribonucleotide II was confirmed by partial alkaline hydrolysis (compare lane 1, containing the hydrolysis products of capped I, and lane 7) (24). The absence of base-catalyzed hydrolysis bands corresponds to the position where the 2Ј-OH was replaced with a 2Ј-deoxysubstituted nucleoside (bracket c, lane 7). Two bands result from this treatment, apparently due to the product having a 2Ј,3Ј-cyclic phosphate or a 3Ј-phosphate terminus. The bands resulting from the alkaline hydrolysis migrate faster through the gel as compared with the same length endonucleasederived cleavage product due to the presence of the 3Ј-phosphate group and possibly due to C-8/N-9 cleavage within the cap base (1).

DISCUSSION
The aim of this work was to elucidate the mechanism of influenza endonuclease. To carry out this study, a method was developed for the synthesis of short, sequence-specific cap 1-containing RNA molecules, which allowed for the site-specific incorporation of nucleoside analogs.
The data presented for the cleavage of capped II indicate that the 2Ј-OH at the cleavage site is important for either recognition or catalysis (Fig. 3). A large number of ribonucleases catalyze the cleavage of the phosphodiester bond between the 3Ј and 5Ј riboses of adjacent nucleosides by using the ribose 2Ј-OH as the nucleophile. This results in the formation of a 2Ј,3Ј-cyclic phosphate-terminated RNA, which slowly hydrolyzes to the 3Ј-phosphate (21). In contrast, the final product of the influenza endonuclease reaction is an RNA with a free 3Ј-OH. This species is the required end product because it is utilized as a primer for the subsequent transcriptase reaction (10). In addition, the data presented in Fig. 2 indicate that the phosphate remains with the six-nucleotide 3Ј-cleavage fragment of I. Thus, the influenza endonuclease carries out the nucleophilic attack of the internucleotidic linkage with a group other than the 2Ј-OH, with the concomitant cleavage of the 3Ј-O-phosphorous bond. This is analogous to the mechanism proposed for the cleavage of RNA in DNA-RNA duplexes by Escherichia coli ribonuclease H (22). When the endonuclease was incubated with another analog of capped I, where the nucleoside at the site of cleavage was replaced with a 2Ј-fluoromodified ribose, no cleavage at this position was observed. In contrast, RNaseH is able to hydrolyze 2Ј-fluoro-substituted RNA, albeit at a reduced rate (22). This suggests that the 2Ј-OH is required for influenza endonuclease, whereas it is not likely to be utilized as the nucleophile at the site of cleavage.
The cleavage of capped 19-base I yielded a biphasic reaction progress curve. Fig. 4A illustrates that there was a burst in the conversion of substrate to product at a rate of 21 h Ϫ1 , followed by a much slower steady-state rate of 2 h Ϫ1 . The presence of a burst was consistent with the rapid accumulation of enzymebound product (E⅐cap-13-mer) followed by a slow breakdown of E⅐cap-13-mer to yield free enzyme. The less than stoichiometric conversion of substrate to product per enzyme determined from the burst amplitude was consistent with the presence of some ribonucleoprotein that was able to bind capped IV but was inactive catalytically.
Previous studies have shown that influenza endonuclease was able to bind but was unable to cleave capped RNA when divalent metal ion was absent from the reaction (23). The progress curves for reactions initiated with capped I or Mg 2ϩ were compared to determine whether productive binding of the capped RNA I was rate-determining for the burst part of the reaction. No significant difference in the reaction progress curves was observed, indicating that the rate of substrate binding does not significantly affect the rate of the burst phase. Thus, the rate-limiting step during the burst phase might be due to a catalytic event or a rate-determining conformational change.
The linear steady-state rate observed in the biphasic reaction progress curve (Fig. 4A) was consistent with the slow release of E⅐cap-13-base to yield free enzyme. In order to con- FIG. 4. Reaction progress curve for the cleavage of capped I by influenza endonuclease. Reactions were carried out as described under "Experimental Procedures." A, protein concentrations were 2.0 (q), 1.6 (Ⅺ), and 1.2 (å) g/ml. B, the reaction was quenched using the standard protocol before being loaded on the gel (q). Alternatively, aliquots were filtered through a nitrocellulose membrane to remove enzyme and RNA-associated enzyme before being loaded (Ⅺ). The data were fit to Equation 1.

FIG. 5.
A, IC 50 curve for m 7 GpppGmUAUUAAUA(3Ј-dA)-3Ј (capped IV) against influenza endonuclease activity using 400 pM m 7 Gppp-GmUUUUUAUUUUUAAUUUUC-3Ј I as substrate. The inhibitor was preincubated with the enzyme for 30 min prior to initiation of the reaction with capped I. The data were fit to the Hill equation and resulted in an IC 50 of 83 Ϯ 5 pM. B, equilibrium binding of capped IV to influenza endonuclease. Binding reactions were carried out as described under "Experimental Procedures." The data were fit to a hyperbolic function using a nonlinear squares algorithm to yield the following best fit parameters: amplitude ϭ 52.0 Ϯ 1.8 pM, K d ϭ 170 Ϯ 16 pM. firm this, a method was developed that removed all the E⅐cap-13-base at various time points during a reaction by filtering an aliquot through a nitrocellulose membrane. If release of the enzyme-bound product were rate-limiting, then the data would originate near the origin and linearly parallel the data obtained for the reaction with the burst. The results from this experiment (Fig. 4B) clearly show that the data from the two methods parallel each other. The slow phase of the reaction with the rate of 2 h Ϫ1 was consistent with the dissociation of the capped 13-base product from the enzyme. It is reasonable that the product dissociates slowly from the enzyme since it is the substrate for the ensuing transcriptase reaction. Also, it would be deleterious to the virus if the endonuclease efficiently catalyzed the cleavage of all the host cellular mRNA molecules, thus killing the cell.
The slow release of the capped RNA product is consistent with the tight binding and potent inhibition by the productanalog, capped IV (Fig. 5). In fact, a recent report suggested that a mixture of short length capped RNAs inhibited the RNA polymerase activity of influenza ribonucleoprotein (23). The capped 10-base IV that was selected for the current study was reported by Plotch and co-workers (10) to be the shortest influenza endonuclease-derived fragment from the cleavage of brome mosaic virus RNA 4. The tight binding of this productanalog also allowed for the determination of enzyme concentration and turnover number. It will be interesting to determine if shorter or linker-substituted capped RNA analogs also exhibit tight binding characteristics. Such compounds might prove useful as specific antiviral agents.
In summary, basic aspects of the influenza endonuclease reaction have now been elucidated with the use of chemically defined capped RNAs. In addition to establishing the bond hydrolyzed at the site of cleavage, endonuclease-catalyzed turnover was quantitatively determined.