Catalysis by RNase P RNA

Metal ions are essential cofactors for precursor tRNA (ptRNA) processing by bacterial RNase P. The ribose 2′-OH at nucleotide (nt) –1 of ptRNAs is known to contribute to positioning of catalytic Me2+. To investigate the catalytic process, we used ptRNAs with single 2′-deoxy (2′-H), 2′-amino (2′-N), or 2′-fluoro (2′-F) modifications at the cleavage site (nt –1). 2′ modifications had small (2.4–7.7-fold) effects on ptRNA binding to E. coli RNase P RNA in the ground state, decreasing substrate affinity in the order 2′-OH > 2′-F > 2′-N > 2′-H. Effects on the rate of the chemical step (about 10-fold for 2′-F, almost 150-fold for 2′-H and 2′-N) were much stronger, and, except for the 2′-N modification, resembled strikingly those observed in the Tetrahymena ribozyme-catalyzed reaction at corresponding position. Mn2+ rescued cleavage of the 2′-N but also the 2′-H-modified ptRNA, arguing against a direct metal ion coordination at this location. Miscleavage between nt –1 and –2 was observed for the 2′-N-ptRNA at low pH (further influenced by the base identities at nt –1 and +73), suggesting repulsion of a catalytic metal ion due to protonation of the amino group. Effects caused by the 2′-N modification at nt –1 of the substrate allowed us to substantiate a mechanistic difference in phosphodiester hydrolysis catalyzed by Escherichia coli RNase P RNA and the Tetrahymena ribozyme: a metal ion binds next to the 2′ substituent at nt –1 in the reaction catalyzed by RNase P RNA, but not at the corresponding location in the Tetrahymena ribozyme reaction.

Metal ions are essential cofactors for precursor tRNA (ptRNA) processing by bacterial RNase P. The ribose 2-OH at nucleotide (nt) ؊1 of ptRNAs is known to contribute to positioning of catalytic Me 2؉ . To investigate the catalytic process, we used ptRNAs with single 2deoxy (2-H), 2-amino (2-N), or 2-fluoro (2-F) modifications at the cleavage site (nt ؊1). 2 modifications had small (2.4 -7.7-fold) effects on ptRNA binding to E. coli RNase P RNA in the ground state, decreasing substrate affinity in the order 2-OH > 2-F > 2-N > 2-H. Effects on the rate of the chemical step (about 10-fold for 2-F, almost 150-fold for 2-H and 2-N) were much stronger, and, except for the 2-N modification, resembled strikingly those observed in the Tetrahymena ribozyme-catalyzed reaction at corresponding position. Mn 2؉ rescued cleavage of the 2-N but also the 2-H-modified ptRNA, arguing against a direct metal ion coordination at this location. Miscleavage between nt ؊1 and ؊2 was observed for the 2-N-ptRNA at low pH (further influenced by the base identities at nt ؊1 and ؉73), suggesting repulsion of a catalytic metal ion due to protonation of the amino group. Effects caused by the 2-N modification at nt ؊1 of the substrate allowed us to substantiate a mechanistic difference in phosphodiester hydrolysis catalyzed by Escherichia coli RNase P RNA and the Tetrahymena ribozyme: a metal ion binds next to the 2 substituent at nt ؊1 in the reaction catalyzed by RNase P RNA, but not at the corresponding location in the Tetrahymena ribozyme reaction.
Endonucleolytic 5Ј maturation of tRNA primary transcripts is catalyzed by the ribonucleoprotein enzyme ribonuclease P (RNase P) 1 in all three domains of life (Archaea, Bacteria, and Eukarya) as well as in mitochondria and chloroplasts (1)(2)(3). Bacterial RNase P enzymes are composed of a catalytic RNA subunit (4), ϳ400 nucleotides (nt) in length, and a single small protein of typically 120 amino acids (5). The only known exception may be the hyperthermophilic bacterium Aquifex aeolicus.
Neither genes for RNase P RNA (rnpB) or protein (rnpA) subunits could be identified in the sequenced genome, nor have biochemical studies provided evidence for the presence of a bacterial-like RNase P activity (6).
Processing of ptRNAs by RNase P results in cleavage products with 3Ј-OH and 5Ј-phosphate termini. Studies with the model RNase P RNA ribozymes from Escherichia coli and Bacillus subtilis have indicated that at least two metal ions (the actual number influenced by the nature and concentration of monovalent cations present) play a specific role in productive enzyme-substrate complex formation and cleavage chemistry (7)(8)(9)(10)(11)(12)(13)(14). A solvent hydroxide or activated water molecule is assumed to be positioned by a metal ion for nucleophilic attack in an S N 2 in-line displacement mechanism (7,15). A 2Ј-deoxyribose substitution at the RNase P cleavage site was previously reported to affect binding of catalytically important Mg 2ϩ and to substantially reduce the rate of catalysis by E. coli RNase P RNA (7, 16 -18). Recent NMR experiments have provided further evidence that the metal ion coordinated with the help of the 2Ј-OH group at nt Ϫ1 is actually "prebound" to ptRNA before complexation with RNase P RNA (19).
Here we have analyzed the role of the 2Ј-OH group at nt Ϫ1 of ptRNA substrates for cleavage by E. coli RNase P RNA (M1 RNA) in more detail by introducing a single 2Ј-amino (2Ј-N), 2Ј-fluoro (2Ј-F), or 2Ј-deoxy (2Ј-H) substitution at this position. We demonstrate that the 2Ј-OH at nt Ϫ1 of the substrate contributes to enzyme-substrate binding in the ground state and to transition state stabilization in particular. 10-fold slower cleavage chemistry of the 2Ј-F-ptRNA compared with the all-ribose control but a 13-fold rate advantage over the 2Ј-H-ptRNA suggests a hydrogen-bonding function for the 2Ј-OH at nt Ϫ1 during catalysis. All three 2Ј modifications at nt Ϫ1 caused a defect in the binding of catalytically relevant Mg 2ϩ . Increased miscleavage at position Ϫ2/Ϫ1 of 2Ј-N-modified substrate variants at lower pH suggests that the 2Ј-amino group becomes protonated and promotes electrostatic repulsion of the Me 2ϩ bound in its immediate vicinity. The extent of miscleavage and the pH value at which we observed 50% miscleavage showed a remarkable dependence on the base identities at nt Ϫ1 and ϩ73 of ptRNA. We further show that the replacement of Mg 2ϩ with Mn 2ϩ strongly improves processing efficiency for substrates with a 2Ј-N substitution at nt Ϫ1, although there is no convincing evidence for a direct metal ion coordination to the 2Ј substituent at nt Ϫ1.

EXPERIMENTAL PROCEDURES
RNA Synthesis and 5Ј 32 P-End-labeling-RNA oligonucleotides with and without single 2Ј-ribose modifications were either synthesized by phosphoramidite chemistry as described (20) or purchased from IBA (Göttingen, Germany). E. coli M1 RNA and the 3Ј-portion of ptRNA Gly missing the first 17 nt of the mature domain (termed tRNA 18 -79 ; see also Fig. 1) were synthesized by T7 transcription; the latter as part of a precursor transcript flanked by self-cleaving 5Ј-and 3Ј-cis-hammerheads to generate homogeneous 5Ј-and 3Ј-ends. 5Ј-end-labeling was performed as described (21). RNAs were purified by denaturing polyacrylamide electrophoresis.
Alkaline and Nuclease P1 Hydrolysis-Alkaline and nuclease P1 hydrolysis assays were performed essentially as described (22) Assembly of ptRNA Variants with Single Site Modifications-4.5 nmol of modified or unmodified RNA oligonucleotide (24 nt), covering the 7-nt 5Ј-flank and the first 17 nt of the mature domain of ptRNA Gly (Fig. 1), 3 nmol of 5Ј-phosphorylated tRNA  , 20 l of 1 M HEPES (pH 7.5), and 20 l of 0.1 M dithiothreitol were combined and adjusted with H 2 O to a final volume of 160 l. The mixture was heated for 3 min at 90°C, followed by incubation for 10 min at 65°C and slow cooling (45 min) to ambient temperature in a metal block removed from the heating apparatus. The solution was then adjusted to 15 mM MgCl 2 , 0.5 mM ATP, 10% (CH 3 ) 2 SO, and 140 units of T4 RNA ligase (Invitrogen) in a final volume of 200 l, and incubated overnight at 16°C. The sample and a 100-l wash of the reaction tube were combined, and RNA was concentrated by ethanol precipitation overnight at Ϫ20°C in the presence of 20 g of glycogen as carrier. RNA pellets were redissolved in 15 l of H 2 0. Three samples prepared in this way were combined (⌺45 l), mixed with loading buffer (45 l; Ref. 23) and loaded onto a 10% PAA/8 M urea gel (gel thickness: 1 mm; pocket width: 20 mm). Ligated pRNAs were detected by UV shadowing, excised from gels, and eluted from crushed gel slices in 200 mM Tris/HCl, 1 mM EDTA (pH 7.0) at 4 -10°C overnight, followed by a second elution for several hours. Eluted ptRNAs were concentrated by ethanol precipitation and redissolved in 50 l of H 2 O.
Kinetics-Standard processing assays were performed under single turnover conditions (5 M M1 RNA, Ͻ 1 nM ptRNA, 1 M NH 4 OAc, 50 mM MES, or HEPES; metal ion concentration and pH as indicated) as described (21). Aliquots withdrawn from enzyme-substrate mixtures were desalted by ethanol precipitation in the presence of 20 g of glycogen before analysis by 20% PAGE/8 M urea. Data analysis and calculation of single turnover rates of cleavage (k obs ) was performed as described (21).
The calculation of error bars in Fig. 3 is illustrated with the following representative example: the k obs values for all-ribose ptRNA Gly CU and ptRNA Gly C H U at pH 5.5 and 0.2 M Mg 2ϩ were 5 Ϯ 0.6 and 0.05 Ϯ 0.015 min Ϫ1 , respectively (each value based on five individual experiments); k rel (OH/H) was calculated as 5/0.05 ϭ 100; the upper and lower limits for k rel were calculated as 5.6/0.035 ϭ 160 and 4.4/0.065 ϭ 68, respectively; thus, k rel is illustrated as 100 with errors of ϩ60 and Ϫ32.

RESULTS
ptRNA Ground State Binding to M1 RNA-Variants of the standard substrate ptRNA Gly CU (C-1, Uϩ73) with a 2Ј-OH, 2Ј-F, 2Ј-N or 2Ј-H substituent at position Ϫ1 (referred to as ptRNA Gly CU, C F U, C N U, and C H U in the following; Fig. 1) were analyzed for binding to M1 RNA in the ground state. The 2Ј modifications had moderate (2.4 -7.7-fold) effects on ptRNA binding, decreasing substrate affinity in the order 2Ј-OH Ͼ 2Ј-F Ͼ 2Ј-N Ͼ 2Ј-H (Table I). This was measured at pH 6.0 in the presence of 15 mM Ca 2ϩ as the divalent metal ion to prevent substantial ptRNA cleavage during binding assays.
Kinetics-The same substrate variants mentioned above were processed under single turnover conditions (see "Experimental Procedures") at saturating enzyme and varying divalent metal ion concentrations, either at pH 7.0 or pH 5.5. The pH of 7.0 was chosen to provide conditions where the 2Ј-amino group is mainly deprotonated. The pK a of the 2Јamino group was determined to be 6.2 (25) and 6.0 (26) in the context of dinucleotides. The pH 5.5 allowed us to analyze cleavage at rate-limiting chemistry for all substrates, including the most efficiently cleaved all-ribose substrate. Ratelimiting chemistry for the latter was inferred from the linear relationship of log k obs over pH in the range of pH 5.0 -6.5 (8). Cleavage efficiency of all-ribose and 2Ј-modified ptRNA Gly CU variants at pH 7.0 decreased in the order 2Ј-OH Ͼ 2Ј-F Ͼ 2Ј-N Ͼ 2Ј-H ( Fig. 2A). The same hierarchy was seen when Mg 2ϩ was replaced with Mn 2ϩ (Fig. 2B). A similar relationship was evident at pH 5.5 (Fig. 2, C and D), although the ptRNA Gly C N U was as inefficiently cleaved as the ptRNA Gly C H U, suggesting that protonation of the 2Ј-amino group at pH 5.5 impaired cleavage efficiency (see below). In Fig. 3, we plotted the data of Fig. 2, A and C as relative cleavage rates (k rel ) versus Mg 2ϩ concentration. The k rel values for processing at pH 7.0 ( Fig. 3A) decreased with increasing [Mg 2ϩ ], but remained rather constant between 0.04 and 0.1 M Mg 2ϩ . At 0.06 M Mg 2ϩ , the cleavage rate of the unmodified substrate was ϳ3-fold, 16-fold, and almost 40-fold higher than that of ptRNA Gly C F U, C N U and C H U, respectively. At pH 5.5 (Fig. 3, B and C), where the chemical step is rate-limiting, a linear relationship of k rel versus [Mg 2ϩ ] is clearly evident at [Mg 2ϩ ] Ն 0.1 M, with the rate for the unmodified ptRNA exceeding that of the modified substrates by a factor of ϳ10 (2Ј-F) and about 150 (2Ј-H, 2Ј-N). We are aware that the unphysiologically high concentrations of Mg 2ϩ used here (up to 0.6 M) may have affected the conformational equilibria of substrate and ribozyme RNAs. However, the steady increase of k obs with increasing [Mg 2ϩ ] up to 0.6 M (Fig. 2, C and D) did not suggest any obvious defects of processing at such high Mg 2ϩ concentrations.
We interpret our results to indicate that (a) the 2Ј-OH at the RNase P cleavage site is involved in binding of catalytically important Mg 2ϩ as previously suggested (7), and (b) that cleavage chemistry of the modified substrates is inherently slower. Evidence for (a) is the finding that the reductions in rate are most pronounced at lower Mg 2ϩ concentrations. This is seen for the experiments performed at pH 7.0 ( Fig. 3A), because at this pH value cleavage efficiencies were high enough to reliably measure rates for the modified ptRNA variants at Mg 2ϩ concentrations in the lower millimolar range.
The relative recovery of catalytic performance (k rel ) for the modified substrate versions at high [Mg 2ϩ ] is more pronounced at pH 7.0 than pH 5.5 (for example, compare the k rel values at 0.1 M Mg 2ϩ in Fig. 3A versus B and C). We think that two factors contributed to this difference. First, the cleavage rates for the most efficient all-ribose substrate were Ͼ 10 min Ϫ1 at pH 7.0 and Mg 2ϩ concentrations exceeding 0.04 M and thus at the limit of precise rate determinations by manual kinetics. Therefore, we have to consider some underestimation of k rel values at higher Mg 2ϩ concentrations in Fig. 3A. Second, the linear relationship between pH and log k obs for single turnover cleavage of all-ribose ptRNA by M1 RNA starts to level off above ϳpH 6.5. Apparently, another step, possibly a conformational change (28), begins to limit the cleavage rate at pH 7.0 for the all-ribose but not for the more slowly cleaved modified substrates. Thus, at pH 7.0 and high Mg 2ϩ concentrations the catalytic defect caused by the 2Ј modifications at nt Ϫ1 seems to be partially masked, whereas results obtained at pH 5.5 provide a better assessment of the catalytic defect, except for ptRNA Gly C N U whose 2Ј-amino group was inferred to be substantially protonated at pH 5.5 (see below).
Cleavage Site Selection-Next, we analyzed if ptRNA Gly C F U, C N U and C H U were cleaved at the canonical site (Ϫ1/ϩ1) by M1 RNA or if the modification directed cleavage to the next unmodified phosphodiester in the 5Ј-direction (position Ϫ2/Ϫ1).  14) were cleaved at Ϫ1/ϩ1 at pH 7.5. However, the 2Ј-N variant showed substantial cleavage at position Ϫ2/Ϫ1 at pH 5.5 (Fig. 4B, lane 12). No substantial cleavage at Ϫ2/Ϫ1 was detected at pH 5.5 for ptRNA Gly CU, C F U, or C H U (data not shown). Remarkably, the cleavage site in ptRNA Gly C H U was shifted to position Ϫ2/Ϫ1 when we used the C292 mutant M1 RNA (Fig. 4D, lane 17), which essentially abolishes base pairing with the tRNA CCA terminus (21,29). In contrast, the all-ribose substrate was still cleaved at position ϩ1/Ϫ1 by the C292 mutant RNA (Fig. 4D, lane 18). This documents the synergistic contributions of the 2Ј substituent at nt Ϫ1 and the tRNA 3Ј-CCA base-pairing interaction to transition state complex formation at the canonical cleavage site, particularly in the context of our substrate that is assumed to form only a weak Uϩ73/U294 interaction with E. coli M1 RNA (30,31).  concentrations. C, Mg 2ϩ -dependent cleavage rates at pH 5.5. D as C but omitting the curve for all-ribose ptRNA Gly CU to magnify the differences among the modified substrate variants. Fig. 2A (A) and Fig. 2C (B and C). k rel is the quotient of k obs for all-ribose ptRNA Gly CU divided by k obs for the respective modified substrate at the corresponding Mg 2ϩ concentration: k obs OH /k obs F (q), k obs OH /k obs N ( or OE), k obs OH /k obs H (Ⅺ). For determination of error bars, see "Experimental Procedures." The data points for k rel OH/N (OE, panel C) are shown separately to avoid commingling of error bars with those for k rel OH/H (Ⅺ) in panel B.

FIG. 3. Secondary plots based on the data shown in
pH-dependent Miscleavage of 2Ј-Amino-modified ptRNA Variants-The pH dependence for processing of ptRNA Gly C N U by M1 RNA revealed substantial miscleavage at pH values Ͻ 6.0 (Fig. 5A). The log-linear relationship of the overall cleavage rate and pH with a slope of 1.1 (Fig. 5B) indicates that cleavage chemistry is rate limiting up to pH 8.0 for the ptRNA Gly C N U substrate. As mentioned before, a very similar relationship with the same slope was reported for processing of a ptRNA Phe with a single 2Ј-H modification at nt Ϫ1 by Bacillus subtilis RNase P RNA in the pH range between 6.1 and 7.8 (27). The results in Fig. 5A suggests that protonation of the 2Ј-amino substituent at low pH causes displacement of catalytic Me 2ϩ due to electrostatic repulsion resulting in utilization of the next unmodified phosphodiester in the 5Ј-direction. We then analyzed if pH-dependent miscleavage of ptRNA Gly C N U also pertains to a variant (ptRNA Gly C N A) with a U to Aϩ73 exchange. This might stabilize the interaction of nt ϩ73 with U294 of E. coli M1 RNA and thus favor cleavage at Ϫ1/ϩ1 (30). Fig. 6 shows the similar pH dependence of miscleavage at Ϫ2/Ϫ1 for ptRNA Gly C N U and ptRNA Gly C N A. Surprisingly, the pH at which 50% miscleavage occurred (termed pH 50% ), increased by 0.3 pH units from pH 5.3 to 5.6 for ptRNA Gly C N A. Even more surprising, variants with a C to U exchange at nt Ϫ1, ptRNA Gly U N A and U N U, showed substantial reductions in miscleavage (Table II). At pH 5.0 and with Mg 2ϩ as the divalent metal ion, these variants were only miscleaved to 53% (U N A) and 14% (U N U), whereas miscleavage amounted to 77% (C N U) and 92% (C N A) for the two other variants. A similar hierarchy of miscleavage was observed in the presence of Mn 2ϩ at pH 5.0 (Table II).
Direct Metal Ion Coordination to the 2Ј Substituent at nt Ϫ1?-We considered the possibility that the 2Ј substituent at nt Ϫ1 may be involved in direct metal ion coordination, although inner-sphere coordination of 2Ј-OH ligands to Mg 2ϩ seems to be rare (32).
Unlike Mg 2ϩ , transition metal ions, such as Mn 2ϩ and Zn 2ϩ , can form stable inner-sphere complexes with nitrogen-containing compounds, for example with the N7 of purine nucleotides (33). A metal ion interaction with the 2Ј substituent of the G nucleophile was inferred for the group I self-splicing introns from Mn 2ϩ and Zn 2ϩ rescue of a reaction using a 2Ј-amino-2Јdeoxyguanosine analog instead of guanosine as the nucleophile (34). To address the question of a potential direct metal interaction with the 2Ј substituent at nt Ϫ1, we determined the cleavage rate for the 2Ј-N-modified substrate variants in the presence of Mn 2ϩ versus Mg 2ϩ . The results are summarized in Table III 1 and 2) is identical to the first 24 nt of ptRNA Gly C F U (Fig. 1). 24-mer and ptRNA-Gly C F U were 32 P-labeled at their 5Ј-ends. Lane 1, partial nuclease P1 digest of the 24-mer, with the lengths (in nt) of some hydrolysis products marked at the left margin; lanes 2 and 3, partial alkaline hydrolysis (OH Ϫ ) of the 24-mer (lane 2) or ptRNA Gly C F U (lane 3), with the lengths (in nt) of some hydrolysis products marked at the right margin; the asterisk marks the missing fragment in the alkaline hydrolysis ladders due to the 2Ј-fluoro modification at nt Ϫ1. Note that nuclease P1, as RNase P, generates 5Ј-phosphate and 3Ј-OH ends, whereas alkaline hydrolysis produces 5Ј-OH and 2Ј,3Ј-cyclic phosphate termini.  7 and 8); note that nuclease P1 failed to hydrolyze at the site of the 2Ј-amino modification (marked by the asterisk at the left margin). Lanes 9 and 10, as lanes 4 and 5 in panel A; lanes 11 and 12: as lanes 9 and 10, except that the processing reaction was performed at pH 5.5. On the right margin, only the two 7and 6-nt long 5Ј-cleavage product generated by M1 RNA processing are indicated, corresponding to cleavage between Ϫ1/ϩ1 and Ϫ2/-1, respectively. C, cleavage site analysis for processing of ptRNA Gly C H U by M1 RNA . Lanes 13 and 15, 5Ј-end-labeled RNA oligonucleotides 5Ј-pC-CCUUU (lane 13) and 5Ј-pCCCUUUdC (lane 15, the latter carrying a single 2Ј-deoxy modification at the 3Ј-terminal C residue) used as length standards, with sizes (in nt) indicated at the left margin; lane 14, 5Ј-cleavage product obtained after processing of ptRNA Gly C H U by M1 RNA at pH 7.0; lane 16, partial alkaline hydrolysis (OH Ϫ ) of ptRNA Gly C H U, with the sizes of some hydrolysis products marked at the right margin, and the asterisk indicating the missing fragment due to the 2Ј-deoxy substitution. D, miscleavage (between nt Ϫ2/Ϫ1) of ptRNA Gly C H U by the G292 to C mutant M1 RNA, but not by wild-type M1 RNA. Lanes 17 and 18, 5Ј-cleavage product obtained after processing of ptRNA Gly C H U (lane 17) or all-ribose ptRNA Gly CU (lane 18) by the C292 mutant M1 RNA at pH 6.0; lanes 19 and 20, as lanes 15 and 13, respectively, in panel C; lanes 21 and 22, as lanes 17 and 18, but using wild-type M1 RNA; note that 5Ј-pCCCUUUdC and 5Ј-pCCCUUUC (both 7-nt long) migrated differently in the denaturing PAA gels (marked by the double arrow at the right margin). Electrophoresis was performed in 30% PAA/8 M urea gels in panels A, B, and C. In panel D, a 25% PAA/8 M urea gel was used and the electrophoresis buffer was 0.5ϫ TBE instead of 1ϫ TBE. completely inverted for the substrates with a 2Ј-N substitution at Ϫ1, with ptRNA Gly C N U being processed at an 82-fold higher single turnover rate in the presence of Mn 2ϩ versus Mg 2ϩ . This was analyzed at pH 6.5, where cleavage occurred predominantly at Ϫ1/ϩ1. The Mn 2ϩ rescue effect was again remarkably influenced by the base identities at Ϫ1 and ϩ73, decreasing in the order C N U Ͼ C N A Ͼ U N U Ͼ U N A (Table III). An important reason for these differences is the particularly low cleavage efficiency of the 2Ј-N-modified variants with a C at Ϫ1 in the presence of Mg 2ϩ . Could the observed Mn 2ϩ rescue effects be interpreted as indicating a direct metal ion coordination to the 2Ј-amino substituent at nt Ϫ1? We argue that this is not the case because a Mn 2ϩ rescue of similar strength (k rel ϭ 52, Table  III) was also observed for the ptRNA Gly C H U carrying a 2Јdeoxy modification at nt Ϫ1, which abrogates a direct metal ion coordination. The absence of a direct metal ion coordination to the 2Ј-amino substituent at nt Ϫ1 is also supported by the inability to stimulate processing of ptRNA Gly C N U by addition of Zn 2ϩ (data not shown).

2Ј-Deoxy and 2Ј-Fluorine
Modifications at nt Ϫ1-In a previous study by Drew Smith and Norman R. Pace (7), a 2Ј-deoxy modification at nt Ϫ1 was calculated to reduce the rate of catalysis by a factor of 3400, which largely exceeds the factor seen in the present study (ϳ150-fold, Fig. 3B)   The pH at which 50% miscleavage occurred (pH 50% ) was shifted upwards by 0.3 pH units from pH 5.3 to 5.6 for ptRNA Gly C N A relative to ptRNA Gly C N U. Note that variants ptRNA Gly U N A and ptRNA Gly U N U showed only 53% and 13.7% miscleavage, respectively, at pH 5.0 (see Table II). possible explanation for this discrepancy is that, in the previous study (7), the cleavage rate of an all-ribose ptRNA with a 5Ј-flank of more than 10 nt and a U residue at nt Ϫ1 was compared with the rate for a ptRNA with a single 2Ј-deoxy A residue as 5Ј flank. A ptRNA with a 1-nt 5Ј-flank was estimated to be cleaved up to 30-fold slower than ptRNA variants with 5Ј-precursor segments of Ն2 nt in the reaction catalyzed by B. subtilis RNase P; in addition, changing the nucleotide at position Ϫ1 from U to G reduced the cleavage rate 70-fold (36). Likewise, B. subtilis RNase P RNA cleaved a ptRNA with a single 2Ј-deoxy C as the 5Ј-flank at a ϳ30-fold slower rate than a variant thereof with four additional ribonucleotides at the 5Ј-end (35). Exchanging the nt upstream of the cleavage site from U to A within an 8 -10 nt long 5Ј-flank reduced k cat in the M1 RNA-catalyzed reaction by a factor of 12-to almost 200fold, depending on the substrate context (13). Even if one assumes that only a minor proportion of these reported reductions in catalytic rate took effect in the previous study (7), such effects on reactivity may easily explain the quantitative differences a 2Ј-deoxy modification at nt Ϫ1 had on the cleavage rate constant in the study by Smith and Pace (3400-fold) relative to the present study (ϳ150-fold, Fig. 3B). It is instructive to compare the results obtained for M1 RNA with those reported for the second large ribozyme model system, the Tetrahymena group I ribozyme. Both RNA enzymes generate cleavage products with 3Ј-OH termini, in contrast to the known small ri-bozymes (hammerhead, hairpin, VS, and HDV ribozyme), which produce 2Ј,3Ј-cyclic phosphates (Ref. 37, and references therein). A difference between the Tetrahymena ribozyme and RNase P RNA, however, is that the former utilizes an exogenous guanosine nucleophile instead of a solvent hydroxide. A 2Ј-F modification at U-1 of the substrate for the Tetrahymena ribozyme caused a 2.5-fold increase in the K d for substrate ground state binding and slowed the chemical step by a factor of 8.5 compared with the all-ribose substrate (38). These values are remarkably similar to the effect of a 2Ј-F modification observed in our study (2.4-fold K d effect, 10-fold slower chemistry; Table I, Fig. 3B). In the Tetrahymena ribozyme reaction, the cleavage rate was 70-fold faster with a 2Ј-F versus a 2Ј-H modification at nt Ϫ1 under rate-limiting chemistry. For M1 RNA, this rate advantage of the 2Ј-F substitution was 13-fold (Fig. 3B), and the 2Ј-H modification had a slightly more pronounced effect on the K d of substrate ground state binding in the M1 RNA-catalyzed reaction (7.7-fold versus 3-fold increase; Table I, Ref. 38). Nevertheless, the similarities are striking: (a) in both systems, the 2Ј modifications at nt Ϫ1 had minor to moderate effects on substrate ground state binding, whereas the major effects were on the chemical step; and (b) a 2Ј-F modification conferred a substantial rate enhancement over the 2Ј-deoxy substituent, but a 2Ј-OH function was superior in both systems. Based on this analogy there is no reason not to argue along the same lines as for the Tetrahymena ribozyme Bases Ϫ1, ϩ73, and ϩ75 of the ptRNA substrate as well as G292 of M1 RNA are depicted in green; green arrows illustrate that base identities at Ϫ1 and ϩ73 as well as disruption of the base-pairing between Cϩ75 and G292 had profound effects on the kinetic phenotype, metal ion utilization, and cleavage site selection (Fig. 4, Tables II and III) of substrates with single 2Ј modifications at nt Ϫ1 (represented by the 2Ј-OH of the all-ribose substrate shown in blue). Putative, catalytically important Mg 2ϩ ions are shown in magenta. For the Mg 2ϩ close to G292/G293/U294 in the L15/16 loop of M1 RNA, see Ref. 12, and references therein. The Watson-Crick base pairing interaction between G292/G293 of M1 RNA and Cϩ74/ϩC75 of tRNA (21,29) as well as the proposed interaction between U294 of M1 RNA and Uϩ73 of tRNA (30) have been depicted. Nt A248, A249, C252, C253, G332, and A333 were shown to be in close vicinity of the substrate nt Ϫ1 in E⅐S complexes, as revealed by short range cross-links (49,50). It should be mentioned here that Michael Harris and co-workers recently reported evidence (51) that the conserved A248 specifically interacts with the nucleotide at Ϫ1 (indicated by the dashed line). The transition state model for hydrolysis of the scissile phosphodiester connecting nt ϩ1 and Ϫ1 is derived from that proposed in (52), according to which Mg B directly coordinates to the pro-Rp phosphate oxygen and simultaneously interacts with the OH Ϫ nucleophile (in orange) and the 2Ј-OH at position Ϫ1. Alternatively, both Mg A and Mg B may directly coordinate to the pro-Rp oxygen, but Mg A instead of Mg B coordinates the OH Ϫ nucleophile (8,53), as indicated by the question mark next to Mg A . Sulfur substitutions at the canonical cleavage site, which essentially abolished catalysis in the presence of Mg 2ϩ , are depicted (S with crossed out red circle). M1 RNA-catalyzed cleavage of a substrate with a single sulfur substitution of the pro-Rp oxygen could be rescued by replacing Mg 2ϩ with Mn 2ϩ or Cd 2ϩ , whereas no such rescue could be obtained for a sulfur substitution of the pro-Sp oxygen (marked Sp); the Sp-phosphorothioate modification also weakened substrate ground state binding 30-fold (8). Sulfur substitution of the 3Ј-bridging oxygen abolished cleavage in the presence of Mg 2ϩ as well as thiophilic metal ions, such as Mn 2ϩ or Cd 2ϩ (54). The aforementioned findings do not rule out metal ion interactions with the pro-Sp oxygen and the 3Ј-bridging oxygen, as denoted by question marks. (39). The all-ribose substrate reacts 10-fold faster than a derivative with a 2Ј-F modification at nt Ϫ1, because there is donation of a hydrogen bond by the 2Ј-OH at nt Ϫ1 in the transition state (possibly to the 3Ј-bridging oxygen), although the 2Ј-OH has a lower electron withdrawing ability. The higher inductive effect of a 2Ј-F versus 2Ј-H substituent explains the higher reactivity of the ptRNA with a 2Ј-F versus 2Ј-H modification at nt Ϫ1. Here we assume that conformational differences of the sugar moieties with different 2Ј substituents do not play a major role due to the low energy barrier for interconversion (40) between the two major sugar puckering states [C3Ј-endo (N) and C2Ј-endo (S)].
Another issue is the hydrogen acceptor function of the 2Јfluoro group, which is predicted based on the availability of lone-pair electrons. There is x-ray structural evidence arguing against fluorine covalently bound to DNA as an effective H bond acceptor (41). This conclusion is based on the finding that a larger proportion of water molecules that form a hydrogen bond to an RNA 2Ј-OH group are located at 3 Å or less from the 2Ј-oxygen atom, whereas the majority of water molecules adopt a distance of Ͼ 3 Å to a 2Ј-fluorine despite the smaller van der Waals radius of fluorine relative to oxygen. Also, water molecules around 2Ј-hydroxyl groups display a more pronounced clustering than those around 2Ј-fluorines (41). However, hydrogen bonds accepted by a 2Ј-fluorine from a strong donor, such as a metal ion-coordinated water molecule, are considered to exist with certainty (41). If this kind of hydrogen bonding interaction takes place in the M1 RNA-catalyzed reaction is, however, unknown at present. Thus, it remains a possibility that the 10-fold lower reactivity of ptRNA with a 2Ј-fluorine relative to a 2Ј-hydroxyl at nt Ϫ1 includes contributions from differences in water structure around the 2Ј substituent and/or an impaired hydrogen bonding acceptor capacity of the 2Јfluorine. It has been shown for the Tetrahymena ribozyme that a hydrogen bond from the 2Ј-OH of A207 to the 2Ј-OH of U-1 stabilizes the transition state (42). Thus, the 2Ј-OH at nt Ϫ1 combines H bonding acceptor and donor functions in catalysis by the Tetrahymena ribozyme. Based on the identical strength of the 2Ј-fluorine effect (10-fold) in the M1 RNA-catalyzed reaction, it is not unlikely that the 2Ј-OH at nt Ϫ1 not only donates but also accepts a hydrogen bond during catalysis.
Metal Ion Binding at nt Ϫ1 and the 2Ј-Amino Substitution-The defect in Mg 2ϩ binding due to 2Ј substitutions at nt Ϫ1 (Ref. 7; Fig. 3) has led to the formulation of a model where hydrogen bond donation from this 2Ј-hydroxyl to a metal ioncoordinated water positions the latter for proton donation to the 3Ј-oxyanion leaving group (7). As outlined below, this model is consistent with our data and represents an alternative to the direct donation of a hydrogen bond from the 2Ј-OH to the 3Ј-bridging oxygen in the transtion state. Further evidence for the binding of a metal ion (termed Me B ; Refs. 8 and 12) in immediate vicinity of the 2Ј substituent at nt Ϫ1 comes from our results with the 2Ј-amino modification at this position, where miscleavage of ptRNA Gly C N U(A) at lower pH (Figs. 5 and 6) suggests electrostatic repulsion of metal ion Me B due to protonation of the 2Ј-amino group. Also, the 2Ј-amino group would have been expected to result in a much better catalytic performance than the 2Ј-deoxy modification due to its higher inductive effect and the capability to function in hydrogen bonding. However, ptRNA Gly C N U was cleaved only at a ϳ2-fold higher rate than ptRNA Gly C H U at pH values Ն7.0 ( Fig. 3A; data for pH 8.0 not shown) where protonation of the 2Ј-amino group is expected to be negligible. Several reasons may contribute to the low catalytic performance of ptRNA Gly C N U. For example, the 2Ј-N substituent may constrain the conformation of the ribose at nt Ϫ1, may create an artificial interaction or may have unfavourable effects on water structure around this functional group. It is interesting to note in this context that the 2Ј-amino modification also blocked hydrolysis by RNase A, although this reaction follows a different mechanism (nucleophilic attack of the 2Ј substituent on phosphorus) than the one catalyzed by RNase P. It was proposed that steric factors, possibly related to the favored C2Ј-endo pucker of the nucleoside analog, may prevent the 2Ј-amino group from being favorably placed in the RNase A active site for nucleophilic attack on phosphorus (43,44).
Remarkable is our finding (Table II) that a C to U exchange at nt Ϫ1 reduced miscleavage at pH 5.0 from 77% (ptRNA Gly C N U) to 14% (ptRNA Gly U N U) or from 92% (ptRNA Gly C N A) to 53% (ptRNA Gly U N A). This correlates with the very low single turnover rate of cleavage for substrates with a 2Ј-amino-2Ј-deoxy C residue at Ϫ1 and in the presence of Mg 2ϩ (Table III). One explanation could be that effective electrostatic repulsion of Mg B (Fig. 7) requires simultaneous protonation of the 2Ј-amino group at nt Ϫ1 and of the cytidine N3 function (unperturbed pK a : ϳ4.5; Ref. 40). The pK a of the N3 function may be significantly perturbed, as observed by NMR for a cytidine in a ribosomal A loop module (apparent pK a ϳ6.4), where N3 protonation also caused subtle conformational changes of A loop structure (45). Likewise, protonation of adenosine N1 (unperturbed pK a close to 4.0; Ref. 40) may be the reason for increased miscleavage when Uϩ73 is exchanged for A (Table II). In summary, our results document an unexpectedly strong dependence of miscleavage on the base identities at nt Ϫ1 and ϩ73, with miscleavage favored in the order C-1 Ͼ U-1 and A73 Ͼ U73, suggesting that the 2Ј substituent at nt Ϫ1, catalytic metal ion cofactors as well as base functional groups of nt Ϫ1 and ϩ73 constitute an intricate and sensitive network in the active site, consistent with results from a recent in depth analysis of the interaction between the substrate nt ϩ73 and nt 294 of M1 RNA (31).
In any case, a 2Ј-amino modification at nt Ϫ1 showed fundamentally different effects in reactions catalyzed by E. coli M1 RNA (this study) and the Tetrahymena ribozyme (39), thus revealing substantial differences in transition state complex geometry for the two ribozymes. In the Tetrahymena ribozyme reaction, a substrate with a 2Ј-amino modification at nt Ϫ1 (termed rS N ) reacted even faster than the all-ribose substrate (rS OH ) at lower pH (between pH 4.5 and 6.5), i.e. under conditions where a substantial proportion of the 2Ј-amino group is protonated (the apparent pK a was determined to be 5.0; Ref. 39). These findings ruled out that the 2Ј substituent at nt Ϫ1 is involved in binding of a metal ion, because a NH 3 ϩ group has no lone pair electrons for a direct metal ion interaction and the positive charge would have been expected to repulse a metal ion. Furthermore, the rS N substrate was cleaved with roughly the same catalytic efficiency as rS OH at pH 7.0, where protonation is low or negligible and another step than the chemical starts to become rate-limiting for cleavage of rS OH (39). In contrast to the Tetrahymena ribozyme, protonation of the 2Јamino group in ptRNA Gly C N U was apparently detrimental to cleavage in the reaction catalyzed by M1 RNA. Also and again in contrast to the Tetrahymena ribozyme, ptRNA Gly C N U was cleaved at extremely low rates at pH 7.0, where, analogous to the Tetrahymena ribozyme reaction (see above), another step than the chemical begins to become rate-limiting for cleavage of the all-ribose substrate. In conclusion, all existing experimental evidence supports the notion that a metal ion binds next to the 2Ј substituent at nt Ϫ1 in the reaction catalyzed by bacterial RNase P RNA, but not in that catalyzed by the Tetrahymena ribozyme.
RNase P RNA Active Site Plasticity-Our results have dem-onstrated a close and sensitive communication between the 2Ј substituent at nt Ϫ1, the nature (Mg 2ϩ or Mn 2ϩ ) of catalytic divalent metal ion(s) and, surprisingly, the base functional groups of nt Ϫ1 and ϩ73 (Tables II and III; Figs. 6 and 7), suggesting a rather dynamic structure of catalytic M1 RNAsubstrate complexes. The disruption of the Cϩ75/G292 base pair in E⅐S complexes caused miscleavage of ptRNA Gly C H U but not of all-ribose ptRNA Gly CU (Fig. 4D), demonstrating the communication between the 2Ј substituent at nt Ϫ1 and the tRNA 3Ј-end "NCCA" interaction with M1 RNA. Our results are even more remarkable since our substrate, a bacterial ptRNA Gly , is a simple class I tRNA with a canonical 7-bp acceptor stem. In comparison, another natural ptRNA substrate that has been extensively characterized, namely E. coli ptRNA Tyr Su3 (2), is more complex because it has the capacity to form a 9-bp acceptor stem and its specific 5Ј-precursor sequence plays an important role in complex formation with M1 RNA (46). Irrespective of such differences in substrate nature, our findings are consistent with those of Leif Kirsebom and co-workers (reviewed in Refs. 2, 13, 31, and references therein) who have characterized in large detail the intricate and flexible interplay of metal ion cofactors with several structural determinants in the substrate and ribozyme near the cleavage site.
The emerging picture of a highly dynamic nature of the RNase P RNA catalytic site correlates with the peculiar biological role of the enzyme. Bacterial RNase P (RNA) differs from other ribozymes, such as the self-splicing group I introns, in its natural trans-cleavage activity and its biologically important capacity to recognize a large variety of ptRNAs, even including non-tRNA substrates in some organisms (reviewed in Ref. 3). The enzyme has to act with similar efficiency on ptRNAs carrying unrelated 5Ј-flanking sequences and which differ in structural details of their mature domains. Some ptRNAs even have to be processed at a site shifted by 1 nt in order to generate a mature tRNA product with an 8-bp instead of a 7-bp long acceptor stem (tRNA His , tRNA SeCys ; Refs. 47 and 48). This broad substrate specificity is thought to have brought about the consequence of a rather flexible active site architecture compared with other large but more specialized ribozymes, such as the group I self-splicing introns. The dynamic nature of RNase P RNA is also a likely explanation for the difficulties associated with attempts to crystallize this ribozyme, to define the constituents of its active site architecture and to pinpoint the number and exact location of catalytic metal ion cofactors.