Identification of the Substrate-binding Sites of 2 (cid:1) -5 (cid:1) -Oligoadenylate Synthetase*

2 (cid:1) -5 (cid:1) -Oligoadenylate synthetases are interferon-in-duced enzymes that upon activation by double-stranded RNA polymerize ATP to 2 (cid:1) -5 (cid:1) -linked oligoadenylates. In our continuing effort to understand the mechanism of catalysis by these enzymes, we used photo affinity cross-linking and peptide mapping to identify the sub-strate-binding sites of the P69 isozyme of human 2 (cid:1) -5 (cid:1) oligoadenylate synthetases. Radiolabeled azido 2 (cid:1) -5 (cid:1) -oli-goadenylate dimers were enzymatically synthesized and used as ligands for cross-linking to the P69 protein by exposure to ultraviolet light. The radiolabeled protein was digested with trypsin, and two ligand-cross-linked peptides were purified by immobilized aluminum affinity chromatography followed by reverse phase high pressure liquid chromatography. The peptides were identified by mass spectrometry and peptide sequencing and were found to correspond to residues 420–425 and 539–547 of P69. To examine the functional importance of the cross-linking sites, specific residues in the two peptides were mutated. When residues in the two sites were mutated individually, ligand cross-linking was selectively eliminated at the mutated site, and the enzyme activity was lost almost completely. Using substrates that can serve either as a donor or

Interferons are potent cytokines with anti-viral and cell growth modulating properties (1,2). These effects of interferons are mediated by the interferon-induced proteins (3). Among them are the family of enzymes called 2Ј-5Ј-oligoadenylate (2-5(A)) 1 synthetases (4,5). These enzymes are inactive, as such, and are activated by an essential co-factor, doublestranded (ds) RNA. The activated enzymes polymerize ATP to produce 2Ј-5Ј-linked oligoadenylates. The 2-5(A)s, in turn, activate a latent ribonuclease, RNase L. Activated RNase L can degrade cellular and viral RNAs and inhibit protein synthesis (6). Three sets of interferon-induced genes encode three size classes of these proteins: small, medium, and large. Members of all of the size classes have been cloned, and their enzymatic properties have been characterized (5,7). Within each size class, multiple members arise as a result of alternate splicing of the primary transcript. Recently it has been shown that one of the alternatively spliced isoforms of small synthetase can act as a pro-apoptotic protein of the Bcl-2 family (8). The enzymatic properties between the members of three size classes vary considerably in terms of the length of the 2-5(A) oligomers synthesized. The small isozymes synthesize only up to hexamers of 2-5(A) (9), whereas the medium isozyme, P69, can synthesize up to 30-mers of 2-5(A) (10). But the large isozyme, P100, makes mostly dimers of 2-5(A) (11). Because dimeric 2-5(A)s cannot activate RNase L, the biological role of P100 remains elusive. Another difference between the three size classes of isozymes is their oligomeric protein compositions. The small isozymes are functional only as tetrameric proteins, whereas the medium isozyme must dimerize for enzymatic activity, and the large isozyme functions as a monomer. We have previously identified specific residues in the carboxyl termini of the small and the medium synthetases, which are required for their oligomerization (12,13). These residues are absent from P100, which does not require oligomerization for activity.
Previously, we have used sequence comparison, molecular modeling, and site-directed mutagenesis to identify the three aspartic acid residues that constitute the catalytic center of these enzymes (13). In the current study, we have used photo affinity cross-linking of a 2-5(A) analogue followed by peptide mapping in conjunction with point mutagenesis to identify two substrate-binding sites of the P69 isozyme. These sites are conserved in other isozymes as well.
HPLC Purification of 2-5(A)-2-5(A) generated by a P69 reaction were separated on a HPLC using an anion exchange column (Rainin Pure-Gel SAX column; 7-m particle size; 500-Å pore size; 10 mm ϫ 10 cm) and a flow rate of 1 ml/min. HPLC Solvent A was 20 mM HEPES, pH 8.0; Solvent B was the same HEPES buffer containing 800 mM NaCl. The gradient used was: t ϭ 2 min, Solvent B ϭ 0%; t ϭ 2.2 min, Solvent B ϭ 10%; t ϭ 30 min, Solvent B ϭ 40%; and t ϭ 32 min, Solvent B ϭ 100%. 2-5(A) dimer peaks from five HPLC runs (see Fig. 1A) were collected. During the peak elution, the UV lamp was turned off to avoid photo activation of the azido group. Pooled peaks were dialyzed against 20 mM HEPES, pH 8.0, using Spectrapor CE Float-A-Lyzer (500 molecular weight cut off) overnight at 4°C. The dialyzed sample was concentrated in a Speedvac (Savant Instrument Inc.) followed by another round of dialysis for 4 h. The concentration of radiolabeled azido 2-5(A) was determined by measuring the A 260 , assuming the ⑀ of 2-5(A) dimer was same as that of ApppA.
Radiolabeled Azido 2-5(A) Dimer Cross-linking to P69 -The standard cross-linking protocol used 0.83 mM radiolabeled azido 2-5(A) dimer, 0.1 mg/ml P69, 20 mM Tris-Cl, pH 7.5, and 20 mM magnesium acetate, pH 7.5. The reaction mixtures were incubated on ice for 30 min and aliquoted in 25 l of volume in a Nunc 96-well mini tray. The tray was placed on ice, and the samples were photolyzed for 2 min with 254-nm radiation from a hand-held UV Mineralight lamp (UVP Inc., San Gabriel, CA) at a distance of 4 cm. Following the cross-linking, the samples were boiled for 1 min in SDS-PAGE sample loading buffer and electrophoresed. The protein bands were visualized by Coomassie Blue staining. For analytical experiments, the extent of cross-linking was determined by exposing dried gels to a PhosphorImager screen followed by scanning and quantification by Imagequant software. In the ApppA competition experiment, increasing concentrations of ApppA were present in the reaction mixture during the incubation prior to cross-linking.
Purification of Cross-linked Peptides-For preparative scale crosslinking, 80 -100 g of purified P69 was used in the standard crosslinking reaction. The samples were then electrophoresed in a preparative scale SDS-PAGE followed by Coomassie Blue staining. The radioactive cross-linked protein bands were excised, washed with water, followed by 0.1% trifluoroacetic acid, 60% acetonitrile, cut into small pieces (1 mm 3 ), and immersed in 100 mM NH 4 HCO 3 containing 10 mM dithiothreitol. Following reduction for 45 min at 55°C, dithiothreitol was removed, and the sulfhydryl groups were alkylated with 55 mM iodoacetamide (in 100 mM NH 4 HCO 3 ) for 30 min at room temperature. The alkylated samples were washed sequentially with 50% acetonitrile containing 50 mM NH 4 HCO 3 and then 10 mM NH 4 HCO 3 for 30 min each. Following the wash, the gel pieces were vacuum dried and rehydrated in 10 mM NH 4 HCO 3 containing 10 g/ml trypsin (Promega) and incubated overnight at 37°C. The peptides were extracted from gels with 0.1% trifluoroacetic acid, 60% acetonitrile (more than 60% of the initial radioactivity was recovered from the gel pieces). The peptide solution was then partially dried to remove acetonitrile and diluted with 1 M NaCl, 1% ammonium acetate, pH 5.8, for aluminum affinity chromatography.
Immobilized Aluminum Affinity Chromatography was done using metal chelating Sepharose (Amersham Biosciences) loaded with aluminum (14). 2.5 ml of metal chelating Sepharose was packed in a 1.5 ϫ 12-cm disposable column and washed several times with water. The column was charged with 30 ml of 50 mM AlCl 3. Before loading the peptides, the column was equilibrated with 30 ml of IAAC running buffer (1 M NaCl, 1% ammonium acetate, pH 5.8). Samples diluted in 5 ml of IAAC running buffer were loaded on the column, and the unbound peptides were washed with 30 ml of IAAC running buffer. A brief, 5-ml wash of 0.1% ammonium acetate was applied to reduce the NaCl concentration in the eluted peptides. The bound peptides were eluted from the column with 15 ml of 10 mM K 2 HPO 4 , pH 7.3. Throughout the chromatography, all of the eluents from the column were collected in 2-ml fractions and monitored for radioactivity by Cerenkov counting. The peptides eluted in fractions 20 -22 (see Fig. 2A) were pooled and concentrated in the Speedvac for further purification by reverse phase HPLC.
Reverse phase HPLC purification of cross-linked peptides was performed with an Applied Biosystems model 120A HPLC system using a Vydac C18 column (5-m particle size; 300-Å pore size; 1 mm ϫ 250 mm), aqueous trifluoroacetic acid/acetonitrile solvents, and a 50 l/min flow rate. The gradient was: t ϭ 2 min, Solvent B ϭ 5%; t ϭ 66 min, Solvent B ϭ 50%; and t ϭ 78 min, Solvent B ϭ 100%. The column eluent was split with 30% directed to the mass spectrometer and the reminder collected automatically in a fraction collector (1 min/fraction). The fractions were analyzed for radioactivity by measuring Cerenkov radiation. The regions of the total ion current corresponding to the radioactive fractions were analyzed for modified peptide mass. The radioactive fractions were also used for peptide sequencing by Edman degradation.
Liquid Chromatography Electrospray Mass Spectrometry-Liquid chromatography electrospray mass spectrometry was performed with a PerkinElmer Sciex API 3000 triple quadrupole mass spectrometer equipped with an ion spray source (15). Nitrogen was used as the nebulization gas (at 40 p.s.i.) and curtain gas and was supplied from a nitrogen Dewar. A scan range of 300 -2000 in the positive ion mode was used with 0.2-atomic mass unit steps, 0.4-ms dwell time/step, 40-V orifice potential, and 5000-V ion spray.
Peptide Sequencing-Peptide sequencing was done by the Cleveland Clinic Molecular Biotechnology core service using an Applied Biosystems model 492 Procise automated protein sequencer.
Production and Purification of P69 and Its Mutants-Site-directed mutagenesis by the mega primer polymerase chain reaction method was used for mutating Tyr 421 , Arg 544 , and Lys 547 of P69. Individual mutant proteins carrying an amino-terminal histidine tag were expressed in insect cells using a baculovirus vector and purified using nickel-nitrilotriacetic acid affinity chromatography as described before for the production and purification of wild type P69 (7).

RESULTS
Preparation of the Ligand-A photoactivable radiolabeled substrate of 2-5(A) synthetases is an ideal ligand for crosslinking. One inherent problem in selecting such a ligand is the fact that the affinity for ATP or other substrates for these enzymes is low (10,13). Consequently, a high concentration of the ligand is required to obtain a substantial amount of the ligand-bound protein. These considerations ruled out the possibility of using commercially available radiolabeled azido ATP as a ligand, because of prohibitive costs. Instead we decided to synthesize our own ligand, a 2-5(A) dimer, pppazidoA 2Ј p* 5Ј dA, using purified P69, 8-azido ATP, and ␣-32 P-labeled dATP. Because dATP does not have an acceptor 2Ј OH group, the desired dimer was the exclusive radiolabeled product. The conditions were developed using a high enzyme concentration and a short incubation time 2 to ensure that almost all input substrates had been converted to dimers containing both ppp-azidoA-p-azidoA and ppp-azidoA-p*-dA molecules. The dimers were purified by HPLC (Fig. 1A) and used as ligands in subsequent experiments.
Characterization of Cross-linking-In our experimental conditions, maximal photo-cross-linking was obtained after 5 min of exposure to UV light (data not shown). For structural analysis we used a subsaturating UV exposure of 2 min to avoid nonspecific cross-linking. The cross-linking of azido 2-5(A) dimer to P69 was saturable with increasing concentrations of the ligand. The dissociation constant (K D ) for the ligand as determined from this experiment was 0.88 mM (Fig. 1C), which agrees well with the binding constant of ApppA to P69 as determined earlier by fluorescence quenching assays (0.82 mM). 2 To establish the specificity of azido 2-5(A) dimer crosslinking to the P69, we used an unrelated protein, alcohol dehydrogenase, under the same cross-linking conditions. As shown in Fig. 1B, alcohol dehydrogenase did not cross-link to the ligand. The specificity of the photo-cross-linking of P69 with the ligand was also tested by competition experiments. ApppA is known to serve as an acceptor and can be elongated with one or two adenine moieties to produce ApApppA or ApAp-ppApA (13). We used increasing concentrations of ApppA to compete out the cross-linking of the radiolabeled ligand. Almost 75% of the cross-linking could be competed out with excess ApppA (Fig. 1D). Because dsRNA is a co-factor for all of the 2-5(A) synthetases, we tested the effect of dsRNA on the ligand cross-linking. Poly(I)⅐poly(C) did not affect the extent of cross-linking (data not shown).
Purification of Radiolabeled Peptides-Once we had established the specificity of cross-linking, we used the same methodology to generate ligand-cross-linked P69 for structural analysis. Approximately 80 -100 g of purified P69 was crosslinked and subjected to SDS-PAGE, radiolabeled bands were excised, and radioactivity was measured. Based on radioactivity measurements and estimating protein amounts by Coomassie Blue staining, the approximate stoichiometry of the cross-linking was 0.45 mol of ligand dimer/mol of P69. This is in good agreement with typical azido ATP cross-linking efficiency found with other polymerases (17). Following in-gel tryptic digestion of cross-linked P69, IAAC and reverse phase HPLC were used to purify the radiolabeled peptides. IAAC has been successfully used by others to partially purify ATP cross-linked peptides (14). In the present study, about 90% of the applied radioactivity was bound to the aluminum column, and about 42% was recovered in fractions 18 -26, specifically eluted with 10 mM phosphate ( Fig. 2A). Peptides in IAAC fractions 20 -22 were pooled, concentrated, and further purified on reverse phase HPLC. Two major radioactive peaks were observed by reverse phase HPLC and are marked as peaks A and B in Fig. 2B.
Identification of the Cross-linked Peptides-Ligand crosslinking to a peptide would produce a mass addition of 834. The mass spectra obtained from radioactive reverse phase HPLC peak A (Fig. 2B) included one major signal at m/z 773.8 (Fig.  3A). The doubly charged m/z 773.8 corresponded to the P69 peptide SYTSQK containing the ligand modification. A weaker singly charged ion from this peptide could also be seen at m/z 1547.4 (MHϩ calculated ϭ 1547.3). Another peak observed was the singly charged m/z 815, which corresponds to the unmodified P69 tryptic peptide FCLFTK (MHϩ calculated ϭ 815.4). The identity of the cross-linked peptide was confirmed by sequencing the peptide present in the fraction 46 of Fig. 2B by Edman degradation. The sequences obtained in the two experiments were SXTSQK and SXTSQKNER, respectively (Fig.  3D). The latter peptide is an extension of the former, produced by incomplete trypsin digestion. In both analyses, no amino acid was identifiable in the second cycle of degradation, suggesting that the ligand was cross-linked to this residue. Indeed, the eluate from this cycle of degradation contained most of the peptide-associated radioactivity, thus confirming that the Tyr residue in SYTSQK (Fig. 3D) was the site of cross-linking. In the peptide sequencing experiments, a minor contamination of the peptide, FCLFTK, was also observed. The high intensity of this contaminating peptide signal in the mass spectrum ( 3A) could be explained by the higher ionizibility of this unmodified peptide compared with the highly negatively charged modified peptide. We concluded that SYTSQK was the authentic cross-linked peptide and FCLFTK was a contaminant because of the following reasons. In the mass spectrum, only the SYTSQK peptide signal contained the ligand as well, and in the Edman sequencing experiment radioactivity was released in the second cycle in which, because of the modification, no Tyr could be detected, although Cys from FCLFTK was detected. The above conclusion was unequivocally validated by the mutational analysis described below.
The mass spectra from reverse phase HPLC peak B (retention time, 57-60 min) (Fig. 4B), contained several peptides but only one possibly modified peptide (Fig. 3B). A strong doubly charged m/z 797.2 and a weaker triply charged m/z 531.4 corresponded to the P69 peptide LKDLIR plus a mass addition of 834, contributed by the ligand. Upon repeating the complete photoaffinity labeling and purification procedure, very similar results were obtained. In the second set of labeling and peptide characterization results, the mass spectra of peak B indicated the modified peptide to be DLIRLVK (Fig. 3C). In this experiment, the level of contaminating peptides was much lower. The two peptides identified in Fig. 3 (B and C) are overlapping peptides generated by incomplete tryptic digestions at two sets of alternate sites (Fig. 3D). Their common region, DLIR, probably contained the site to which the adduct was cross-linked.

Properties of P69 with Mutations in the Substrate-binding
Sites-To confirm the substrate binding functions of the identified regions, we mutated several residues in these regions of P69 and studied the 2-5(A) cross-linking and enzyme activity of the mutant protein. Because the modification site in peak A was tentatively identified as Tyr 421 , this residue was mutated. For the peak B region, we were unable to specifically identify amino acid residues that cross-linked to the ligand. Two residues, Arg 544 and Lys 547 , were selected for mutation, because they were positively charged residues and could be the docking sites for the negatively charged ligand. A triple mutant P69, containing mutations at Tyr 421 , Arg 544 , and Lys 547 , was expressed in insect cells using the baculovirus vector, and the mutant protein was purified to homogenity. The triple mutant protein could not cross-link the ligand and was enzymatically inactive (Fig. 4, A, B, and D), although it could dimerize and bind dsRNA (data not shown). These results indicated that the identified substrate-binding sites are functionally important and that they are required for maintaining enzyme activity of the protein.
For examining the contributions of each of the two sites in ligand cross-linking and enzyme activity, two additional mutants were generated. The mutant Y421P had a mutation only in the peptide in peak A, and the mutant R544Y/K544A had mutations only in the peptide in peak B. Each mutant retained partial ability to cross-link the ligand (Fig. 4, A and B). Analysis of the cross-linked peptides generated from the mutants showed that peak A was lost in the Y421P mutant and peak B was reduced in the R544Y/K547A mutant (Fig. 4C). This result confirmed that the introduced mutations had selectively dis- rupted binding of the ligand to the mutated sites. Moreover, because the binding and the cross-linking of the ligand to the nonmutated sites of these mutants remained unaffected, it can be concluded that the two sites can bind substrates relatively independently and that they are not part of the same site. Binding to each site was, however, required for enzyme activity; the mutants had less than 2% enzyme activity as compared with the wild type protein (Fig. 4D). These results demonstrated that both of the ligand-cross-linked sites of P69 individually contribute to the enzyme activity of the protein. In the experiments shown in Fig. 5, we continued this mutational analysis by testing the properties of additional point mutants (Fig. 5A). As expected, all five mutants were partially defective in ligand cross-linking (Fig. 5B). Similar to the Y421P mutant, the Y421A mutant had a very low enzyme activity (Fig. 5C, bar  3). Point mutants R544A and R544Y were also equally inactive (Fig. 5C, bars 4 and 5), indicating that the Arg residue at position 544 is the crucial functional determinant of this domain. In contrast, point mutant K547A retained about 8% of enzyme activity (Fig. 5C, bar 6). Thus, for maintaining enzyme activity, Tyr 421 and Arg 544 were identified as the most important residues at the two substrate-binding sites.
Evidence Regarding the Nature of the Two Substrate-binding Sites-Between the two substrate-binding sites, one presumably serves as the acceptor-binding site and the other as the donor-binding site. We wanted to identify them; however, there is no absolute binding specificity for these sites, as reflected by the observed binding of the cross-linking ligand to both sites, although it could not serve as a donor. For this reason, we took advantage of the low residual activities of the single site mutants. We examined the kinetic properties of appropriate mutants, using as substrates, dATP, which can only donate dAMP residues, and A 5Ј ppp 5Ј A, which can only accept them. Keeping the concentration of one substrate high, we changed the concentration of the other substrate and measured the rates of the enzyme reactions. Examples of the results of such experiments are shown in Fig. 6. This series of investigations enabled us to determine the K m values for different substrates for each mutant. As reported previously, the Wt protein has a lower K m for ApppA and a higher K m for dATP as compared with that for ATP (13). For both Y421P and Y421A, the ATP K m was increased 4-fold; similarly the K m for ApppA was also increased, but that for dATP remained relatively unchanged. In contrast, for the R544A and R544Y pair, it was the K m for dATP that was 2.5-fold higher. For K547A mutant, as expected from its retention of considerable activity, the K m for dATP was only slightly elevated (Table I). The above results indicated that mutations of Tyr 421 affected ApppA binding, whereas those of Arg 544 affected dATP binding. We, therefore, tentatively identified the substrate-binding site at Tyr 421 as the acceptor site and that at Arg 544 as the donor site. DISCUSSION 2-5(A) synthetases are a unique family of enzymes because of their ability to catalyze 2Ј-5Ј phosphodiester bond formation. Previously we identified the catalytic center of these enzymes (13). In the present study, we have used photo affinity crosslinking followed by peptide mapping to identify putative substrate-binding sites in P69, one isozyme of this family. P69 is a nonprocessive nucleotidyl transferase, and it is expected to bind two substrate molecules simultaneously to join them by a 2Ј-5Ј phosphodiester bond. One site should bind the acceptor molecule whose 2Ј OH will be linked to the 5Ј PO 4 of the other molecule bound to the donor site. In addition to ATP and 2-5(A), many other nucleic acids containing a penultimate adenine can serve as the acceptor. Similarly, ATP and other nucleotide triphosphates can serve as donors. By virtue of their specific chemical structures, some of these molecules can only accept or only donate. For example, dATP can only be a donor in the reaction because of the lack of a 2Ј OH residue. Similarly, A 5Ј ppp 5Ј A is used as a specific acceptor because its 5Ј phosphate groups are blocked. Although such compounds can function selectively as donors or acceptors, their binding to the two sites is not as selective. Thus, no diagnostic compound is known that specifically binds to one site but not the other. However, previous studies by us and others (13,18,19) have shown the existence of the two substrate-binding sites whose affinities toward specific ligands are different.
We decided to use radiolabeled azido 2-5(A) dimers as ligands for cross-linking for several reasons. 2-5(A)s have a higher affinity for the enzyme than ATP; thus the cross-linking would be more specific, and the yield of cross-linked protein would be higher. The ATP K m of the enzyme is 2.1 mM (10). Thus, a very high concentration of radiolabeled azido ATP would be necessary to have significant cross-linking. The presence of the azido group in our ligand made it highly crosslinkable to the protein, and the presence of radioactivity made it easy to follow during purification. We used a high concentration of purified P69 to make radiolabeled dimers from 8-azido ATP and radiolabeled [␣-32 P]dATP. After several trials, we were able to achieve a reaction condition where all of the substrate was exhausted and the major reaction product was 2-5(A) dimers. The incorporation of dATP acted as a chain terminator. This dimeric 2-5(A) mixture also had azido 2-5(A) dimers in which two azido ATP dimerized. They were not separated from the radiolabeled ligand but could not be detected because of the absence of radioactivity. When the ligand was purified and used for cross-linking with P69, it showed specific labeling of the P69 protein. The labeling followed a hyperbolic saturation with an apparent dissociation constant (K D ) of 0.88 mM, compared with the 0.82 mM K D obtained from fluorescence quenching experiment. 2 The cross-linking could also be specifically competed by another substrate, ApppA, indicating that the sites of cross-linking were authentic. The co-factor dsRNA did not affect cross-linking, which was in agreement with our earlier observation that dsRNA does not affect substrate binding, as judged by fluorescence quenching of the protein, in response to substrate binding. 2 The total tryptic digest of cross-linked P69, when separated on a reverse phase HPLC column, always gave two distinct radioactive peaks (data not shown). But the complexity of the chromatogram made it impossible to identify the cross-linked peptides. When an extra IAAC purification step was introduced before HPLC, the basic two-peak characteristic remained. As mentioned in the results, we were able to identify and confirm the peptide present in the peak A region. But the peak B region of the chromatogram was always broad, indicating heterogeneity. Our efforts to identify the cross-linked peptides in this region by Edman sequencing were not successful because the yield of the cross-linked peptide in this region was low, and there were other unmodified peptides present in this region. However, the modified peptides identified in this region by mass spectrometry were consistent. To confirm our identification of the substrate-binding sites and to functionally test their roles in substrate binding, we mutated three residues in the two putative substrate-binding sites. The mutant showed no cross-linking and was enzymatically inactive. This provided evidence that these residues are involved in substrate binding to P69, which is essential for its enzyme activity. Individual mutations in each of the two sites eliminated cross-linking of the ligand to only the mutated site, thus demonstrating that binding of the ligand to each site is independent of binding to the other site. That binding to each site is functionally required for enzyme activity was also established by the observed loss of enzyme activity of these mutants. Neither of these sites corresponds to the ATP-binding site identified by Kon and Suhadolnik (18) for a different isozyme of 2-5(A) synthetase. The significance of the site identified by Kon and Suhadolnik is unclear, because mutations in that site did not destroy the enzyme activity (9,20).
The reaction sequence of 2-5(A) synthetases enzymes can be visualized as a two-step reaction. In the first step, the acceptor 2-5(A) or ATP, which will accept the incoming phosphate group at its 2Ј OH, and the donor ATP, which will be donating its 5Ј  a Wild type and mutant proteins were purified as His-tagged proteins. 0.01 mg/ml of each protein was used for assays.
b ApppA K m values were determined with constant dATP concentration (4 mM) and increasing ApppA concentrations. c dATP K m values were determined in the presence of constant ApppA (4 mM) and increasing dATP concentrations. a phosphate, bind to two respective sites. In the second step, the catalysis of the 2Ј-5Ј phosphodiester bond takes place between the donor and the acceptor, resulting in a one nucleotide longer 2-5(A) molecule. Thus, we can theoretically expect two substrate-binding sites in the enzyme defined as the acceptorbinding site and the donor-binding site. In the literature there has also been evidence for two substrate-binding sites for these enzymes with different affinities (13,18,19). The two sites of P69 identified in this study most probably do not constitute two parts of the same site, because mutations in one did not abolish ligand cross-linking to the other (Fig. 4C), and mutants in those two sites had distinctly different affinities for acceptors and donors ( Fig. 6 and Table I). Having identified the two substrate-binding sites in P69, we wanted to see which one of these two sites is the acceptor-binding site and which one is the donor-binding site. In the absence of any substrate that can bind to only the acceptor site or only the donor site of this enzyme, we used an indirect approach to identify them. Because of their chemical nature, ApppA can only be used by P69 as an acceptor in an enzyme reaction and dATP only as donor. Although ApppA and dATP can bind to both the acceptor and the donor sites, the successful reaction will take place only when they bind to the acceptor and the donor sites, respectively. Thus, when we monitor the formation of 2-5(A) between these two substrates, the K m for ApppA will indirectly reflect the binding efficiency of ApppA to the acceptor site, and similarly the dATP K m will represent its binding efficiency to the donor site. We measured the ApppA and dATP K m for several mutants to identify which mutations had affected the acceptor and the donor binding. The Tyr 421 mutants showed an increased ApppA K m compared with the Wt enzyme. However, the dATP K m for these mutants remained almost unchanged.
This indicated that Tyr 421 mutations affect the acceptor binding properties of P69. On the other hand, when we tested the Arg 544 mutants, the dATP K m was increased, but the ApppA K m was not substantially affected, signifying that this region is involved in the donor binding.
When the primary structures of the peptides present in peaks A and B were compared with the corresponding regions of other isozymes, a strong sequence conservation was observed (Fig. 7A). The sequences in the putative donor site were almost totally conserved including the critical Arg 544 and Lys 547 residues. Arg 544 was replaced by a Leu in the amino-terminal half of P69, providing further evidence that this half is functionally inert (13), although there is strong sequence homology between the two halves of the protein. The sequence conservation in the acceptor site was less pronounced (Fig. 7A). The Tyr residue, to which the ligand was cross-linked in P69, was replaced by another aromatic residue, Phe, in most of the other isozymes. This suggests the possibility of a base stacking interaction between the adenine moieties of the acceptor and the aromatic residue at the site. In contrast, the critical interaction at the other site could be ionic between the acidic phosphate residues of the donor and the basic amino acid residues Arg 544 and Lys 547 . Both of the above sites are located in the same region of P69 where the catalytic triad of three Asp residues is located (Fig. 7B). In the modeled structure of this region (13), Tyr 421 will be located at the end of the second ␤-sheet, very close to the catalytic center of the protein. The region in peak B, located further downstream (Fig. 7B), can form a long ␣-helix, similar to the nucleotide-binding helix region identified in poly(A) polymerase (21). In that enzyme, residues present in this region were shown to interact with the ␥-phosphate group of ATP (22). It is clear that the characteristics of the catalytic center and the two substrate-binding sites of P69 conform to the general structure of organizations of the nucleotidyl transferases, the large enzyme family to which the 2-5(A) synthetase belong (23).