Metal ion specificities for folding and cleavage activity in the Schistosoma hammerhead ribozyme

Abstract

The effects of various metal ions on cleavage activity and global folding have been studied in the extended Schistosoma hammerhead ribozyme. Fluorescence resonance energy transfer was used to probe global folding as a function of various monovalent and divalent metal ions in this ribozyme. The divalent metals ions Ca²⁺, Mg²⁺, Mn²⁺, and Sr²⁺ have a relatively small variation (less than sixfold) in their ability to globally fold the hammerhead ribozyme, which contrasts with the very large difference (>10,000-fold) in apparent rate constants for cleavage for these divalent metal ions in single-turnover kinetic experiments. There is still a very large range (>4600-fold) in the apparent rate constants for cleavage for these divalent metal ions measured in high salt (2 M NaCl) conditions where the ribozyme is globally folded. These results demonstrate that the identity of the divalent metal ion has little effect on global folding of the Schistosoma hammerhead ribozyme, whereas it has a very large effect on the cleavage kinetics. Mechanisms by which the identity of the divalent metal ion can have such a large effect on cleavage activity in the Schistosoma hammerhead ribozyme are discussed.

INTRODUCTION

The hammerhead ribozyme (HHRz) is one of the best-studied small ribozymes. It was first discovered in the satellite RNA of tobacco ringspot virus (Buzayan et al. 1986) and avocado sunblotch viroid (Hutchins et al. 1986) and has been implicated in rolling-circle replication of plant satellite viruses (Forster and Symons 1987; Haseloff and Gerlach 1988), where it performs both self-cleavage and self-ligation reactions. The HHRz is also found in repetitive satellite DNA in Schistosoma mansoni, newts, salamanders, and crickets (Zhang and Epstein 1996; Ferbeyre et al. 1998; Rojas et al. 2000) and more recently in the genome of Arabidopsis thaliana, where it is expressed in several tissues (Przybilski et al. 2005).

The HHRz consists of three helical regions surrounding a 13-nucleotide (nt) catalytic core, which performs a trans-esterification reaction where site-specific RNA cleavage produces a 2′,3′-cyclic phosphate and a 5′-hydroxyl group. This “minimal” ribozyme requires relatively high concentrations of divalent metal ions for efficient cleavage in vitro (i.e., ≥10 mM Mg²⁺) (Hertel et al. 1994). Inclusion of the naturally occurring tertiary interaction between loop motifs in stems I and II (Fig. 1) leads to much faster cleavage rates, and this interaction has been shown to be essential for efficient catalysis in vivo (De la Pena et al. 2003; Khvorova et al. 2003). In vitro kinetic studies on these “extended” HHRzs show that the loop–loop interaction can increase the observed cleavage rates under physiological conditions by over a factor of 100 (De la Pena et al. 2003; Khvorova et al. 2003; Canny et al. 2004).

FIGURE 1.

(A) The wild-type Schistosoma hammerhead construct used for the kinetics studies. (Black) The ribozyme strand; (gray) the substrate strand. The cleavable substrate strand has either a ³²P on the 5′-end or a fluorescein label on the 3′-end and is cleaved at C₁₇, as indicated by the arrow. (B) The noncleavable wild-type Schistosoma hammerhead ribozyme construct used for FRET studies. Cy3 was coupled to a modified U in loop II. Cy5 was coupled to the modified 3′-end of the substrate strand. The substrate strand has a 2′-deoxyribose on C₁₇ to prevent cleavage. (C) The UUCG control construct used for FRET studies. Cy3 was coupled to a modified U in loop II. The same noncleavable substrate strand was used for the UUCG control and the wild-type complex shown in B.

Comparison of the X-ray structures of the minimal HHRz (Pley et al. 1994; Scott et al. 1995) with the recent structure of an extended HHRz (Martick and Scott 2006) shows that inclusion of the loop–loop tertiary interaction leads to a substantial conformational rearrangement in the catalytic core. The conformation of the catalytic core in the extended hammerhead rationalizes much of the extensive biochemical data that were in serious disagreement with the structure of the minimal HHRz (McKay 1996; Blount and Uhlenbeck 2005; Nelson and Uhlenbeck 2008). Biophysical studies, such as FRET and EPR, have also been used to probe the mechanism for increased catalytic activity in the extended HHRz. These studies showed that the inclusion of the tertiary interaction leads to a lower metal ion requirement for folding of the extended than the minimal HHRz (Penedo et al. 2004; Heckman et al. 2005; Kim et al. 2005; Kisseleva et al. 2005). Together, the biochemical, biophysical, and structural data support a model where the loop–loop tertiary interaction increases the population of catalytically active species of the HHRz (Khvorova et al. 2003; Martick and Scott 2006; Nelson and Uhlenbeck 2006, 2007) and where divalent metal ions help to stabilize formation of the loop–loop tertiary interaction.

Previous studies on the minimal HHRz indicate that divalent metal ions are primarily involved in stabilizing the catalytically active structure and play only a small role in the chemistry of the cleavage/ligation reaction of the HHRz (Murray et al. 1998; Curtis and Bartel 2001; O'Rear et al. 2001). It was earlier hypothesized that all catalytic RNAs are metalloenzymes (Pyle 1993; Yarus 1993), where metal ions are required to perform chemistry in the catalytic reaction. The Tetrahymena group I intron is thought to be such a metalloenzyme since it has no activity in monovalent cations alone and phosphothioate rescue experiments have shown that there are multiple divalent metal ions within the core that are required for catalytic activity (for reviews, see Cech and Golden 1999; Hougland et al. 2006). Recent structural data support this model where the X-ray structure of the Azoarcus group I ribozyme (Stahley and Strobel 2005) has two Mg²⁺ ions in the active site, which are poised to carry out the classical two-metal ion catalytic mechanism for a phosphoryl transfer reaction (Steitz and Steitz 1993). Other ribozymes such as the hairpin ribozyme, the HDV ribozyme, and the ribosome have been shown to use RNA functional groups to act as general acids or general bases in catalysis (for recent reviews, see Fedor and Williamson 2005; Bevilacqua and Yajima 2006; Hoogstraten and Sumita 2007; Scott 2007; Walter 2007). Based on biochemical and structural studies, RNA functional groups have also been proposed to act in the catalytic mechanism of the HHRz (Han and Burke 2005; Martick and Scott 2006).

FRET (Penedo et al. 2004) and EPR (Kim et al. 2005) studies have shown that global folding occurs at much lower Mg²⁺ concentrations in the extended HHRz compared to the minimal HHRz. Previous biochemical studies have also shown that the identity of the divalent metal ion has a significant effect on the cleavage rate for the minimal HHRz (Dahm and Uhlenbeck 1991; Dahm et al. 1993) and a SELEX-derived extended HHRz (Roychowdhury-Saha and Burke 2006). To try to better understand the roles of metal ions in the extended HHRz, the observed rate constants for cleavage (k _obs) and global folding of the Schistosoma HHRz were measured for various divalent and monovalent metal ions. These results show that the identity of the divalent metal has a very large effect on the cleavage kinetics, whereas it has only a small effect on global folding of this HHRz.

RESULTS

Cleavage activity of the Schistosoma HHRz with different metal ions

Previous studies have shown that the inclusion of the loop–loop interaction in the HHRz leads to higher catalytic activity at lower Mg²⁺ concentrations (De la Pena et al. 2003; Khvorova et al. 2003; Canny et al. 2004) and that divalent metals are much more efficient at promoting catalysis than monovalent metal ions (Nelson et al. 2005). To test the effect of metal ion identity on catalysis, single turnover cleavage experiments were performed on the Schistosoma HHRz construct shown in Figure 1A in reaction buffer containing 1.0 mM of one of the following divalent ions: Ca²⁺, Cd²⁺, Co²⁺, Mg²⁺, Mn²⁺, Sr²⁺, or Zn²⁺. The fraction of cleaved substrate was monitored as a function of time and fit to Equation 1 yielding k _obs for cleavage under these conditions (0.1 M NaCl, at pH 7.0, 50 mM Tris, 25°C) (Table 1; Supplemental Fig. S1). These k _obs values vary by more than 10,000-fold (from 0.018 min⁻¹ for Sr²⁺ to ∼200 min⁻¹ for Mn²⁺) with the following trend: Mn²⁺ > Co²⁺ > Zn²⁺ ≈ Cd²⁺ > Mg²⁺ > Ca²⁺ > Sr²⁺. Single turnover cleavage kinetics were also performed on the Schistosoma HHRz in the presence of 600 mM of the monovalent ion Na⁺ or NH₄ ⁺ and yielded upper limits of k _obs < 0.02 min ⁻¹ (Table 1).

TABLE 1.

Metal dependence of k _obs for cleavage in the Schistosoma HHRz

As seen in Table 1, the fraction of substrate cleaved varies for the different metal ions from a plateau of 0.36 for Co²⁺ to 0.69 for Mg²⁺. The interpretation of the plateau is complicated because this plateau is not simply determined by the internal equilibrium for cleavage and ligation (k _cleave/k _ligate) but is also affected by the Schistosoma HHRz having a significant population of catalytically inactive species (Canny et al. 2007). These different metal ions may be differentially affecting the population of active ribozymes, which would lead to changes in the cleavage plateau. However, as previously discussed (Canny et al. 2007), the k _obs here is determined from the rapidly cleaving population, which is a minimum of 36% of the HHRz molecules for all metals (Table 1).

Global folding of the noncleavable Schistosoma HHRz with different metal ions

In addition to accelerating the rate of cleavage at a given Mg²⁺ concentration, the extended HHRz also globally folds at lower concentrations of Mg²⁺ when compared to a ribozyme lacking the tertiary interaction (Penedo et al. 2004; Kim et al. 2005). To examine if the large variation in k _obs for different divalent metal ions results from each metal differentially inducing global folding, FRET was used to monitor folding of the noncleavable wild-type hammerhead construct shown in Figure 1B. Figure 2A shows the FRET efficiency (E _FRET) for the Schistosoma HHRz (filled symbols) as a function of concentration of the following divalent metals: Ca²⁺, Mg²⁺, Mn²⁺, or Sr²⁺. The FRET results clearly show that these four divalent ions have similar efficiencies for globally folding the Schistosoma ribozyme, which is in contrast to the very large variation in k _obs for cleavage for these same metal ions (Table 1). FRET experiments were not performed with Co²⁺, Zn²⁺, and Cd²⁺, for which k _obs values were determined, due to interference with the fluorescence measurements from the colored Co²⁺ and limited solubility at higher concentrations of Zn²⁺ and Cd²⁺.

FIGURE 2.

(A) FRET efficiency (E_FRET) versus divalent metal ion concentration for folding of the noncleavable Schistosoma HHRz (filled symbols) wild-type and (open symbols) UUCG control constructs in (circles) Mn²⁺, (diamonds) Mg²⁺, (squares) Ca²⁺, or (triangles) Sr²⁺. (B) FRET efficiency versus monovalent metal ion concentration for folding of the noncleavable wild-type construct in (squares) Na⁺ or (circles) NH₄ ⁺. Error bars for replicate measurements are within the symbols.

FRET was also used to monitor global folding of the noncleavable UUCG control HHRz construct (Fig. 1C), which cannot form the loop–loop tertiary interaction. Like the wild-type ribozyme, an increase in E _FRET was observed with increasing metal ion concentration for Ca²⁺, Mg²⁺, Mn²⁺, and Sr²⁺ (open symbols in Fig. 2A), and all of these divalents again show comparable efficiencies for folding. However, higher concentrations of these metals are required to globally fold the control ribozyme compared to the wild-type ribozyme.

To more quantitatively compare folding with different metal ions, the FRET data were fit to a two-state folding equilibrium assuming fully cooperative metal ion binding (Equation 2 in Materials and Methods). As seen in Table 2, all the divalent metal ions have very similar values for both M_1/2, the apparent dissociation constant for the metal ion, and n, the Hill coefficient for metal-binding induced folding of the wild-type HHRz (M_1/2 ranges from 0.87 to 2.0 mM, and n ranges from 0.41 to 0.65). Furthermore, the divalent ions show a similar change in FRET efficiency (ΔE _FRET ranging from 0.22 to 0.25) upon folding, indicating that the global fold of the HHRz is comparable in all of these metals (Supplemental Table S1). Together these results show that the identity of the divalent metal ion has little influence on global folding of the HHRz.

TABLE 2.

Metal dependence of global folding of the Schistosoma hammerhead ribozyme

The FRET-based folding data for the UUCG control ribozyme construct as a function of Ca²⁺, Mg²⁺, Mn²⁺, or Sr²⁺ (Fig. 2A) were also fit to the two-state folding equilibrium (Eq. 2) and again showed small ranges of M_1/2 (5.2–12 mM) (Table 2), n (0.7–1.0) (Table 2), and ΔE _FRET (0.13–0.17) (Supplemental Table S1). The M_1/2 values for the UUCG control are fivefold to sixfold higher and the Hill coefficients are larger for the UUCG than the wild-type HHRz. The average ΔE _FRET for the UUCG control of 0.16 is lower than for the wild-type construct (ΔE _FRET ∼0.25), which is not unexpected given that the position of the donor fluorophore, Cy3, is altered in the UUCG loop (Fig. 1C).

To assess how the loop–loop interaction affects global folding in the presence of only monovalent cations, FRET studies were performed for both the wild-type (Fig. 2B) and UUCG control (data not shown) noncleavable HHRz constructs as a function of Na⁺ or NH₄ ⁺ concentration. The FRET data for monovalent ions were fit to Equation 2, yielding M_1/2 values for the wild-type HHRz in Na⁺ and NH₄ ⁺ of ∼170 mM and ∼245 mM with Hill coefficients of 1.3 and 1.1, respectively (Table 2). The UUCG control HHRz showed higher M_1/2 values for both Na⁺ and NH₄ ⁺ of >500 mM and >800 mM and Hill coefficients of 1.0 and 1.1, respectively (Table 2). Thus, as observed for the divalent metals, the HHRz containing the loop–loop interaction requires considerably lower concentrations of monovalent metal ions to achieve its global fold.

Inactive populations in the Schistosoma HHRz

In addition to the very large changes in k _obs for the wild-type HHRz, the divalent metal ions show a variation in the fraction of substrate cleaved (ranging from 0.36 to 0.69) (Table 1). As previously reported, the Schistosoma HHRz has an inactive population that slowly converts to an active population (Canny et al. 2007). To determine if interconversion of inactive and active hammerhead species involves a change in global folding, the FRET of the noncleavable Schistosoma HHRz system was monitored over time. As seen in Figure 3A, the E _FRET values were essentially unchanged over 8 h in 1 mM Mg²⁺ (<3% change). On the other hand, kinetic experiments with a cleavable substrate show a substantial increase in the fraction of substrate cleaved (from 0.30 to 0.65), yielding an apparent k _obs = 0.003 min⁻¹ over the same 8-h time period (Fig. 3B). The FRET data here only monitor global folding of the HHRz, i.e., stems I and II coming together, and may not be sensitive to local conformational changes. Thus, an inactive-to-active transition involving a local conformational change in the loop–loop tertiary interaction or in the catalytic core would not generally be detected by this assay.

FIGURE 3.

(A) Plot of FRET efficiency of the noncleavable wild-type Schistosoma HHRz FRET construct as a function of time in 1 mM Mg²⁺. (B) Plot of fraction cleaved substrate as a function of time for the wild-type Schistosoma HHRz (chemically synthesized Cy3-labeled ribozyme strand and transcribed substrate strand) in 1 mM Mg²⁺. The line represents the fit to Equation 1 and gives a k _obs of 0.003 min⁻¹.

To rule out multiple turnover ribozyme reactions being responsible for the apparent inactive-to-active conversion in the single-turnover kinetics experiments, pulse-chase nondenaturing gel experiments were performed to monitor the rate of substrate dissociation (Fedor and Uhlenbeck 1992). These experiments showed no evidence for dissociation of the noncleavable substrate strand from the transcribed or chemically synthesized ribozymes (see Supplemental Fig. S2). Thus, the slow increase in the fraction of substrate cleaved in Figure 3B is not due to substrate dissociating from an inactive ribozyme and binding to an active ribozyme in these single-turnover kinetic experiments. In addition, the ribozyme-noncleavable substrate complexes for both transcribed and chemically synthesized ribozymes show a single band in the nondenaturing gel, so there is no evidence for conformational heterogeneity in these complexes (Supplemental Fig. S2).

Cleavage activity with different divalent metal ions in a globally folded HHRz

Although divalent metal ions are normally required for efficient folding of RNAs, RNAs sometimes fold in high concentrations of monovalent ions (Draper 2004). To try to separate the effects of the identity of the divalent metal ions on global folding from the effects of the identity of the metal ion on cleavage activity, single-turnover cleavage experiments were performed in a reaction buffer containing 2 M NaCl and 1.0 mM Ca²⁺, Mg²⁺, or Mn²⁺ (Table 1). The FRET data in Figure 2B show that the wild-type Schistosoma HHRz is essentially fully folded in the presence of 2 M NaCl, so any variation in k _obs is not due to global folding. The k _obs in 2 M NaCl, 1.0 mM Mn²⁺ was too fast to measure manually at pH 7.0, so kinetic experiments were repeated with 0.15 mM Mn²⁺ at pH 6.0 yielding a k _obs of 3.5 min⁻¹. A linear dependence of k _obs on Mn²⁺ concentration and log-linear dependence with pH was observed over these ranges (data not shown), yielding a calculated k _obs of 230 min⁻¹ for Mn²⁺ at 1 mM (pH 7.0). Thus, the k _obs for Ca²⁺, Mg²⁺, and Mn²⁺ still vary by >4600-fold even in the globally folded HHRz (2 M NaCl) (Table 1). Furthermore, the fraction of substrate cleaved in a given metal is unchanged in both the 0.1 M and 2 M NaCl backgrounds (Table 1), indicating that higher concentrations of monovalent ions do not affect the populations of rapidly cleaving ribozymes.

DISCUSSION

Metal ions are important for catalytic activity of the HHRz (Dahm and Uhlenbeck 1991; Murray et al. 1998). Previous studies on a minimal hammerhead system required higher than physiological divalent metal ion concentrations for efficient catalysis (Hertel et al. 1994). More recently, it was shown that formation of a tertiary interaction distant from the catalytic core of the HHRz allows for efficient cleavage and ligation at physiological Mg²⁺ concentrations (De la Pena et al. 2003; Khvorova et al. 2003). Although the size and specific sequence of this tertiary interaction varies, all of the natural hammerhead systems have the potential to form a tertiary interaction between loop motifs on stems I and II (Khvorova et al. 2003; Przybilski et al. 2005). One model for how inclusion of these additional motifs generates faster kinetics is that there is a rapid equilibrium between highly active and low or inactive species in the HHRz and that formation of the tertiary interaction leads to a higher population of catalytically active molecules (Khvorova et al. 2003; Martick and Scott 2006; Nelson and Uhlenbeck 2006). The effects of the identity of the metal ions on both the cleavage kinetics and the global folding of the extended Schistosoma HHRz are studied here.

The identity of the metal ion has a very large (>10,000-fold) effect on the k _obs for cleavage in the Schistosoma HHRz, where the divalent ions follow the trend of Mn²⁺ > Co²⁺ > Zn²⁺ ≈ Cd²⁺ > Mg²⁺ > Ca²⁺ > Sr²⁺ (at 1 mM metal ion, 50 mM Tris, 0.1 M NaCl, pH 7, and 25°C) (see Table 1). Previous studies on the Schistosoma HHRz showed an ∼100-fold difference in k _obs for Mg²⁺ and Mn²⁺ (Osborne et al. 2005), similar to the results here. Except for Cd²⁺, this trend is the same as what was previously observed for a SELEX-derived rapidly cleaving HHRz with a stable hairpin loop-internal bulge tertiary interaction, which had k _obs of Mn²⁺ > Cd²⁺ > Co²⁺ > Zn²⁺ > Mg²⁺ > Ca²⁺ > Sr²⁺ (at 1 mM metal ion, 50 mM Tris, pH 7.4, and 37°C) (Roychowdhury-Saha and Burke 2006). A somewhat different trend for metal ions was previously observed for a minimal HHRz with k _obs of Mn²⁺ ≈ Cd²⁺ > Mg²⁺ > Co²⁺ ≈ Zn²⁺ > Ca²⁺ > Sr²⁺ (at 2 mM metal ion, 50 mM Tris, 0.5 mM spermine, pH 7.5, and 25°C) (Dahm et al. 1993). Interestingly, the identity of the divalent metal ion has a much larger effect on the cleavage rates in the extended HHRz compared with the minimal HHRz. Only a relatively small effect on k _obs (∼25-fold change) was observed for the various divalent metals in the minimal HHRz (Dahm et al. 1993), whereas a much larger effect in k _obs was observed for the same divalent metals in the Schistosoma HHRz (>10,000-fold change) and the tertiary-stabilized SELEX-derived HHRz (>5000-fold change) (Roychowdhury-Saha and Burke 2006). Thus, although these three hammerhead systems all show the fastest k _obs with Mn²⁺, intermediate k _obs with Mg²⁺, and slower cleavage with Ca²⁺ and Sr²⁺, inclusion of the loop–loop tertiary interaction leads to a much higher sensitivity of k _obs to the identity of the divalent metal ion.

FRET was used here to monitor global folding of the HHRz to examine if the large variations in k _obs with different divalent ions arise from different efficiencies of global folding. In contrast to the cleavage kinetics, the FRET data in Figure 2A unambiguously show that the identity of the divalent ion has little effect on the global folding of the Schistosoma HHRz. This is confirmed by fitting the data to Equation 2, which shows that there are small ranges for both the apparent dissociation constants for metal binding, M_1/2, and the Hill coefficients for the different divalent ions (Table 2). As previously discussed (Misra and Draper 2002), the two-state model used to derive Equation 2 involves multiple assumptions for metal-induced folding of RNA, and these assumptions may not be valid for folding of the HHRz. Therefore, although the data in Table 2 can be used to compare the overall effects that different metal ions have on folding, they cannot be used to probe the mechanism of folding this RNA. Independent of whether the folding of the HHRz is well described by the two-state model (or follows a more complex pathway), the raw FRET data in Figure 2A clearly demonstrate that global folding of the Schistosoma HHRz is independent of the identity of the divalent ion. It is important to note that this FRET assay only probes the proximity of stems I and II and does not give information on formation of the specific tertiary interaction observed in the X-ray structure of the Schistosoma HHRz (Martick and Scott 2006). Thus, the divalent metals could be differentially stabilizing a local conformational change required for efficient cleavage or could be stabilizing slightly different local structures, which then result in the large variation in k _obs. The UUCG control HHRz, which lacks the tertiary interaction, requires higher concentrations of divalent metal ions to globally fold than the wild-type HHRz (approximately sixfold higher) (Table 2), but folding is also independent of the identity of the divalent metal ion (Fig. 2A).

Previous studies showed that the minimal and extended constructs of the sTRSV HHRz have similar k _obs in 4 M Li⁺, from which it was proposed that the loop–loop interaction may not be able to form in the presence of only monovalent cations (Nelson et al. 2005). Cleavage was measured here for the Schistosoma HHRz in the presence of 600 mM Na⁺ or NH₄ ⁺ and yielded an upper limit on k _obs of <0.02 min⁻¹. The FRET studies show evidence for global folding of the Schistosoma HHRz with monovalent ions, consistent with previous folding studies on this ribozyme (Penedo et al. 2004). As seen in Table 2, the wild-type Schistosoma HHRz globally folds at threefold lower metal ion concentration than the UUCG control for both Na⁺ and NH₄ ⁺. However, the FRET experiments cannot determine whether the molecules are forming the correct tertiary interaction required for cleavage or whether the monovalent ions simply reduce electrostatic repulsion between stems I and II.

Various biochemical and structural studies have provided strong evidence for metal ion-binding site(s) within the core of the hammerhead ribozyme (for reviews, see Feig and Uhlenbeck 1999; Hoogstraten and Sumita 2007). Cd²⁺ rescue experiments led to the prediction that in the transition state a single metal ion coordinates the phosphate of A9 and N7 of G10.1 and the phosphate at the cleavage site in the transition state (Wang et al. 1999). This prediction was essentially validated in the X-ray structure of the extended HHRz (Martick and Scott 2006), where the phosphate oxygens at A9 and the cleavage site C1.1 are separated by 4.3 Å and thus are poised to bind a single metal (Wang et al. 1999). A subsequent X-ray structure of the Schistosoma HHRz (Martick et al. 2008) showed a Mn²⁺ ion with inner sphere coordination to both the phosphate oxygen of A9 and the N7 of G10.1. Although no bridging metal ion was observed between the A9 and C1.1 phosphates, these phosphates are still separated by only 4.3 Å. EPR studies also showed evidence for a single Mn²⁺ ion-binding site with ∼10 μM affinity in the minimal HHRz at 1 M NaCl (Horton et al. 1998). A follow-up study identified this as the Mn²⁺ coordinated to the A9 phosphate and the N7 of G10.1 (Vogt et al. 2006). Furthermore, an EPR study on an extended HHRz (Kisseleva et al. 2005) showed a single Mn²⁺ binding site with even higher affinity, where the binding affinity for the Mn²⁺ has increased by >100-fold compared to the minimal HHRz (K _d < 10 nM for Mn²⁺ binding in 0.1 M NaCl). These results demonstrate that the loop–loop interaction of the extended HHRz dramatically increases the affinity for binding of a Mn²⁺ ion. Achieving such high affinity (K _d < 10 nM) likely requires a highly optimized metal-binding site in the RNA, and such a site could also show high metal ion selectivity, potentially resulting in the large variation in k _obs measured here for different metal ions.

To try to understand what property of the divalent metal ion gives rise to the changes in the k _obs for the Schistosoma HHRz, Figure 4 shows plots of log(k _obs) versus pK _a of the hydrated metal ion, ionic radius, or absolute hardness of the metal ion. The pK _a plot shows the best correlation for all the metals. There is a good linear correlation for pK _a versus log(k _obs) for the Schistosoma HHRz (R = 0.93) (Fig. 4A) with a slope of approximately −1. A similar linear correlation of log(k _obs) and pK _a was previously observed for the SELEX-derived tertiary-stabilized HHRz (Roychowdhury-Saha and Burke 2006). Although this trend could arise from a metal hydroxide acting directly as the general base for the 2′-OH at the cleavage site for the HHRz (Dahm et al. 1993), there are various lines of evidence that this is not the mechanism of action of metal ions in the HHRz. Multiple studies have shown that the minimal HHRz efficiently cleaves in high concentrations of monovalent metal ions (Murray et al. 1998; Curtis and Bartel 2001; O'Rear et al. 2001; Takagi et al. 2004). The monovalent ions have very high pK _a values, strongly suggesting that the metal hydroxide does not directly act as the general base for the 2′-OH in the cleavage reaction (Murray et al. 1998; Curtis and Bartel 2001). One possibility for how the pK _a of the divalent metal (or related parameter such as ionic radius) could effect the k _obs is if there is direct inner-sphere metal binding (Pontius et al. 1997) to the 2′-O of G8, where, as previously proposed, this hydroxyl then acts as a general acid in the catalytic mechanism of the HHRz (Martick and Scott 2006).

FIGURE 4.

(A) Plot of the pK _a of the hydrated metal ion as a function of the log(k _obs) for the Schistosoma HHRz in 1 mM metal ion, 50 mM Tris, 0.1 M NaCl (pH 7.0). The line represents the best fit between log(k _obs) and pK _a for the Schistosoma HHRz (slope=−0.96, R = 0.93). (B) Plot of the ionic radius of the metal ion as a function of log(k _obs) for the Schistosoma HHRz. (C) Plot of the absolute hardness of the metal ion as a function of log(k _obs) for the Schistosoma HHRz. The values for pK _a, ionic radius, and absolute hardness are from Feig and Uhlenbeck (1999).

There are several examples where the ionic radius of a metal ion has been shown to correlate with the binding affinity of metals to an RNA (Bukhman and Draper 1997; Feig et al. 1999; Curtis and Bartel 2001), so the log(k _obs) for the Schistosoma HHRz was plotted as function of ionic radius (Fig. 4B). Except for Mg²⁺, there is a good linear correlation of log(k _obs) with ionic radii (R = 0.92). Mg²⁺ fits very poorly in this trend, where it has much slower kinetics than predicted from its ionic radius. A log-linear relationship between k _obs and the ionic radius of Li⁺, Na⁺, K⁺, Rb⁺, and Cs⁺ was observed in the minimal HHRz (Curtis and Bartel 2001) from which it was interpreted that these metals differentially stabilize the catalytically active structure. Furthermore, a specific metal-binding site on G5 in the minimal HHRz also showed a log-linear correlation between binding affinity and the ionic radius for a series of divalent metals (Feig et al. 1999). One explanation for why Mg²⁺ fits so poorly in Figure 4B is that although it has an ionic radius similar to some of the transition metals, Mg²⁺ is very “hard” compared to these metals. This is illustrated in Figure 4C, which shows that for the Schistosoma HHRz there is a good linear correlation (again except for Mg²⁺) between log(k _obs) and the absolute hardness (Feig and Uhlenbeck 1999) of the divalent metal (R = 0.98). The N7 of purine preferentially coordinates soft versus hard metals, so the data in Figure 4, B and C, can be qualitatively rationalized if there is a critical metal-binding site that coordinates to a purine N7 in the extended HHRz. Therefore, the various divalent metal ions could differentially stabilize an active (or inactive) conformation within the core, leading to the large differences in the k _obs.

To test whether the large differences in k _obs in the Schistosoma HHRz could be eliminated by using monovalent ions to replace a structurally important metal-binding site, the k _obs was measured in 2 M NaCl and 1 mM Ca²⁺, Mg²⁺, or Mn²⁺ ions. As seen in Figure 2B, 2M NaCl results in global folding of the HHRz, and this high concentration of monovalent ions can also compete with binding of other divalent metals ions (Draper 2004). Paralleling what was observed in 0.1 M NaCl, there is still a very large variation in k _obs for the divalent metals (>4600-fold) (Table 2). Such large differences for these divalent metals contrast with what was previously observed for two structurally important divalent metal-binding sites in the P4–P6 domain of the Tetrahymena group I intron (Travers et al. 2007). In that study, the metals Ca²⁺, Mg²⁺, and Mn²⁺ showed only a low level of specificity for two well-defined divalent metal ion-binding sites in 2M NaCl (less than sixfold variation in M_1/2 for these metals). Thus the catalytic activity of the extended HHRz is much more sensitive to a specific property of the divalent metal ion (e.g., pK _a, ionic radius or coordination geometry) than metal binding in this P4–P6 domain.

There are several examples of metal-binding sites that also show high selectively for the identity of the divalent metal ion in a pre-folded RNA. For example, a metal-binding site at G5 in the minimal HHRz shows moderate selectivity (∼250-fold) for different divalent metal ions in a background of 10 mM Mg²⁺ (Feig et al. 1999). Another example is a 58-nt fragment of the rRNA, where tertiary folding was measured as a function of Mg²⁺, Ca²⁺, Sr²⁺, and Ba²⁺ in 1.6 M NH₄Cl (Bukhman and Draper 1997). These metals showed a 60-fold variation in their apparent binding affinities, and there was a log-linear relationship between binding affinities and ionic radius of the metal. There are also examples of larger ribozymes where certain divalent ions are required for catalytic activity. The Tetrahymena group I ribozyme has a strict requirement for the divalents Mn²⁺ or Mg²⁺ (Grosshans and Cech 1989). The Escherichia coli RNase P RNA is active with Mg²⁺ and Mn²⁺ in 1 M NaCl and still shows substantial activity with Ca²⁺ (Smith et al. 1992). However, with the exception of the Tetrahymena group I ribozyme, the Schistosoma HHRz has a higher level of divalent metal ion selectively (>4600-fold for Ca²⁺ compared to Mn²⁺ in 2 M NaCl) than exhibited by other catalytic RNAs.

The results here demonstrate that the changes in k _obs in the Schistosoma HHRz with different metal ions do not arise from differential effects on global folding, defined here as stems I and II coming together. Potential explanations for the variation in k _obs with metal ion identity include: (1) there is a structurally important metal-binding site required for catalysis that is very sensitive to the identity (i.e., properties) of the divalent ion; (2) there is a rapid equilibrium between a catalytically active and inactive species for the HHRz (Khvorova et al. 2003; Nelson and Uhlenbeck 2007) and the various divalent metal ions differentially affect the populations and/or conformations of these species; or (3) there is a specific, selective metal-binding site that helps to stabilize the negative charge in the transition state. Previous studies on the minimal hammerhead comparing kinetics of mutants/modifications in 10 mM Mg²⁺ and 4 M Li⁺ concluded that divalent metal ions played only a small direct chemical role in catalysis (O'Rear et al. 2001). However, divalent metal ions may play a larger direct chemical role for divalents in the extended than the minimal HHRz. One candidate for a structural (or catalytic) metal is the Mn²⁺ ion that coordinates to the nonbridging R_p oxygen of A9 and the N7 of G10.1 in the X-ray structures of both the minimal and extended HHRzs (Pley et al. 1994; Martick et al. 2008). This metal is close to the active site in the X-ray structure of the Schistosoma HHRz (Martick et al. 2008) and could help to stabilize the negative charge in the transition state, so differential affinities, or a slightly different coordination/conformation, of the metal ions in this site could alter the catalytic activity.

MATERIALS AND METHODS

Sample preparation

The ribozyme and substrate RNA oligonucleotides used in the cleavage studies were prepared by in vitro transcription with T7 RNA polymerase from synthetic DNA templates (Integrated DNA Technologies) and purified using 20% denaturing polyacrylamide gel electrophoresis as described (Milligan and Uhlenbeck 1989). To remove trace amounts of divalent metal ions, the RNAs were heated to 90°C in 50 mM Tris, 100 mM NaCl, and 10 mM EDTA (pH 7) and cooled to room temperature. The concentration of EDTA was then reduced by multiple rounds of buffer exchange using a 3K Microcon centrifugal device (Millipore) into reaction buffer (50 mM Tris, 100 mM NaCl, 0.1 mM EDTA at pH 7.0). Prior to use, the reaction buffer was run over a chelating resin (Chelex; Sigma) to remove any trace divalent metal ions. The substrate strand (Fig. 1A) was labeled with either fluorescein at the 3′-end (Qin and Pyle 1999) or with a ³²P phosphate group at the 5′-end using T4 polynucleotide kinase and buffer exchanged as described above. To prevent oxidation, the metal ion solutions were prepared by diluting stock solutions into the reaction buffer, where the stock solutions were kept at −20°C and low pH until ready for use. The stock solutions of metal ions were examined for contaminating metals by inductively coupled plasma optical emission spectroscopy. The RNAs used in FRET studies (Fig. 1B,C) were chemically synthesized (Dharmacon) with a 5-amino-allyl modification on the C5 of an internal U in the ribozyme strand and a 2′-deoxyribose at the cleavage site and a six-atom amino linker on the 3′-end of the substrate strand. The Cy3 (donor) or Cy5 (acceptor) N-succinimidyl ester fluorophores (Amersham Biosciences) were coupled to the RNA and purified as previously described (Downey et al. 2006) and then exchanged into reaction buffer as described above.

Single-turnover cleavage experiments on the Schistosoma HHRz

The observed rate constants for cleavage, k _obs, for the Schistosoma HHRz were measured under single turnover conditions in the presence of various metal ions, using the cleavable construct in Figure 1A. A 10:1 ratio of ribozyme strand to either fluorescein-labeled or ³²P-labeled substrate strand was annealed in the reaction buffer by heating for 2 min to 70°C and slowly cooling to 25°C in a thermocycler (Eppendorf). The final concentration of ribozyme strand was 0.5–1 μM. All kinetic experiments were performed at 25°C.

In manual experiments, the cleavage reaction was initiated by adding 90 μL of the annealed ribozyme-substrate complex to 10 μL of the appropriate concentration of metal ion in reaction buffer. Ten-microliter aliquots were removed at various time points and quenched by addition of 10 μL of 80% formamide and 25 mM EDTA. For metals that had too rapid cleavage kinetics to be measured by manual methods (Mn²⁺, Co²⁺, Zn²⁺, Cd²⁺, and Mg²⁺), a Kin-Tek quench-flow model RQF-3 instrument was used. In these experiments, 15 μL of annealed hammerhead complex was rapidly mixed (2 msec dead-time) with 15 μL of two times the appropriate concentration of metal ion in reaction buffer followed by a rapid quench with 90 μL of solution containing 80% formamide and 25 mM EDTA. To separate cleaved and uncleaved substrate strands, the quenched samples were boiled for 1 min at 90°C and loaded on a 20% denaturing polyacrylamide gel. The gels were scanned for either fluorescence or ³²P exposure using a Typhoon Imager (Amersham Biosciences), and the fraction of cleaved substrate was quantified using ImageQuant software (Molecular Dynamics). These data were then fit to Equation 1 using the programs Excel (Microsoft) or KaleidaGraph (Synergy):

where F _cl(t) is the fraction cleaved at time t, f _i is the initial fraction cleaved at t = 0, Δf is the change in the fraction cleaved between t = 0 and t = ∞, and k _obs is the observed rate constant for cleavage. The k _obs represents an approach to equilibrium and therefore is the sum of the cleavage and ligation rate constants. The parameters f _i, Δf, and k _obs were varied to give the best fit to the experimental data.

Substrate dissociation assay

Pulse-chase experiments were used to test for substrate dissociation from the transcribed and chemically synthesized ribozyme strands (Fedor and Uhlenbeck 1992). The 0.5 μM ribozyme strand and trace (<0.05 μM) ³²P-labeled noncleavable substrate strand were annealed as described above. The annealed hammerheads were incubated in reaction buffer containing 1 mM Mg²⁺ for 30 min at room temperature. Then a 75 μM (2 μL) unlabeled noncleavable substrate strand was added to 2-μL aliquots of annealed HHRz complex to initiate the chase reactions, which were then incubated for various times before loading on the gel. One microliter of 30% glucose was added to each sample as loading buffer, including 2-μL aliquots of chase-free complex and substrate-only solutions, and the samples were loaded on a 12% nondenaturing gel containing 1 mM Mg²⁺. The radioactive bands were identified using PhosphorImager analysis as described above.

Design of Schistosoma HHRz constructs for probing global folding by FRET

FRET was used here to monitor global folding and formation of the loop–loop tertiary interaction in the Schistosoma hammerhead ribozyme–substrate complex. Various constructs with a fluorophore-modified U residue in stem II or in loop II were tested for cleavage activity. Single-turnover cleavage kinetics in 1 mM Mg²⁺ were measured on these modified ribozymes, and only the construct with the Cy3 fluorophore on the U in loop II (Fig. 1B) showed the same k _obs for cleavage as the unmodified, unlabeled construct (data not shown). Modeling of this complex using the X-ray structure of the Schistosoma HHRz (Martick and Scott 2006) indicates that the donor Cy3-fluorophore in loop II and the acceptor Cy5-fluorophore on the 3′-end of the substrate (stem I) are separated by ∼30 Å in the docked complex, which is well below the Förster radius for this FRET pair (Sabanayagam et al. 2005). To separate folding from the effects of catalytic activity, the FRET experiments were performed using a noncleavable substrate that has a 2′-deoxyribose at the cleavage site (see Fig. 1B).

A control construct where loop II was substituted with a UUCG tetraloop was used to test the importance of the loop–loop interaction on the global folding of the extended HHRz. For the FRET studies, a Cy3 fluorophore was coupled to a C5-modified U residue in the second position in the UUCG tetraloop (Fig. 1C). This control ribozyme strand was annealed to the same noncleavable substrate that was used with the wild-type ribozyme strand above.

Although the measured k _obs for cleavage was the same for the chemically synthesized Cy3-modified ribozyme and the in vitro transcribed ribozyme, these two ribozymes showed differences in the fraction of cleaved substrate in single-turnover cleavage experiments. In 1 mM Mg²⁺, the constructs have plateaus of 0.40 and 0.70 fraction of cleaved substrate after 10 min for the chemically synthesized and transcribed ribozyme strands, respectively; and plateaus of 0.60 and 0.85 fraction of cleaved substrate after 24 h (data not shown). A chemically synthesized ribozyme strand without the C5-modified U also showed the lower fraction cleaved, thus the chemically synthesized ribozymes have a higher inactive (or slowly cleaving) population than their transcribed counterpart. RNAs often form inactive populations, and previous studies of the Schistosoma HHRz showed an ∼30% “inactive” population for transcribed RNAs, where this “inactive” population converts slowly to an active species (Canny et al. 2007). Thus, the hammerhead constructs used here for the folding experiments have a somewhat higher population of catalytically “inactive” species than the constructs employed in the kinetic studies; however, both constructs show the same k _obs for cleavage for the rapidly cleaving population.

Metal-induced folding of the Schistosoma HHRz as probed by FRET

Stock solutions of Schistosoma HHRz for ensemble FRET measurements were prepared by annealing equal volumes of 1 μM Cy3-labeled ribozyme strand with 1.1 μM Cy5-labeled substrate strand (Fig. 1B,C) in reaction buffer by heating for 2 min to 70°C and slowly cooling to 25°C in a thermocycler. These stock solutions were diluted to ∼25 nM in 0.5 mL, and fluorescence measurements were performed on a Photon Technology International spectrofluorometer at 25°C. The donor fluorophore was excited at 500 nm, and emission was measured at both 560 nm (donor emission) and 660 nm (acceptor emission) with slit widths for the emission and excitation monochromators of 4 mm and 8 mm, respectively. A total of 20 pts/sec for 20 sec at each emission wavelength was recorded and averaged for each metal ion concentration. The FRET efficiency, E _FRET, was calculated as I _A /(I _A + I _D), where I _A is the average acceptor intensity and I _D is the average donor intensity.

Metal ion titrations were performed by adding 2–10 μL of a concentrated (25 or 250 mM) metal ion solution in reaction buffer to a 0.5 mL sample of the hammerhead complex. The E _FRET values were determined for each metal ion concentration after correcting for dilution effects and were fit to Equation 2, which assumes a two-state metal-induced folding model consisting of an equilibrium between unfolded and folded states:

where E _i is the FRET efficiency in the reaction buffer (no added metals ions), ΔE _FRET is the overall change in FRET efficiency induced between no metal ion added and infinite metal ion concentration, M _1/2 is the apparent dissociation constant for the metal ion, n is the Hill coefficient, [M^z+] is the metal ion concentration, and z is the charge of the metal ion used in the titration. The parameters E _i, ΔE _FRET, M _1/2, and n were varied to give the best fit to the experimental data.

To determine if metal ions lead to the quenching of the fluorophores independent of FRET, experiments were performed where only the Cy3 (donor) or Cy5 (acceptor) dyes were incorporated into the hammerhead complex, and the fluorescence was monitored as a function of metal ion concentration (data not shown). For Mg²⁺, Ca²⁺, and Sr²⁺, there was no significant change in the fluorescence intensity of the donor-only or acceptor-only constructs as a function of metal ion concentration. However, at higher concentrations of Mn²⁺ (>10 mM), for both the donor-only and acceptor-only constructs, there was a linear reduction in the fluorescence intensity with Mn²⁺ concentration, with 10% quenching of original fluorescence at the highest concentration (∼60 mM). Therefore, when measuring the fluorescence of the HHRz in the presence of Mn²⁺, a correction was applied based on the linear relationship between fluorophore quenching and Mn²⁺ concentration.

SUPPLEMENTAL DATA

Supplemental material can be found at http://www.rnajournal.org.