Quantitative studies of Mn²⁺-promoted specific and non-specific cleavages of a large RNA: Mn²⁺-GAAA ribozymes and the evolution of small ribozymes

Abstract

Manganese (Mn²⁺) promotes specific cleavage at two major (I and III) and four minor (II, IV, V and VI) sites, in addition to slow non-specific cleavage, in a 659-nucleotide RNA containing the Cr.LSU group I intron. The specific cleavages occurred between G and AAA sequences and thus can be considered Mn²⁺-GAAA ribozymes. We have estimated rates of specific and non-specific cleavages under different conditions. Comparisons of the rates of major-specific and background cleavages gave a maximal specificity of approximately 900 for GAAA cleavage. Both specific and non-specific cleavages showed hyperbolic kinetics and there was no evidence of cooperativity with Mn²⁺ concentration. Interestingly, at site III, Mg²⁺ alone promoted weak, but the same specific cleavage as Mn²⁺. When added with Mn²⁺, Mg²⁺ had a synergistic effect on cleavage at site III, but inhibited cleavage at the other sites. Mn²⁺ cleavage at site III also exhibited lower values of K (Mn²⁺ requirement), pH-dependency and activation energy than did cleavage at the other sites. In contrast, the pH-dependency and activation energy for cleavage at site I was similar to non-specific cleavage. These results increase our understanding of the Mn²⁺-GAAA ribozyme. The implications for evolution of small ribozymes are also discussed.

INTRODUCTION

Divalent metal cations are generally essential for RNA structure, function and catalysis (1–5). These ions can non-specifically surround RNA and shield the negative charges of phosphates. Divalent cations are also located at specific sites in RNA, often where sharp bending of the phosphate backbone or non-canonical helices occur (6–12). Specific metal binding typically involves not only phosphates, but also nucleotide bases of RNA (3) and are important in interlacing distant structural domains (4,5). In addition to the structural role, metals also play central roles in the reactions catalyzed by certain ribozymes (1,5).

The simplest ribozyme-catalyzed reaction is the cleavage of a phosphodiester bond by the adjacent 2′-OH to form a 2′,3′-cyclic phosphate and a 5′-OH, as exemplified by the reactions catalyzed by small naturally occurring ribozymes (e.g. hammerhead, HDV and hairpin ribozymes) (13,14). It has been proposed that Mg²⁺ catalyzes the reaction by polarizing the 2′-OH group, which then initiates nucleophilic attack on the phosphorus, and by stabilizing the incipient transition state (15,16). Divalent metals such as Pb²⁺ (17), Zn²⁺ (18,19), Mn²⁺ (20) and even Mg²⁺ (19,21) can also promote non-biologically relevant site-specific cleavages and/or non-specific degradation of RNA, depending on the conditions. These metal-promoted scission events are similar to the small-ribozyme promoted cleavages, because both leave 2′,3′-cyclic phosphate and 5′-OH termini. Perhaps, nature has utilized the intrinsic reactivity of the 2′-OH and through natural evolution the inefficient and less specific cleavage complexes have lead to the efficient ribozymes we see today. In fact, the in vitro selection of the leadzyme (Pb²⁺-ribozyme) can be regarded as mimicking such an evolutionary process (22).

The smallest, but not well-understood, ribozyme is the Mn²⁺-GAAA ribozyme (23). The minimal structure for activity was shown to be a seven-nucleotide motif, GAAA/UUU (the G is unpaired and the A/Us are Watson–Crick pairs) (24). Mn²⁺ or Cd²⁺ (but not Mg²⁺ or other divalent metals) promoted specific cleavage between G and AAA (23,24). Although Mn²⁺ and Mg²⁺ have similar physicochemical properties and for many natural ribozymes Mn²⁺ can replace Mg²⁺ in supporting their functions (5), this ribozyme prefers Mn²⁺. Mn²⁺ is ‘softer’ and coordinates better with nucleotide bases than Mg²⁺ (3,5), suggesting that there might be sequences (or motifs) which Mn²⁺ preferentially binds and cleaves. There have been comparatively few studies of this ribozyme (25,26), although it has been found in some interesting places (27,28).

Further study of the Mn²⁺-GAAA ribozyme is of interest for several reasons. First, since specificity is an important criterion for enzymes and Mn²⁺ also promotes non-specific RNA cleavage, it is important to know how well GAAA is distinguished from other sequences. Second, the GAAA tetramer is known for forming a stable tetraloop (29), which can also dock with other RNA structures as part of tertiary folding (30,31). Also, partially- to highly-conserved GAAA sequences are found in certain small [e.g. hammerhead (32) and hairpin (33)] and large [e.g. group I introns and RNase P (34,35)] ribozymes. Thus, the question arises as to whether these GAAA sequences bind Mn²⁺ or other metals. Third, in the early stages of the proposed ‘RNA World’ (36,37), the complexity of RNA molecules was probably <100 nt (38). The small size of the Mn²⁺-GAAA ribozyme suggests that specific metal-catalyzed RNA cleavage could have been a metabolic tool of the early ribo-organisms (28). Thus, it is of interest to know how a very small ribozyme enhances cleavage rates.

We detected multiple specific cleavages of a large RNA (23S.3 RNA) with Mn²⁺, while studying the in vitro self-splicing of the group I intron in the chloroplast 23S rRNA gene of Chlamydomonas reinhardtii, Cr.LSU (39,40). Upon mapping the cleavage sites in this 659 nt RNA, we found that they all contained GAAA. Thus, we devised a method to estimate the rates of specific cleavage at several sites and background cleavage under different Mn²⁺/Mg²⁺ concentrations, pH and temperature regimes. These data quantify the specificity of cleavage at GAAA sequences and reveal new properties and variants of the Mn²⁺-GAAA ribozyme.

MATERIALS AND METHODS

In vitro RNA synthesis and purification

Construction of plasmid pGEM23S.3 was described previously (39). The plasmid contains a 623-bp insert in the pGEM3zf(+) vector (Promega). The insert is comprised of a shortened form of the Cr.LSU intron (522 bp, lacking most of the ORF) flanked by exonic sequences. To synthesize the full-length RNA (659 nt), the plasmid was linearized with HindIII. For some experiments, it was digested with XhoI to produce a 3′-truncated RNA of 481 nt. The transcription reactions (50 µl) were incubated at 37°C and contained 1 mM of each rNTP (A,U,C,G), 3 mM MgCl₂, 2 mM spermidine, 10 mM dithiothreitol (DTT), 40 mM Tris–HCl pH 7.5, 2 µg of linearized plasmid DNA and 100 U of T7 RNA polymerase (Boehringer Mannheim). The relatively low MgCl₂ concentration was to prevent self-splicing of the RNA during transcription. To prepare internally ³²P-labeled RNA, 10 µCi (1 µl) of [α-³²P]GTP (ICN; specific activity, ∼3000 Ci/mmol) were added. After 2 h, the RNA was extracted with phenol/chloroform and precipitated twice with ethanol.

Full-length transcripts were purified by denaturing PAGE (41). The RNA band was located by UV shadowing or by exposing the gel to X-ray film, excised and eluted by incubating the gel slices in 10 mM Tris–HCl pH 7.5, 0.2 M NaCl, 1% (w/v) SDS, 1 mM EDTA at 37°C for several hours. The RNA was precipitated with ethanol after the addition of carrier glycogen (Boehinger Mannheim), re-precipitated twice with sodium acetate and ethanol and dissolved in 10 mM Tris–HCl pH 7.5.

Mn²⁺-dependent cleavage

The standard reaction mixture for Mn²⁺-dependent cleavage contained 3 mM MnCl₂ (MnCl₂·4H₂O, BDH), 0.2 M KCl, 50 mM Tris–HCl pH 7.0 (adjusted at 47°C; pK_a of Tris is 7.5 at 47°C) and 1–15 nM ³²P-labeled RNA (∼1000–15 000 c.p.m.) in a final volume of 5 µl. Variations in these conditions are indicated in the legends or text. Immediately prior to the reaction, the RNA (in 10 mM Tris–HCl pH 7.5) was denatured by heating to 95°C for 1.5 min, followed by controlled cooling (–0.5°C/s) to 25°C. Reaction mixtures were set up by aliquotting the renatured RNA into non-siliconized, 0.65-ml microfuge tubes (PGC Scientific), followed by an aliquot of a freshly prepared mixture of the Tris–HCl, MnCl₂ and KCl. Pre-mixing MnCl₂ with monovalent salt and buffer for long periods of time was avoided since it can accelerate the oxidation and precipitation of Mn²⁺. The standard reactions were incubated at 47°C for 0–100 min, stopped by adding 1.2 vol of gel loading buffer [80% (v/v) formamide, 0.1 M EDTA pH 8.0, 0.1% (w/v) xylene cyanol, 0.1% (w/v) bromophenol blue], heated to 65°C for 3 min, quick-cooled and separated by denaturing PAGE (41).

The following buffers were used to examine the pH dependence: MES, adjusted with NaOH, for pH 5.5, 6.0 and 6.5; HEPES, adjusted with NaOH, for pH 6.5, 7.0 and 7.5; Tris, adjusted with HCl, for pH 7.0, 7.5 and 8.0; and (2-N-cyclohexylamino)ethanesulfonic acid, adjusted with NaOH (CHES) for pH 8.0, 8.5 and 9.0. The final concentration of each buffer was 50 mM. Tris–HCl was used in the temperature-dependence experiments. The pH of the buffers was adjusted at 47°C (or other usage temperature).

Kinetics of cleavage at multiple sites

The cleavage reactions, typically ∼5000–8000 c.p.m. of internally ³²P-labeled RNA, were separated by denaturing PAGE on long (50 cm) sequencing gels. All of the specific cleavage products and the unreacted 23S.3 RNA were quantified with a Phosphorimager (Molecular Dynamics) and ImageQuant software (Molecular Dynamics). A range of uncleaved RNA amounts (10–15 000 c.p.m.) were analyzed in parallel and used to establish a standard curve for obtaining the c.p.m. of the RNA bands.

Initial rates of cleavage at specific sites were used to obtain the k_obs. In general, data from cleavage of less than the first 20% of the input RNA was used, and under most conditions >80% of the RNA cleaved during this time was at specific sites. However, at the upper extremes of pH, Mn²⁺ concentration and temperature, non-specific degradation became significant (up to 60% of the cleaved RNA). Thus, the measurements of specific cleavage products (in c.p.m.) were corrected (in all conditions) for the contribution of non-specific cleavage. The rate of non-specific cleavage of 23S.3 RNA was also of interest for comparative purposes and was estimated as follows. It was assumed that the rate of non-specific cleavage was approximately equal for all phosphodiesters in 23S.3 and that at the early time points degradation of the intact RNA was principally by single, non-specific cleavage events. Thus, the following equation applied:

ln((ΣR)/R₀) ≈ –658·k_obs·t 1

where R₀ is the initial amount of radioactive 23S.3 RNA, ΣR is the sum of the specific cleavage products and the remaining intact RNA (i.e. the RNA that was not non-specifically cleaved), 658 is the number of phosphodiesters in 23S.3 and t is time. The ratio (ΣR)/R₀ is related to the fraction of 23S.3 RNA non-specifically cleaved by the following:

f_nsp = 1 – ((ΣR)/R₀) 2

where f_nsp is the fraction of 23S.3 non-specifically cleaved at the early time points. This ratio declines according to the rate of non-specific cleavage of the RNA. Thus, the slopes of semi-log plots of (ΣR)/R₀ with time provided the k_obs for non-specific cleavage.

The direct measurements of specific cleavage products provided an underestimate of the real site-specific cleavage rate because the specific cleavage products were also subjected to non-specific cleavage during the reaction. Thus, to obtain the k_obs for cleavage at a specific site, the term S, which gives the corrected amount of a specific cleavage product, was used:

S = [(R_f /m_f)/(R₀ /m_P)]/[1 – f_nsp·(m_f /m_p)] 3

where R₀ is the amount (in c.p.m.) of the initial 23S.3 RNA, R_f is the amount of the specific cleavage fragment, m_P is the number of Gs in 23S.3 RNA, m_f is the number of Gs in the cleavage fragment and f_nsp is the fraction of 23S.3 non-specifically cleaved (which is described by equation 2). The components of ‘S’ can be understood as follows: the numerator, [(R_f /m_f)/(R₀ /m_P)], is the observed fraction of 23S.3 RNA specifically cleaved at the site in question, whereas the denominator, [1 – f_nsp(m_f /m_p)], corrects for the effect of non-specific cleavage on the specific cleavage product. The latter component is based on equation 2 and takes into account the fact that the rate of disappearance of the specific cleavage product is length-dependent [hence, the (m_f /m_p) term]. Note that S is independent of reaction volume and has no units. At the initial time points, dS/dt ≈ k_obs. Thus, a reasonably good estimate of k_obs was obtained from the slope of S versus initial time plots. Specific cleavage at different sites in 23S.3 RNA, which were estimated simultaneously, were treated as independent events. The rate constants determined by this method were not significantly affected by RNA concentrations over the range employed (1–50 nM). The derivation and validity of these equations was discussed in more detail elsewhere (42). Kaleidograph software (Synergy Software) was used for curve fitting.

Primer-extension analysis

Primer-extension analysis of RNA was performed as described by Christopher and Hallick (43), except for the following modifications. The reactions contained 0.6 µCi of [α-³²P]dCTP (ICN; specific activity, 3000 Ci/mmol), 100–150 ng of RNA and 10 ng of non-radioactive oligonucleotide primer. The following oligonucleotides were used: #8, (5′-dATATTTTGTATTGATAAGATG-3′), which anneals to nt 176–196 of the intron in 23S.3; #37, (5′-dCAGGAGTCGGCGTATTA-3′), which anneals to nt 406–422 of the intron in 23S.3; and #95, (5′-dTGCCTGCAGGTCGACTCTAGA-3′), which anneals to the 3′ exon, 16–36 nt from the 3′ splice site. The extension reactions were performed for 30 min at either 42°C (oligo #8), 47°C (oligo #37) or 50°C (oligo #95). Sequence ladders were generated by adding a dideoxynucleotide (final concentration, 10 µM) to reactions with untreated RNA. The reactions were stopped by adding 7.5 µl of gel loading buffer (see above), heating to 95°C and denaturing PAGE.

Enzymatic sequencing of end-labeled RNA

Non-radioactive 23S.3 RNA was prepared, cleaved with Mn²⁺ under standard conditions, and then end-labeled with T4 polynucleotide kinase and [γ-³²P]ATP (41). The RNAs were purified by denaturing PAGE as described above, cleaved with base-specific RNases (44) and the products separated on 20% denaturing gels. The wet gels were transferred to used X-ray film, wrapped in Saran wrap and exposed to X-ray film (BioMax MS, Kodak) at –70°C with DuPont Cronex intensifying screens.

RESULTS

Mn²⁺-dependent cleavage of 23S.3 RNA

Figure 1A is a linear diagram of 23S.3 RNA, showing the positions of the specific cleavage sites (identified below). Figure 1B shows the cleavage pattern obtained when 23S.3 RNA was incubated with Mn²⁺ under standard cleavage conditions (3 mM MnCl₂, 0.2 M KCl, 50 mM Tris–HCl pH 7.0, 47°C) for 0–80 min. There were four major fragments and a number of minor ones. Detailed mapping revealed six cleavage sites, I–VI (see below). The major cleavage sites are I and III, and the minor ones, sites II, IV, V and VI (Fig. 1A). Accordingly, the RNAs derived from a single cleavage of the precursor are annotated by site and whether they contain the 5′ or 3′ end of the RNA [e.g. I(5′), I(3′), etc.]. RNAs produced by cleavage of 23S.3 at two sites are denoted with a hyphen; for example, the 228-nt RNA denoted I-III, is produced by cleavage at both major sites (I and III). All of the products of single cleavages, with the exception of the 558-nt site VI(5′) fragment (which was too weak to be identified with confidence) and most of the expected products of double-cleavage (one major-one minor) could be found. Finally, the RNAs are also named in order of descending size (f1, f2, etc.).

Figure 1.

Mn²⁺-dependent cleavage of 23S.3 RNA. (A) Map of 23S.3 RNA and the positions of Mn²⁺-promoted cleavage sites. The transcript is 659 nt, consisting of a shortened Cr.LSU intron (line, 522 nt) and flanking exons (open boxes, 93-nt 5′ exon and 44-nt 3′ exon) (36). The major Mn²⁺-cleavage sites are indicated with solid arrows and the minor sites with open arrows. Below are diagrams of the two major cleavage events (sites I and III), showing the nomenclature of cleavage products. (B) Time course of the cleavage of internally ³²P-labeled 23S.3 RNA. The reaction was performed under standard conditions and analyzed by denaturing PAGE and autoradiography. The sizes (in nt, on right) of these products were first estimated by comparing to size markers with an accuracy to ± 3 nt (data not shown), and then determined from the locations of six cleavage sites (Fig. 2). The cleavage products were named according to fragment size (right) or identity (left). The autoradiograph was overexposed to reveal the presence of weak cleavage products.

The parameters of the standard cleavage reaction, i.e. MnCl₂, monovalent salt, pH and temperature, were chosen to obtain the greatest amount of specific cleavage in a 40-min reaction with minimal requirements. The effects of Mn²⁺ concentration, pH and temperature on the specific cleavage of 23S.3 RNA are presented below. Monovalent salts (KCl or NaCl), up to at least 0.5 M, increased the proportion of specific cleavage by reducing background cleavage (data not shown).

Localization of the cleavage sites

To facilitate mapping of the cleavage sites, preliminary experiments were performed by cleaving 5′-end-labeled 23S.3 RNA with Mn²⁺ or by post-cleavage labeling of non-radioactive cleavage products with [γ-³²P]ATP and polynucleotide kinase. These results indicated that the major cleavage sites were ∼150 and ∼375 nt, respectively, from the 5′-end of 23S.3 and that minor cleavage sites were located between the two major sites and between the second major site and the 3′-end of 23S.3 RNA (data not shown).

This information was used to design oligonucleotides for primer extension analysis (Fig. 2A). Figure 2B shows the results with oligonucleotide #8. A major product was obtained specifically with Mn²⁺-treated RNA (Fig. 2B, lanes 6 and 7), which indicated that cleavage occurred between G¹⁴⁹ and A¹⁵⁰. To precisely locate the other major cleavage site (and minor sites), oligonucleotide #37 (Fig. 2A) was used. Figure 2C, lanes 6 and 7, show that a doublet of major bands and two minor products were specifically obtained with Mn²⁺-treated RNA. The positions of the doublet suggested that there were adjacent cleavage sites between U³⁷⁶ and G³⁷⁷ and between G³⁷⁷ and A³⁷⁸, respectively (however, see below). The positions of the two minor cDNAs indicated that cleavages also occurred between G³⁴⁴ and A³⁴⁵ and between G³⁸⁵ and A³⁸⁶. Primer extension was also performed on the f7 cleavage product purified by denaturing PAGE. The result (Fig. 2C, lane 5) was the same as that obtained with total, cleaved 23S.3 RNA (Fig. 2C, lanes 6 and 7), confirming that f7 is a product of the second major cleavage. The absence of the two minor cDNA products from the extension of f7 is consistent with their resulting from cleavages of 23S.3 RNA at other sites. Oligonucleotide #95 was used to map cleavage sites in the 3′ portion of 23S.3 RNA. Figure 2D and E show that Mn²⁺-dependent cleavage occurred between G⁴⁹² and A⁴⁹³ and between G⁵⁵⁸ and A⁵⁵⁹, respectively. The signal at the latter site was weak, but reproducible nonetheless.

Figure 2.

Localization of the Mn²⁺-cleavage sites by primer extension and direct RNA sequencing. (A) Map of 23S.3 RNA showing the locations of the oligonucleotides used for primer extension. (B–E) Primer extension analysis with oligonucleotide #8 (B), #37 (C) and #95 (D and E). Unlabeled 23S.3 RNA was cleaved with Mn²⁺ under standard conditions for the indicated times and then primer extension was performed using [α-³²P]dCTP. The sequence ladders (U, A, C, G) were generated with the same primers on untreated 23S.3 RNA. Solid arrows, extension products induced by Mn²⁺. The corresponding 23S.3 sequences are indicated to the left and the cleavage sites marked by asterisks. In lane 5 of (C), primer extension was performed on gel-purified f7 RNA (Fig. 1). (F and G) Direct sequencing of the 5′-termini of cleavage products f10 (F) and f7 (G), respectively. The end-labeled RNAs (Fig. 1C, lane 2) were gel-purified, treated with base-specific RNases and analyzed on a denaturing 20% gel. Lanes: M, [γ-³²P]ATP; S, starting (untreated) RNA; L, ladder generated by alkaline hydrolysis; C, RNA treated with RNase CL₃ (digests at Cp↓N); A/U, RNA treated with RNase PhyM (digests at Ap↓N and Up↓N); G, RNA treated with RNase T₁ (digests at Gp↓N); and A, RNA treated with RNase U₂ (digests at Ap↓N). The sequences of the 5′-ends of the RNAs are indicated to the right.

Direct RNA sequencing was used to independently confirm positions of the Mn²⁺ cleavage sites. The size (∼230 nt) of the f10 RNA (Fig. 1) suggested that it was derived by double-cleavage of 23S.3 RNA at the two major sites. This RNA was isolated after Mn²⁺ cleavage and 5′-end-labeling and subjected to enzymatic sequencing. The results, which are shown in Figure 2F, indicate that the first 5 nt of f10 are 5′-AAACC-3′. This result is consistent with the primer extension analysis and confirms the position of the first major cleavage site. The primer-extension data also suggested that the second major cleavage site consisted of two adjacent cleavages, one preceding and one following G³⁷⁷, and that there was approximately equal cleavage at these two linkages. This was examined further by subjecting end-labeled f7 RNA to enzymatic sequencing. The results, which are shown in Figure 2G, indicate that this RNA begins with As and there was no evidence for an RNA with a 5′-terminal G. These data indicate that the cleavage at the second major site is not a doublet but is similar to the others in occurring primarily between G and A. The doublet of bands produced during primer extension can be attributed to reverse transcriptase adding a non-templated nucleotide to the end of the cDNA (45). It is not known why this occurred more efficiently with this substrate than with the other major extension products. Finally, cleavage after G⁴⁹² (Fig. 2D) was examined by direct sequencing of end-labeled f13 RNA (Fig. 1B) and the results were consistent with the primer extension analysis (data not shown).

Table 1 shows the sequences at the six cleavage sites; GAAA is the consensus sequence and cleavage occurred between G and AAA. It should also be noted that these are the only GAAA sequences in 23S.3 RNA. Thus, specific cleavage of 23S.3 RNA by Mn²⁺ occurred primarily and at all GAAA sequences.

Table 1. Sequences at the Mn²⁺ cleavage sites in 23S.3 RNA.

Site	Sequencea	Locationb
Ic	GGU149G↓AAACCU	J4/5
II	UCU344G↓AAACCG	P6b extension
IIIc	AGU377G↓AAAAAA	P6b extension
IV	AAA385G↓AAAUCG	P6b
V	UGG492G↓AAAGUA	P7.2
VI	UAC558G↓AAAGUA	L9

^aArrow is the cleavage site; consensus sequence is underlined.

^bRefers to secondary structures of the Cr.LSU intron.

^cMajor cleavage site.

Quantitative characterization of cleavage events

The 23S.3 RNA decreased with first order kinetics (t = 13 ± 1 min under standard conditions; Fig. 1B) during Mn²⁺-cleavage and was apparently homogeneous. Thus, we developed an approach to estimate the observed rate constants (k_obs) for cleavage at multiple sites simultaneously using data collected at early time points. Figure 3A shows the early accumulation of products of specific cleavage at sites I, III, IV and V; cleavage rates at sites II and VI were too low to be reliably estimated. All four lines are close to linear over this period. Under most conditions, >80% of the cleavage at early time points was due to site-specific cleavage; however, at the upper extremes of Mn²⁺ concentration, pH and temperature, non-specific cleavage became significant. Thus, the measured values of specific cleavage products were corrected for the effect of non-specific degradation (see Materials and Methods). This practice significantly expanded the range of each parameter that was analyzed. It also provided non-specific cleavage rates for comparisons.

Figure 3.

Kinetics of Mn²⁺-dependent cleavage at sites I, III, IV and V: dependence on Mn²⁺ concentration. (A) Quantification of the accumulation of specific cleavage products during the initial time points. Mn²⁺ cleavage of 23S.3 RNA was performed as described in Figure 1B. CPM, counts per minute. Rates of specific cleavage at sites I, III, IV and V were obtained by measuring the accumulation of cleavage products I(5′), III(3′), IV(3′) and V(3′), respectively. These fragments were chosen because I(5′) does not contain any other specific cleavage sites and, although III(3′), IV(3′) and V(3′) do, secondary cleavage was insignificant (<10% of primary cleavage). (B) The observed rate constants (k_obs) for cleavage at sites I, III, IV, V and non-specific were determined as described in Materials and Methods. The cleavage conditions were standard except for varying the Mn²⁺ concentration. The data were fit with the equation k_obs = k_cat ([Mn²⁺]/(K + [Mn²⁺])), where k_cat is the rate constant for RNA cleavage at saturating Mn²⁺ concentrations and K is the Mn²⁺ concentration at which k_obs is one-half of k_cat. (C) The specificity of cleavage at sites I and III as a function of Mn²⁺ concentration. Specificity is the ratio of k_obs for site-specific cleavage to k_obs for non-specific cleavage. The data are from (B).

Figure 3B shows the rate constants (k_obs) for cleavage at sites I, III, IV and V as a function of Mn²⁺ concentration. Hyperbolic curves were obtained and there was no evidence of cooperativity. Cleavage rates at the major sites (I and III) plateaued at ∼3 mM Mn²⁺. The slopes of log–log plots of the data before saturation gave values close to one for cleavage at the specific sites and for non-specific cleavage.

The kinetic parameters, k_cat and K, for cleavage at each site are given in Table 2. Site I had the highest k_cat and site IV the lowest, but with only a 6-fold difference between them. The lowest K (0.3 mM) belonged to site III, suggesting a greater affinity for the metal at this site compared to the others. Thus, although the k_cat for site III is only 1.8-fold greater than for site V, the considerably greater cleavage at site III under standard conditions (Fig. 1) can be explained by the fact that the Mn²⁺ concentration was nearly saturating for site III, but not for site V.

Table 2. Kinetic parameters for Mn²⁺-dependent cleavage of 23S.3 RNA at sites I, III, IV and V.

	I	III	IV	V	Non-specific
kcat (× 103 min–1)	18	9	3	5	0.14
Ka (mM)	1	0.3	3	3	7

The values are means of four determinations that varied no more than 20%.

^aK is the Mn²⁺ concentration at which k_obs = k_cat.

Figure 3C shows how cleavage specificity varied as a function of Mn²⁺ concentration for the major cleavage sites. Specificity is defined as the ratio of k_obs for specific cleavage over the k_obs for non-specific cleavage. The graphs show that specificity peaked at low (0.3–0.5 mM) Mn²⁺ concentrations and the highest specificity was for site III (0.3 mM Mn²⁺), where the cleavage rate was ∼900-fold greater than non-specific cleavage.

Figure 4A shows the effect of varying pH on Mn²⁺-dependent cleavage. Cleavage rates at specific sites (and non-specific cleavage) increased linearly from pH 5.5 to 8.5, except for site I, which plateaued at pH 8 (Fig. 4A). The slopes of the pH–k_obs plots in the linear portion of the curves were 1 for site I and 0.8 for sites IV, V and non-specific cleavage. These results indicate that, for cleavage at these sites, a single deprotonation event is the major rate-limiting step (46). Interestingly, the slope of the pH–k_obs plot for site III was only ∼0.5. One possible explanation of such a low pH–rate dependency is that a protonation step (which is required for the departure of the 5′-OH leaving group) is not much faster than the deprotonation step. In this regard, we note the evidence of Zhou et al. (47) that departure of the 5′ leaving group was limiting for a hammerhead ribozyme. The rate plateau at high pH could result from some precipitation/oxidation of the Mn²⁺ (48) or a change in the rate-limiting step.

Figure 4.

The pH and temperature effects. (A) The pH-dependence of Mn²⁺-dependent cleavage at sites I, III, IV and V (compared to non-specific, nsp, cleavage). 23S.3 RNA was incubated under standard cleavage conditions except for varying the pH from 5.5 to 9.0. The reactions between pH 6.5 and 8.0 were performed with two different buffers, but the data were plotted with the same symbol. The lines were drawn using regression analysis on the linear portions of the curves. (B) Arrhenius plots of the effect of temperature on Mn²⁺-dependent cleavage at sites I, III, IV and V (compared to nsp cleavage). 23S.3 RNA was cleaved under standard conditions, except for varying the temperature from 25 to 70°C. The lines were drawn by performing regression analysis on the data obtained from 30 to 65°C.

The effect of temperature on Mn²⁺-dependent cleavages are shown in Figure 4B. The Arrhenius plots for cleavage at sites I, III, IV and V are linear over at least a 30°C range, suggesting that the same rate-limiting step was involved at different temperatures. The activation energies, E_a, for Mn²⁺-dependent cleavage, obtained from the slopes of the plots, are 29.9, 15.1, 29.6 and 32.8 kcal/mol for cleavage at sites I, III, IV and V, respectively. The activation energies for sites I and IV are similar to non-specific (29.6 kcal/mol) and, therefore, most of the rate enhancement is attributable to an increase in the frequency factor, which can be described as the frequency of collisions with the correct orientation. It is also interesting to note that the activation energy for site III cleavage is substantially (∼2-fold) lower than for the other sites, including site I.

Effects of Mg²⁺on Mn²⁺-dependent cleavage

Figure 5A shows the effects of increasing Mg²⁺ concentration on Mn²⁺-dependent cleavage of internally-labeled 23S.3 RNA. The cleavages at sites I, IV and V were inhibited by Mg²⁺ (>1 mM), whereas cleavage at site III was first stimulated at low Mg²⁺ (<25 mM) and then gradually decreased at higher Mg²⁺ concentrations. Interestingly, as Figure 5A and B (lanes 11, 12 and lane 10, respectively) shows, Mg²⁺ alone promoted cleavage at site III and at the same phosphodiester as Mn²⁺.

Figure 5.

The effects of Mg²⁺ on Mn²⁺-dependent cleavage at specific sites. Mg²⁺ induces cleavage at site III. (A) Analysis with ³²P-labeled RNA. Internally-labeled 23S.3 RNA was incubated under standard cleavage conditions plus the indicated concentrations of MgCl₂ in mM (lanes 2–10). The RNA was also incubated with MgCl₂ as sole divalent cation (lanes 11 and 12) and without divalent cations (lane 1). The major Mn²⁺-cleavage products are labeled to the left as in Figure 1. The group I ribozyme-related reaction products are: C, circularized intron; E–I, 5′ exon–intron; I, linear intron; 5′E, 5′ exon. The RNA species (*) migrating just above the I-III product is due to Mn²⁺ cleavage of the linear intron at site III. The sizes of the RNAs are given (in nt) between lanes 10 and 11. (B) Primer extension analysis of non-radioactive RNA. Unlabeled 23S.3 RNA was incubated with Mn²⁺ and/or Mg²⁺ at the indicated concentrations (the other conditions were standard) and then primer extension was performed with oligonucleotide #37 (Fig. 2) and [α-³²P]dCTP. The extension products derived from cleavage of the RNA at sites III and IV are indicated to the left. The sequence ladder (lanes 1–4) was generated with the same primer on untreated 23S.3 RNA.

Figure 5A, lane 5, also shows that as the Mg²⁺ concentration was raised to 12.5 mM, the group I ribozyme activity of 23S.3 RNA was induced, as indicated by the appearance of characteristic products [linear and cyclized intron and 5′ exon molecules (39)]. A ligated exon molecule was not formed to any significant extent because there was no guanosine in the reaction. This result indicates that Mg²⁺ promoted 23S.3 RNA to fold into a native conformation.

Rates of Mg²⁺-dependent cleavage at site III were quantified by using a XhoI-truncated 23S.3 RNA, which terminates in P7.2 and has no group I ribozyme activity. The kinetic parameters of Mn²⁺-dependent cleavage at sites I and III were essentially unchanged in this shortened RNA (data not shown). The k_cat and K of Mg²⁺-dependent cleavage (at site III) were ∼10-fold lower and ∼16-fold higher, respectively, than those of Mn²⁺-promoted cleavage at site III (data not shown). These data indicate that the binding and catalysis by Mg²⁺ at site III are much weaker than Mn²⁺.

DISCUSSION

Extraordinary specificity at GAAA sites

The primary sequence of this 659-nt RNA contains 229 different tetramers (out of a possible 256), including 11 single-point-variants of GAAA (out of a possible 12), yet none were efficiently cleaved by Mn²⁺. In addition, cleavage of several other RNAs (650–1650 nt) with Mn²⁺ gave products consistent with the distribution of GAAA sequences (unpublished data). Indeed, we have shown that Mn²⁺ cleaved 23S.3 RNA at the major-site GAAAs ∼900-fold faster than other sequences.

Mn²⁺-binding at cleavage sites

The binding of Mn²⁺ to these sites is probably confined to the tetranucleotide and does not involve a large structural rearrangement. This is suggested by the lack of cooperativity in the rate–concentration curves. Hyperbolic kinetics were also observed in the cleavage of a pentanucleotide (GAAACp) by poly(U) and Mn²⁺ (25). It is unlikely that the 0.2 M KCl in the cleavage reactions suppressed a cooperative effect, because there was evidence of such an effect at site III by Mg²⁺, although not by Mn²⁺. It should be noted that the low Hill coefficient (n = ∼1.0), which we and others (25) have observed, does not necessary preclude models with multiple Mn²⁺ binding to GAAA (3,24). The numbers of RNA-bound metals estimated by activity assays usually are less than the numbers observed in crystallography (5,11,49–51). Moreover, two divalent metals can bind in a metal-binding pocket created by only a few nucleotides (11,12).

Cleavage efficiency and GAAA conformation

In an attempt to explain the differences of cleavage efficiencies at each site, the secondary structures of these sites were inferred from the established secondary structures for group I introns (34). As Figure 6 shows, sites I, IV, V and VI lie in the conserved J4/5, P6b, P7.2 and L9 regions (Fig. 6A), respectively, whereas sites II and III are in a non-conserved extension of P6b (Fig. 6B). The two major cleavage sites, I and III, are located within an internal loop and a four-way junction, respectively, structures which often bring about turns and twists of RNA helices (1,5) and could possibly mediate strong metal binding. This may account for the high cleavage efficiencies at these sites. Interestingly, the AAA trinucleotide at these sites apparently does not pair with UUU as in sites IV and V and in the original ribozyme (23,24,27). Presumably, the GAAA at these sites can adopt the active conformation without pairing with UUU. The weaker cleavages at sites IV and V are consistent with the notion that regular RNA helices usually are not ideal locations for specific metal binding (4). The weakest cleavage (at site VI ) is in a stable tetraloop (30,31), which may bind divalent metals only weakly (54). The extremely poor cleavage at this site may be due to the absence of available sites for metal coordination; the N-7 atoms of A2 and A3, which appear to be important for the GAAA/UUU ribozyme (25), should be involved in interactions within the tetraloop (30,55). It is not clear from the secondary structure of site II why cleavage at this site was weak. However, this site, which is located in a terminal loop, may be involved in forming a pseudoknot with ³¹⁸GGUUU³²¹ (Fig. 6B). It is also possible that some of the proposed secondary structure in the P6b extension (Fig. 6B) is incorrect, since it is based only on a thermodynamic prediction (unlike the conserved core in Fig. 6A).

Figure 6.

Secondary structure of the intron portion of 23S.3 RNA. (A) Secondary structure of the conserved core of the Cr.LSU intron. The nomenclature for structural elements is from Burke et al. (52); only the pertinent structures are labeled. The nucleotides are numbered from the 5′-end of 23S.3. The GAAA tetramers are boxed, shaded and labeled with Roman numerals (I, V, VI). (B) Proposed structure of the P6b extension. The structure was generated using the optimal/suboptimal algorithm of Jaeger et al. (53) incorporating the requirement for the phylogenetically conserved P6. This was the only structure obtained under optimal or suboptimal (±5%) conditions. The free energy of this structure is –21.8 kcal/mol. The start of the ORF, which terminates in J6/7, is indicated. The GAAA tetramers (II-III) are boxed and shaded.

While the secondary structures of the cleavage sites can be approximately correlated with cleavage efficiencies, the tertiary structure at these sites will determine the actual metal binding and cleavage. Thus, it would be informative to know how Mn²⁺ affects tertiary folding. Unfortunately, Fenton reagents, which have been applied to study the tertiary folding of Tt.LSU (50) and phage (56) group I ribozymes, cannot be used in the presence of Mn²⁺. Alternatively, we examined the structures of RNA where the conformation of GAAA is known. The crystal structure of the P4–P6 domain of the Tt.LSU group I intron has a GAAA at J4/5 and there are two Mg²⁺ in this region (12). One is very close (2.25 Å) to a phosphate oxygen (Op-G112) of the G and to the bases of the G and first A (4.18 Å to N7-G112, 4.15 Å to N7-A113 and 4.81 Å to N6-A113). The other Mg²⁺ could form an outer-sphere coordination with the phosphate of the first A (4.16 Å to Op-A113). The GAAA in P4–P6 is also in a non-canonical helix (12,57,58); however, the 2′-OH of the G could not be polarized by these two Mg²⁺, because of improper orientation and distance. In the hammerhead ribozyme, the conserved GAAA is in a non-canonical helix (8,9,51) (but different from the above) and Mg²⁺/Mn²⁺ binds to the phosphate of the first A (and possibly to N7 and pro-Rp oxygen of the second A) (59). In the model of the hairpin ribozyme, the highly conserved GAAA sequence is also placed in a non-canonical helix (60) (different from the above two) and a hydrated Mn²⁺ coordinates with the phosphate and nucleotide base of the G (61). Finally, although inner-sphere metal binding to a GAAA tetraloop has not been observed in crystal structures (30,31,55), a hydrated Mg²⁺ could be bound by forming hydrogen bonds with phosphates and the N7 of the G (54). Thus, divalent metals have been observed binding to GAAA in different geometries, although none of the above explains the metal binding and cleavage mechanism of the Mn²⁺-GAAA ribozyme. Possibly, they only represent ground state conformations. It would be interesting to know if the J4/5 GAAA in the Tt.LSU intron can be efficiently cleaved by Mn²⁺, as in 23S.3 RNA. If so, its conformation would possibly be closer to the conformation of the active Mn²⁺-GAAA ribozyme than that of the known GAAA tetraloops.

Further comparisons of specific cleavages at different sites in 23S.3 and the small Mn²⁺-GAAA ribozyme

The rates of cleavage at sites IV and V in 23S.3 RNA are close to that of the oligonucleotide Mn²⁺-GAAA ribozyme [k_cat ≈ 4–5 × 10^–3 min^–1 (24,25)], but are ∼2–4-fold slower than cleavage at sites I and III (Table 1). It should be noted that in the oligonucleotide Mn²⁺-GAAA ribozyme (23–25) and in cleavage sites IV and V of 23S.3, the GAAAs are all predicted to be mainly in helices. Such a consistency suggests that the method we have devised to obtain cleavage rates is reliable. Also, the higher cleavage rates at sites I and III further support the notion that, in a large RNA, the GAAA does not have to pair with UUU to form an active conformation.

On the other hand, there are substantial differences between the characteristics of the cleavages at sites I and III, although both are efficient. Most noticeably, Mg²⁺ alone promoted cleavage at site III, but not site I (or the other sites). This result disputes the notion that Mg²⁺ cannot promote cleavage of the GAAA ribozyme, as suggested from studies of a small RNA (23,24). Other differences include lower values for activation energy, K of the Mn²⁺ requirement, and slope of the pH–rate curve for site III. These characteristics are consistent with cleavage at site III being catalyzed by a more advanced ribozyme than at site I. Site III could be binding two metals (62) as suggested from the synergistic effect with Mg²⁺ and have a potentially more complex secondary structure; such metal binding could also lower the activation energy for cleavage.

Background versus specific cleavages

Since enzyme-catalyzed reactions usually have a lower activation energy than non-catalyzed reactions, it was somewhat surprising to find that the activation energy for specific cleavage at site I (and sites IV and V) was similar to that of background cleavage. However, reactions can also be accelerated by increasing the concentration of reactants and by properly orienting the reactants at the catalytic sites. It appears that by forming a metal-binding pocket at site I, the local concentration of Mn²⁺, and the probability of the right orientation, was raised. The weaker cleavages at sites IV and V may reflect the frequencies of conformational transitions from a regular helix to a non-canonical helix (as adopted by site I).

Metal binding in group I introns

In the present study, we have identified two major Mn²⁺-binding sites in 23S.3 RNA and, interestingly, one is located in the conserved J4/5 region where a GAAA sequence is found in many group I introns (34). In the crystal structure of the P4–P6 domain of Tt.LSU, a Mg²⁺ was located near the phosphate of the G in the J4/5 GAAA sequence (12), and in 23S.3 RNA, Mg²⁺ inhibited Mn²⁺-promoted cleavage at this site. Taken together, the data suggest that J4/5 (or part of J4/5) is in (or near) a divalent metal binding site in many group I introns. When the J4/5 GAAA of 23S.3 RNA was mutated to GACA, specific cleavage by Mn²⁺ was abolished and the Mg²⁺-requirement for in vitro self-splicing was raised dramatically (T.-C. Kuo, S. Holloway and D. Herrin, unpublished data; 63). Thus, the posited role of metal binding at J4/5 is structural, rather than catalytic. It is not known, however, if the conformation of J4/5 GAAA (or the whole 23S.3 RNA) is the same in Mn²⁺ as in Mg²⁺. This question arises because Mn²⁺ alone does not support the self-splicing of Cr.LSU, Cr.psbA2 or Cr.psbA3 (42) and at high concentrations Mn²⁺ inhibits the self-splicing of Tt.LSU (64), Cr.psbA1 and Cr.psbA4 introns (42). Also, whereas Mn²⁺ promoted specific cleavage in the conserved J8/7 region of a T4 phage group I intron (65), it did not in 23S.3. Thus, not all self-splicing group I introns fold the same in the presence of Mn²⁺. More structural studies are required to address the effects of Mn²⁺ on folding of 23S.3 RNA.

Evolution of the RNA World

The specificity of Mn²⁺-promoted GAAA cleavage suggests that in the ancient RNA World, metal-dependent site-specific cleavage could have been performed by ribo-organisms with genome sizes of only ∼100 nt (38). The likelihood of forming a simple metal-binding pocket (e.g. cleavage site I in this study) must be relatively high, in part because other divalent metals besides Mn²⁺ are known to promote site-specific cleavages in RNA of different sizes [e.g. 30-nt leadzyme (22), 76-nt tRNA (17,20), 377-nt RNase P (19) and rRNAs, 16S and 23S, (66)]. However, in environments replete with metals that can also promote non-specific RNA degradation, the population and complexity (i.e. genome size) of ribo-organisms would increase only if they were equipped with highly specific ribozymes. Therefore, under the appropriate selection pressure, more complex ribozymes (e.g. cleavage site III and leadzymes) would have evolved. The hairpin and hammerhead ribozymes could be regarded as two of nature’s final choices. The cleavage reactions catalyzed by these ribozymes have low activation energies [e.g. 13–22 and 18–19 kcal/mol for hammerhead and hairpin ribozymes, respectively, (67–70)], relatively high k_cat values [1.5 and 2.1 min^–1 for hammerhead and hairpin ribozymes, respectively (71,72)] and very low pH-dependency [e.g. hairpin ribozyme (70)]. More strikingly, they can use monovalent salts to promote catalysis (73). Finally, it should be noted that the fact that GAAA sequences do not have to be paired with UUU sequences to affect metal binding and cleavage increased the potential usefulness of this motif in the ancient and modern RNA Worlds.