Insertion of an N7-methylguanine mRNA Cap between Two Coplanar Aromatic Residues of a Cap-binding Protein Is Fast and Selective for a Positively Charged Cap*

The N7-methylated guanosine (m7G) cap structure, which is found at the 5′ ends of mature eukaryotic mRNAs, is critical to a myriad of biological processes. The twenty structures of complexes of cap nucleosides and nucleotides and methylated bases with the vaccinia virus VP39, a cap-specific RNA 2′-O-methyltransferase, which we have determined previously, have revealed the atomic basis of cap binding. The precise insertion and tight fitting of the m7Gua moiety of the cap between two parallel aromatic residues that are spaced only 6.8 Å apart governs the high specificity of binding. Here we report the investigation of the reaction mechanism of VP39 with three capped ligands (m7G, m7GpppG, and m7GpppGA3) by fluorescence stopped-flow technique. Cap binding is a simple one-step mechanism with very fast association rate constant (∼107 m–1 s–1). Moreover, the pH dependence on the association rate constant of m7G binding indicates that only the positively charged keto tautomer of the cap is recognized and bound. The association and dissociation rate constants and affinity constants of the three ligands do not vary greatly, demonstrating that binding is achieved almost entirely by the interactions of m7Gua with two aromatic residues in a cation-π sandwich.

The ability of proteins to discriminate alkylated from nonalkylated nucleic acids is of paramount importance in numerous biological processes, including DNA repair, pre-mRNA splicing, nucleocytoplasmic transport, mRNA translation, cap-dependent ribose methylation, and influenza virus transcriptional priming (briefly summarized in (1)). VP39, a vaccinia virus protein, provides an excellent and well defined system to study mRNA recognition at the molecular level. The protein acts in the processing of both ends of nascent mRNA transcripts. At the 3Ј end, the protein serves as a processivity factor for the vaccinia virus poly(A) polymerase (2,3), whereas at the 5Ј end, it acts, together with the S-adenosylmethionine coenzyme, as a cap 0 (m 7 G(5Ј)pppN⅐⅐⅐)-specific (nucleoside-2Ј-O-)-methyltransferase (3,4). To perform the enzymatic methylation of the 2Ј-hydroxyl group of the first transcribed base, VP39 must specifically recognize the m 7 Guanine (m 7 Gua) 1 moiety of the capped nucleotide (m 7 G), while also binding a segment of the mRNA transcript immediately downstream of the cap in a sequence-nonspecific manner. The failure of VP39 to catalyze the 2Ј-O-methylation of capped RNAs whose cap nucleotide contains an unmethylated guanine (4), makes the enzyme particularly suitable for investigating the specific recognition of an alkylated, positively charged base. Crystallographic studies have also revealed no binding of unmethylated nucleosides and nucleotides (5,6).
Our determination of over 20 x-ray structures of VP39 complexed with a variety of methylated nucleobases, nucleotides, and oligonucleotides have provided a comprehensive understanding at the atomic level of the process by which a protein interacts with a methylated, positively charged nucleobase of the cap and with an mRNA transcript in a sequence-nonspecific manner (1,(5)(6)(7)(8). The cap-binding site is located in a pocket at one end of the active site groove. The series of structures of ternary complexes of VP39, S-adenosylhomocysteine and m 7 G, m 7 Gp, m 7 Gpp, m 7 GpppG, or m 7 GpppG(A) 5 showed virtually identical binding mode of the m 7 Gua moiety of the cap. This binding mode has the following major atomic features: i) The m 7 Gua is sandwiched between two aromatic side chains of Tyr-22 and Phe-180 (Fig. 1A). These two side chains are oriented parallel to one another and spaced 6.8 Å apart in both the ligand-free and ligand-bound protein. The intercalated m 7 Gua moiety is in a perfect van der Waals contact distance (3.4 Å) from each flanking aromatic residue. ii) The methyl group at the N7 position makes a van der Waals contact or a weak CH⅐⅐⅐O hydrogen bond with a main chain peptide carbonyl oxygen (Fig. 1A). iii) The polar groups of m 7 Gua are engaged in five hydrogen bonds, four directly from the endocyclic N(1)H and exocyclic N(2)H 2 groups to the carboxylate side chains of Glu-233 and Asp-182 and one indirectly by way of a water molecule to the exocyclic carbonyl oxygen at position 6. iv) Binding of the ligands m 7 G, m 7 Gp, m 7 Gpp, and m 7 GpppG is mediated almost entirely by the interactions with the m 7 Gua nucleobase as the proximal groups (ribose, phosphates, and the second G) show little or no electron density and thus appear highly mobile.
Subsequent to the VP39 structure, the structure of another cap-binding protein, the messenger RNA 5Ј cap binding protein eI4FE was elucidated by both x-ray crystallography and NMR techniques (9,10). Although the features of cap recognition by eIF4E, based on its crystal structure with bound m 7 Gpp cap analog, are fundamentally analogous to those of VP39, some details of the recognition differ between the two proteins in the following manner (see also Ref. 1 for details). i) The m 7 Gua moiety is also sandwiched by two aromatic side chains, but they come from Trp residues. ii) The N7-methyl group is engaged in a nonpolar van der Waals contact with a carbon atom of a Trp residue. iii) The polar groups of m 7 Gua moiety are involved in only three hydrogen bonds. The C6 position carbonyl oxygen accepts a hydrogen bond from a backbone NH group. Only one carboxylate group from a Glu residue makes hydrogen bonds with the N(1)H and N(2)H 2 groups. Asp-182, one of the two acidic residues in VP39 that makes hydrogen bond with the m 7 Gua moiety, has no counterpart in eIF4E, and it appears to be essentially dispensable to cap binding and VP39 function (6,7).
The prevailing view emanating from a variety of studies, including extensive crystallographic and functional analysis of wild-type and mutant VP39, is that the enhanced double stacking or cation-(or more appropriately "cation-sandwich") interactions between the two aromatic residues and methylated nucleobases play a dominant role in cap recognition and binding (1,6,7). This mechanism requires a cap in the positively charged form. Although the two acidic residues in the cap-binding slot are apparently essentially dispensable to cap binding and VP39 function (5, 6, 11), they play primarily a role in molecular recognition by fixing the rotational orientation of the alkylated nucleobases about an axis perpendicular to that of the plane of the nucleotide ring.
Kinetic studies of ligand binding are necessary for a deeper understanding of the cap mRNA recognition process and reaction mechanism. In the present study, we investigate the kinetics of cap binding to VP39 by rapid mixing, fluorescence stopped-flow technique. This study has been facilitated by the replacement of Phe-180 by a tryptophan to provide a fluorophore ( Fig. 1B) (6,8). In addition we also address a key question concerning which tautomer (keto or enolate) of the cap is recognized and bound by investigating the effect of pH on the kinetics.

EXPERIMENTAL PROCEDURES
Materials-Site-directed F180W VP39 was purified as previously described (6). Nucleosides m 7 G was purchased from Sigma-Aldrich, m 7 GpppG from New England Biolabs. The compound m 7 GpppGAAA was kindly provided by the late Dr. Alec E. Hodel.
Equilibrium Ligand Binding to F180W VP39 as a Function of pH-Fluorescence titration measurements with F180W VP39 were conducted using an SLM/Aminco model 4800 spectrofluorimeter as described (6,8). With excitation wavelength set at 282 nm, ligand titrations as a function of pH were carried out at 20°C by following the decrease in fluorescence at 330 nm after adding microliter aliquots of concentrated ligand solution to a 2 ml solution of 2.5 M of protein, 0.2 M NaCl, and the appropriate buffer. To correct for possible photodecomposition, the fluorescence decrease was measured, in parallel, for an equivalent protein solution to which aliquots of buffer only were added. For each addition of ligand or buffer, an average of 10 individual measurements was recorded. K d values were determined by Scatchard analysis. The following buffers in 0.1 M concentration were used in the pH-dependent binding activity measurements: sodium citrate (pH 5.5), sodium cacodylate (pH 6.0 and 6.5), HEPES (pH 7 and 7.5), Tris-HCl (pH 8 and 8.5). Deviations assessed from duplicate or triplicate assays are Ͻ5% at pH 5-7 and about 10 -20% at pH 7.5-8.5.
Ligand-binding Kinetics-The rapid reaction between VP39 and ligand were followed by a Bio-Sequential DX-18MV stopped-flow apparatus (Applied Photophysics, Leatherhead, United Kingdom). The dead time of mixing was ϳ1.5 ms. The excitation wavelength was 282 nm, with 2-mm slits, and the emission wavelength range was selected by an interference filter with a band pass between 320 and 340 nm to eliminate the background fluorescence due to the base ligand which has fluorescence signal at 370 and 400 nm. Depending on the pH of the experiment, the protein concentration was either 1 or 2.5 M, and the ligand concentration ranged from 5-400 M. Experiments were performed at 5-6°C in solutions containing 0.2 M NaCl and 0.1 M buffer (sodium citrate (pH 5.5), sodium cacodylate (pH 6.0 and 6.5), HEPES (pH 7 and 7.5), and Tris-HCl (pH 8 and 8.5). For each stopped-flow measurement, 5-10 repeat shots were averaged. Exponential fittings were performed by the software package supplied with the equipment. Nonlinear regression analysis of pH-dependence of observed rates was conducted using the fitting routines resident in the SigmaPlot (SPSS Science, Chicago, IL.). Numerical integration was conducted using a simulation software package SCoP program (Simulation Resources, Inc., Redlands, CA).

RESULTS
One-step Binding Mechanism-Cap analog binding studies in solution have been greatly facilitated by replacing one of the two aromatic residues (Phe-180) that sandwich the m 7 Gua moiety of capped nucleotides and oligonucleotides with a Trp residue to provide a fluorescent reporter group ( Fig. 1) (6,8). The fluorescence emission maximum is at 330 nm with 55% quenching, with no shift in the emission maximum, occurring upon the addition of excess m 7 G. Ligand binding and functional activities of the mutant protein (named F180W VP39) have been characterized by a variety of techniques, including x-ray crystallographic, fluorescence measurements, isothermal calorimetry, and surface plasmon resonance. The F180W VP39 showed no apparent defects in methyltransferase catalytic activity or VP39-capped RNA interactions and no perturbation in the cap-binding pocket structure (6, 7). Moreover, the structures of the wild-type and F180W VP39s complexed with m 7 G showed no difference in ligand binding (6,7). The same is true for the binding of the m 7 GpppG dinucleotide (7,8). In all four different complex structures, only the electron density of the m 7 Gua moiety is well ordered, indicating that binding of the two ligands is driven almost entirely through the interaction with the m 7 Gua.
The kinetics of m 7 G cap analog binding to the F180W VP39 measured in a stopped-flow apparatus, was followed as a decrease in fluorescence intensity. As the binding was too fast to be measured at 25°C, the kinetic measurements were performed at 4Ϫ5°C. Typical kinetic data are depicted in Fig. 2A. At faster rates (i.e. higher m 7 G concentration), a significant proportion of the binding reaction occurred during the dead time (1.5 ms) of the apparatus. The observed fluorescence changes clearly follow a standard single exponential kinetics ( Fig. 2A). This kinetic behavior is consistent with a simple one-step reaction mechanism (Reaction 1), with only one event (the binding of the cap) being monitored by the fluorescence quenching kinetic measurement. REACTION 1 P and L represent the protein and ligand, respectively. The rate equation defining Reaction 1 is as follows in Equation 1.
The y-axis intercept of the linear plot of the observed rate, k obs , versus [L] or [m 7 G] (Fig. 2B) for the reaction at pH 5.5 gives the value of the dissociation rate constant k Ϫ1 (69 s Ϫ1 ), and the slope yields the value of association rate constant k 1 (3 ϫ pH Effects on the Kinetic Rate Constants and Equilibrium of Binding-To assess the effects of pH on the rates of reaction between VP39 F180W and m 7 G and thus the ionization state of the bound ligand and/or charged residues in the binding site, kinetic measurements were conducted as a function of pH. The fluorescent quenching kinetic data at pH Ͼ 5.5 also fit a single exponential function, and the observed rates also increased linearly with [m 7 G] (data not shown).
To check whether changing buffers at different pH affects the kinetic data by the changes in cation concentrations from the buffer, we performed the following stopped flow control experiment. m 7 G binding kinetics to 1 M VP39 was measured in 0.05 M or 0.1 M buffers at either pH 5.5 (citrate or cacodylate) or pH 7.0 (HEPES or Tris). The m 7 G concentrations after mixing were 10 and 15 M for pH 5.5 and 7.0, respectively. The four different buffer combinations at pH 7.0 did not make any significant difference in measured kinetics. The observed rates at 6°C were in the range of 180 -200 s Ϫ1 , and the observed fluorescence change was very similar for all four conditions. Thus changing buffers between Tris and HEPES and varying the buffer concentrations from 0.05 to 0.1 M produced no effect on the outcome of the kinetic measurements. Similar results were obtained at pH 5.5 using citrate and cacodylate buffers at concentrations of 0.05 and 0.1 M, but the observed fluorescence change in the cacodylate buffer was about 30% lower than that found in citrate buffer. The observed rates, 200Ϫ240 s Ϫ1 , however, did not show any significant changes in all four buffer combinations. These control experiments eliminate the possibility that the different ionic strength or ion species introduced by the buffer affect the observed rates. As long as the pH is maintained at the intended value the observed rate remains the same regardless of the buffer strength and composition. We have also shown previously that ionic strengths do not affect significantly the equilibrium binding of m 7 G (6).  It is evident from Table I and Fig. 3A that k 1 is dependent on pH from 5.5 to 8.5 in a manner that follows the standard Henderson-Hasselbach relationship with a pK value of 6.8. In contrast to the pH dependence on k 1 , the k Ϫ1 values are nearly constant (56 Ϯ 15 s Ϫ1 ) over the pH range investigated (Fig. 3A).
The apparent K a (ϭk 1 /k Ϫ1 ) values derived from the kinetic data at 5°C are not greatly different from those obtained by equilibrium fluorescence titration at 25°C (see "Experimental Procedures") (Fig. 3B). More importantly, the pH dependence of K a derived from both methods shows a similar pK of about 7 (Fig.  3, A and B).
Binding Kinetics of Longer Cap Analogs-We have extended our kinetic study to include capped analogs longer than m 7 G. Restricted by reagent availability, binding of only two analogs (the dinucleotide m 7 GpppG and the substrate m 7 GpppG(A) 3 ) was investigated at pH 6.0 (Table II). This study was further prompted by the different functional and structural features that each ligand exhibits as a potential substrate for the methylase activity of VP39. Although the dinucleotide has the methylation target (2Ј-OH) from the second G, it is not a substrate for VP39 (11). This finding is in full agreement with the crystallographic study showing that the second G is not visibly anchored in the catalytic site in the bound structures of both wild-type and F180W proteins (7,8). In fact, like m 7 G, only the entire m 7 Gua moiety of the dinucleotide showed a well resolved electron density in these structures. In sharp contrast, m 7 GpppG(A) 3 is an active methylation substrate (5). Moreover, the crystal structure of a ternary complex with the m 7 GpppG(A) 5 , a substrate that behaves like m 7 GpppG(A) 3 , and S-adenosyl homocysteine revealed a well defined electron density for the entire substrate and several polar interactions, including those with the ␤ and ␥ phosphates of the triphosphate linkage and the entire region of the first three transcribed bases (5). The anchoring provided by the interactions with the three transcribed bases is apparently a prerequisite for an active substrate.
Our kinetic analysis demonstrates that both the k 1 and k Ϫ1 rate constants decrease in the order m 7 G Ͼ m 7 GpppG Ͼ m 7 GpppG(A) 3 (Table II). Both rate constants decrease in such a manner that the apparent association constant (K a ϭ k 1 /k Ϫ1 ) values do not vary greatly. DISCUSSION The one step mechanism of ligand binding to VP 39 is consistent with many crystallographic data showing the absence of structural rearrangement following binding of m 7 G and several other cap analogs to both wild-type and F180W VP39s (1, 5-8). The uppermost association rate constant (k 1 of ϳ3 ϫ 10 7 M Ϫ1 s Ϫ1 ) observed at pH 5.5 and 6.0 is about 2 orders of magnitude slower than the diffusion rate limit, which appears fast for a binding process that depends almost entirely on the precise and tight insertion of the m 7 Gua nucleobase ring between two aromatic side chains (Trp-180 and Tyr-22) that are spaced only 6.8 Å apart (Fig. 1). The electron-rich clouds of both aromatic side chains, which possess a permanent electrostatic quadrupole (12), must therefore impart a strong attractive force for the electron-deficient m 7 Gua moiety. It is this force that in fact dominates cap binding by cap-specific proteins that rely on stacking interactions (5,6). Although hydrogen bonds contribute little to the strength of cap binding, they play an important role in dictating rotational orientation in the plane of the stack that apparently depends solely upon the arrangement of polar moieties of the nucleobase with respect to the hydrogen bondable side chains arrayed around the capbinding slot of VP39 (6).
The stopped flow reaction kinetics of the binding of m 7 GpppG to eIF4E, although it could not be ascribed strictly to either a one-step or two-step mechanism, indicates also a rapid association rate constant (13), about 10-fold faster than that to VP39 (Table II). The difference in rate constants may in reality be insignificant because the experiment with eIF4E was conducted at higher temperature (20°C). The slightly faster bind-  Table I). B, effect of pH on the apparent association constant K a (ϭk 1 /k Ϫ1 ) (filled circles) (data from Table I) and that determined by equilibrium binding (open circles) using fluorescence technique (see "Experimental Procedures"). C, Dependence of k obs on pH. To obtain experimental k obs values (filled circles) as a function of pH, 2 M protein was mixed with 30 M ligand following the conditions described under "Experimental Procedures." The solid line is the best fit by nonlinear regression according to Equation 3 that contains three floating parameters (k 1 , k Ϫ1 , and K of the ionization of the ligand according to Reaction 3 (see "Discussion").
ing could also be attributed to the involvement of two tryptophan residues of eIF4E in the stacking interactions (9). Tryptophan is believed to be the most efficacious in stacking interactions because its maximum intensity of negative electrostatic potential for indole side chain is greater than for Phe or Tyr side chains (Ref. 12, see also Ref. 8).
The most straightforward explanation for the effect of pH on m 7 G binding is that the protonated form at the N(1) position (also identified as the "keto tautomer" in reference to the neutral oxygen at position 6) is the bona fide ligand of VP39 and that the decrease in the k obs , association rate constant (k 1 ) and affinity in going from pH 5.5 to 8.5 (Fig. 3) reflects the ionization of the N(1)H group of the nucleobase to form the "enolate tautomer" (in reference to the deprotonated N(1) with the ensuing negative charge delocalized within the N1-C6-O6 region). The pK of about 7 for this ionization is close to that of about 7.5 for the unbound m 7 G ligand (6). Moreover, the pK of 7 is well removed from the normal pKs of carboxylic side chains (Asp-182 and Glu-233), the only ionizing groups in the cap-binding slot directly involved in the interactions with m 7 Gua, reaffirming that this represents the pK for the N(1)H group. We have also previously demonstrated that the two acidic residues are essentially dispensable to cap binding and VP39 function (1,5,6,11).
In contrast to the pH dependence on k 1 , the k Ϫ1 values are nearly constant (Fig. 3A). This result is to be expected if only one form of m 7 G, the keto tautomer, is bound. The kinetic measurements show no evidence for an additional process that could be ascribed to competing binding of the enolate form of m 7 G. If a competing reaction did exist, the k Ϫ1 values obtained at different pH would be expected to vary in a systematic manner, with the values determined at the extreme pHs reflecting the binding of the keto form at low pH and the enolate form at high pH. The data do not support this supposition. In fact, as further analyzed below, no binding of the enolate tautomer is expected.
In light of the effect of pH (from 5.5 to 8.5) on k 1 (Fig. 3A), which indicates binding of only the protonated m 7 G species, Reaction I can be modified as follows, where LH and L represent the N1 protonated (keto) and deprotonated (enolate) forms, respectively, of m 7 G, L o corresponds to the sum of both forms (Equation 2), and P refers to the protein. A nonlinear regression was conducted for the pH dependence of k obs according to Equation 3 to obtain the optimal values for the three floating parameters k 1 , k Ϫ1 , and K as defined in Reaction 3. The k 1 (3 ϫ 10 7 M Ϫ1 s Ϫ1 ) at pH 5.5 resulting from the nonlinear regression is very close to that determined experimentally (Table I). The k Ϫ1 of 52 s Ϫ1 also agrees very well with that of 56 s Ϫ1 , which is the average of the experimental values obtained at different pHs (Table I). Moreover, plots of k obs versus pH derived from the nonlinear regression and the experimental measurement are very similar (Fig.  3C). This finding strongly supports Reaction 2 or one step binding of only the protonated m 7 G to VP39. That these plots closely resemble the one shown in Fig. 3A for the effect of pH on k 1 indicates that the changes in k obs are correlated with changes in k 1 .
To fortify our conclusion on the binding mechanism represented by Reaction 2, numerical integration of Equations 2 and  a For the stopped flow measurements of the binding of the first three ligands, standard error for k 1 and k Ϫ1 values were obtained from linear regression to the observed (k obs ) rates measured at various ligand concentrations.
3 was performed to simulate the kinetic data obtained at different pH values. Initial total concentrations of VP39 and m 7 G are known (Table III), as are the amounts of protonated keto (LH) and deprotonated enolate (L) species at a given pH from the known K values (Reaction 3). Table III lists the k obs values obtained from the simulation at three pHs (5.5, 7.0, and 8.5) and different total ligand concentrations and, for comparison, those obtained experimentally. All k obs values from the simulations, which used a fixed set of k 1 and k Ϫ1 values, are very similar to those obtained experimentally. The amount of protonated m 7 G predicted by the simulation indicates that, even at pH as high as 8.5, there is more than sufficient amount of the active ligand to saturate all binding sites. This additional crosscheck by computer simulation provides further compelling evidence that the reaction between VP39 and m 7 G is represented by Reactions 2 and 3. It is also abundantly clear that the pH effect is simply a perturbation that modulates the concentration of the protonated keto or positively charged form of m 7 G. How does the deprotonation of the N(1)H of the m 7 Gua in the basic pH range and resulting formation of the enolate tautomer diminish ligand binding? We attribute this to the formation of a mesoionic or charge neutral m 7 Gua (14), bearing the positive charge that is locked into the five membered ring's N7-C8-N9 region and the negative charge that is delocalized within the six membered ring's N1-C6-O6 region. The absolute requirement of a positive charged keto form of the m 7 Gua for the double stack or cation-sandwich would preclude binding of the mesoionic form. In fact, because the N1-C6-O6 region overlaps with both stacking aromatic residues, which is more extensive with Tyr-22 than the replacement Trp-180 (Fig. 1B) (8), a negative charge within this region would totally prohibit cap binding.
Prior to the crystal structure determination of eIF4E with bound m 7 Gpp (9), extensive solution studies of the binding of a variety of N7 substituted cap analogs, including the effect pH, were conducted (15,16). These studies gave rise to a proposed model of cap recognition and binding whose major features require the enolate tautomer form of the cap and electrostatic interaction between the negatively charged O6 and a positively charged side chain. The increase in ligand affinity in going from pH 6 to the pH (7.6) of maximum binding activity was attributed to the formation of the enolate tautomer, whereas the decrease in affinity at pH Ͼ7.6 was ascribed to the deprotonation of the positively charged side chain, believed to be from a histidine residue. However, the cap-binding slot of eIF4E contains no histidine residue (9). A more serious flaw of the model is that a mesoionic cap, the form attributed to be bound to eIF4E at the pH of maximum binding activity, would, as discussed above, disallow the double stacking sandwich from ever forming. The atomic features of cap interaction with eIF4E as revealed by the x-ray crystallographic data (9) (also briefly described in the "Introduction") is fully consistent with preferential binding of the positively charged keto form, a mechanism that is completely compatible with that demonstrated for VP39.
It is noteworthy that both the association and dissociation rate constants (and affinity constants) of the three short ligands do not vary greatly (e.g. by orders of magnitude). This may be related to the findings that most of binding energy of short capped analogs and substrates is derived from the interactions of m 7 Gua, which is dominated by the cation-sandwich of the cap by two aromatic residues.
The finding that the association rate constant of m 7 GpppG is 8-fold slower than that of m 7 G, but close to that of the m 7 GpppG(A) 3 substrate indicates that the extended region of m 7 GpppG beyond m 7 G can put almost as much restriction or "steric effect" on ligand association as the longer substrate. Additional stable interactions (or "the connection energy" (17)) may be formed to slow down the dissociation of the longer ligands. The roughly 20-fold decrease in the dissociation rate constant of the m 7 GpppG(A) 3 substrate relative to the m 7 G is most likely attributed to additional stable interactions associated with the groups downstream of the m 7 Gua group as seen in the binding of m 7 GpppG(A) 5 substrate (7). The considerable interplay between the negative contribution of steric effect and positive contribution of the connection energy is dramatically demonstrated by the kinetics of binding of the 20 nucleotide capped poly(A) long RNA substrate as deduced by surface plasmon resonance technique, which indicates, relative to the dinucleotide binding, about two and four orders of magnitude slower association and dissociation rate constants, respectively (Table II). The approximately three orders of magnitude greater affinity (based on the kinetic data) of the long RNA substrate have been attributed to the involvement of four different binding sites (8,11), of which only one (the cap-binding slot) is mainly engaged in binding the dinucleotide or shorter ligands (7,8).
In conclusion, we have demonstrated using fluorescence stopped flow technique that the binding of cap analogs to VP39, which is dominated by the precise and tight insertion of the m 7 Gua moiety into two stacked aromatic side chains, is fast. Moreover, the decrease in the association rate constant and equilibrium binding constants in the basic pH range clearly indicates binding of only the keto tautomer or positively charged form of the m 7 Gua group of the capped ligands. These results further demonstrate for the first time that incorporation of an additional negatively charged center following deprotonation of the N(1)H group at high pH and concomitant formation of the enolate tautomer disallow binding.