Characterizing multiple metal ion binding sites within a ribozyme by cadmium-induced EPR silencing

Abstract

In ribozyme catalysis, metal ions are generally known to make structural and∕or mechanistic contributions. The catalytic activity of a previously described Diels-Alderase ribozyme was found to depend on the concentration of divalent metal ions, and crystallographic data revealed multiple binding sites. Here, we elucidate the interactions of this ribozyme with divalent metal ions in solution using electron paramagnetic resonance (EPR) spectroscopy. Manganese ion titrations revealed five high-affinity Mn²⁺ binding sites with an upper K_d of 0.6±0.2 μM. In order to characterize each binding site individually, EPR-silent Cd²⁺ ions were used to saturate the other binding sites. This cadmium-induced EPR silencing showed that the Mn²⁺ binding sites possess different affinities. In addition, these binding sites could be assigned to three different types, including innersphere, outersphere, and a Mn²⁺ dimer. Based on simulations, the Mn²⁺-Mn²⁺ distance within the dimer was found to be ∼6 Å, which is in good agreement with crystallographic data. The EPR-spectroscopic characterization reveals no structural changes upon addition of a Diels-Alder product, supporting the concept of a preorganized catalytic pocket in the Diels-Alder ribozyme and the structural role of these ions.

Metal ions are of critical importance for the structure and function of biological macromolecules. Elucidating the function of individual ions is a challenging endeavor and particularly complicated in the case of multiple bound ions of one sort. First, it is often difficult to establish the precise number of bound ions, as their affinities may vary over several orders of magnitude. Second, spectroscopic techniques yield superpositions of all binding sites. Third, attempts to separate the individual binding sites by changing the experimental conditions may influence all sites, and therefore perturb the systems.

Most catalytically active RNAs are known to require divalent metal ions to establish full catalytic activity. This requirement, however, does not necessarily imply a mechanistic participation of the metal ions in catalysis, as their roles can also be purely structural. Examples for both cases are well described. For the extended hammerhead and hairpin ribozyme, crystallographic and mechanistic data suggest primarily a structural role of Mg²⁺ ions (Martick and Scott, 2006; Bevilacqua and Yajima, 2006), whereas in the group I intron, metal ions may be directly involved in the catalytic mechanism (DeRose, 2002). While these investigations provide an understanding of RNA’s strategies to catalyze phosphodiester chemistry, very little is known about how RNA accelerates other types of chemical reactions. One of the best-characterized artificial ribozymes catalyzes carbon-carbon bond formation by Diels-Alder reaction, a [4+2]-cycloaddition between an electron-rich anthracene diene and an electron-deficient maleimide dienophile, with high activity and selectivity (Seelig and Jäschke, 1999). This Diels-Alder ribozyme is active as a true catalyst exhibiting multiple turnover behavior and saturation-type kinetics, and further investigations revealed stereoselective formation of individual product enantiomers (Seelig et al., 2000). From the original ∼150 nucleotide long sequences, a 49-nt minimal ribozyme motif could be extracted, which was found to be fully active [Fig. 1a]. High catalytic activity required the presence of both monovalent and divalent cations (Seelig and Jäschke, 1999).

Figure 1. (a) Secondary structure of the 49-nt Diels-Alder ribozyme and (b) graphical representation of the tertiary fold in the crystal.

(c) Stereo view of the Mg²⁺-soaked crystal structure of the Diels-Alder ribozyme (Serganov et al., 2005, resolution 3.0 Å). Mg²⁺ ions are shown as spheres together with their coordinated water molecules.

The three-dimensional architecture of the Diels-Alder ribozyme was analyzed using mutation studies, a combination of probing techniques, and x-ray crystallography (Keiper et al., 2004; Serganov et al., 2005). The tertiary structure of the ribozyme is formed by an unusual tertiary interaction between the asymmetric internal loop and the 5^′ terminus, resulting in the formation of a nested pseudoknot [Fig. 1b]. Moreover, the probing data indicated that the tertiary structure does not change upon substrate or product binding, thereby suggesting a preformed architecture of the Diels-Alder ribozyme. In the crystal structure of this ribozyme, eight Mg²⁺ ions were observed [Fig. 1c], out of which two are suggested to be due to crystal packing (Serganov et al., 2005). Interestingly, no Mg²⁺ ions were found in the immediate vicinity of the catalytic pocket, indicating that metal ions do not directly participate in catalysis, and a recent computational study attributes the proficiency of the ribozyme-catalyzed reaction to the stabilization of reactive ground state conformations in the ribozyme active site (Zhang and Bruice, 2007). However, the catalytic mechanism is still not fully understood, and metal(II) ions distant from the bond-breaking or bond-forming site have been shown to influence catalysis in other systems (Sigel and Pyle, 2007).

In the present study, we investigate the divalent metal ion binding sites of the 49-nt Diels-Alder ribozyme in solution using electron paramagnetic resonance (EPR) methods. EPR spectroscopy has been extensively utilized to probe metal ion binding sites in proteins (Reed and Markham, 1984; Ubbink et al., 2002; Calle et al., 2006) and recently also in nucleic acids, employing the paramagnetic Mn²⁺ instead of the physiological, but EPR spectroscopically silent Mg²⁺ (Vogt et al., 2006; Schiemann et al., 2003; Kisseleva et al., 2005). Due to the similar radii, Lewis acidity, and coordination chemistry of Mn²⁺ and Mg²⁺ ions, this substitution often causes no or only minor changes in the biological or chemical function. (Reed and Poyner, 2000). However, their coordination to RNAs shows some distinct differences (Freisinger and Sigel, 2007), making a case-by-case evaluation necessary. For the Diels-Alder ribozyme, this approach seems to be suited as the ribozyme was found to retain catalytic activity in the presence of Mn²⁺ (Seelig and Jäschke, 1999). To characterize the multiple metal ion binding sites in this system, we displace individual Mn²⁺ ions by EPR-silent Cd²⁺ ions, allowing the spectroscopic characterization of the remaining bound Mn²⁺ions.

RESULTS AND DISCUSSION

Quantification of Mn²⁺ binding sites at room temperature

In aqueous solution, Mn²⁺ forms [Mn(H₂O)₆]²⁺ ions, which give rise to a six-line continuous wave (cw) EPR signal at room temperature (Abragam and Bleaney, 1986), with the signal intensity being proportional to the concentration of Mn²⁺. The binding of Mn²⁺ to RNA slows down the rotation of the complex and creates an asymmetric ligand field, which both lead to excessive line broadening and ultimately to the disappearance of the signal (Horton et al., 1998). This effect is exploited here to quantify the Mn²⁺ sites in the Diels-Alder ribozyme by monitoring the EPR signal intensity while titrating Mn²⁺ into the buffered ribozyme solution (Supplementary Fig. S1). For comparison, Mn²⁺ was also titrated to the same buffer solution in the absence of RNA, and the binding isotherm was constructed from the difference of both curves and fitted to Eq. 1 (see “Experimental” section). This analysis revealed that in solution, the Diels-Alder ribozyme possesses five high-affinity Mn²⁺ binding sites with an upper and apparent (Sigel and Griesser, 2005) dissociation constant K_d of 0.6±0.2 μM, which is within a typicalrange for ribozyme-bound Mn²⁺ ions (Horton et al., 1998; Schiemann et al., 2003; Kisseleva et al., 2005). Titrations of the Diels-Alder ribozyme at different monovalent salt concentrations ranging from 0.1 to 4.3 M always led to five Mn²⁺ binding sites with an unchanged K_d≤0.6 μM (data not shown), demonstrating the specificity of the binding sites for divalent ions. This specificity differs markedly from the behavior of the hammerhead ribozymes for which only a single high-affinity binding site with a K_d of 4 μM prevails at high monovalent ion concentrations.

Low temperature studies on the Mn²⁺ binding sites

While the five high-affinity Mn²⁺ ions bound to the ribozyme show no EPR signal at room temperature (see above), lowering the temperature to 4.2 K slows the Mn²⁺ relaxation, and characteristic EPR spectra can be recorded (Fig. 2). The ribozyme with one Mn²⁺ ion bound ([ribozyme∕1 Mn²⁺] complex) yields an anisotropic Mn²⁺ spectrum with several features that spreads from a magnetic field value B₀=100 to 6500 G. Similar Mn²⁺ spectra were previously observed for Mn²⁺-oxalate and -pyruvate complexes (Reed and Cohn, 1973), Mn²⁺ dioxygenase (Boldt et al., 1997), and Mn²⁺ superoxide dismutase and are distinctive for monomeric Mn²⁺ ions with a highly asymmetric coordination sphere, characterized by a large zero-field splitting D (Whittaker and Whittaker, 1991; Rusnak et al., 1999; White et al., 2001; Copik et al., 2005). One typical feature of such Mn²⁺ centers is a signal at g≈4.0 with a weakly resolved hyperfine splitting of roughly six lines and an average hyperfine coupling constant A≈90 G (Griscom and Griscom, 1967; Reed and Markham, 1984; Haddy et al., 1992; Ananyev and Dismukes, 1997), as observed for this ribozyme (Fig. 2). Although simulations of such manganese cw EPR spectra are complicated, the agreement between simulation and experiment (Supplementary Fig. S2) confirms the presence of a monomeric Mn²⁺ ion, with a large zero-field splitting D, here of ≈1500 G (see Supplementary Material for the complete set of parameters and Supplementary Fig. S2).

Figure 2. Cw X-band EPR spectra of the [Diels-Alder ribozyme∕ x Mn ²⁺ ] complexes with x =1, 2, 3, 4, 5 at 4.2 K.

The samples contained 0.2 mM Diels-Alder ribozyme in Tris-HCl buffer (pH 7.3, 0.3 M NaCl). EPR conditions: X-band frequency 9.424 GHz; microwave power 0.2 mW; modulation amplitude 22 G; scans 10.

The spectra of the [ribozyme∕1 Mn²⁺], [ribozyme∕2 Mn²⁺], and [ribozyme∕3 Mn²⁺] complexes appear to be identical to each other, only the EPR signal intensity increases with the number of bound Mn²⁺ ions, indicating that all three ions are monomeric and that their electronic structures are comparable (Fig. 2).

The spectrum changes dramatically when the ribozyme∕Mn²⁺ ratio reaches 1:4. The intense features at g=1.1 (6000 G) and 1.3 (5277 G) in the monomeric spectra are now strongly diminished and shifted by about 100 G to lower field. The resonances at g=1.6 (4265 G) and 2.17 (3095 G) disappear, whereas new broad and intense signals show up at g=1.8 (3665 G), 2.3 (2891 G), and 3.0 (2249 G). In addition, the resonances below 2000 G change their appearance and intensity. This new spectrum can be assigned to an electronically coupled dimer which is formed by two out of the four Mn²⁺ ions. The two remaining monomeric Mn²⁺ ions have weak spectral intensities and are hidden under the strong signal of the Mn²⁺ dimer.

Spectra with similar general appearance and intense broad resonances in the region between 2000 and 4600 G have been reported for model complexes of Mn²⁺-saldien (Mabad et al., 1986; Kitajima et al., 1991) and for rat liver arginase (Reczkowski and Ash, 1992; Khangulov, 1998) for which the dimeric Mn²⁺ centers are well established. A shift of the two high-field transitions, as observed here, has also been attributed to the formation of a dimeric Mn²⁺ center (Khangulov et al., 1995; White et al., 2001). In addition, dinuclear Mn²⁺ centers are often identified by a characteristic 11-line pattern with a splitting of 44 G (Epel et al., 2005), which is, however, not always observed (Golombek and Hendrich, 2003). In the case of the [ribozyme∕4 Mn²⁺] complex, the fine transitions at g≈1.6 and g≈1.85 may indicate such an average hyperfine splitting. However, the resolution of the splitting is too weak to be used as direct evidence for dimer formation. To further support the idea of a Mn²⁺-dimer formation, we have simulated the spectrum of the [ribozyme∕4 Mn²⁺] complex (see Supplementary Material for the complete set of parameters and Supplementary Fig. S3). The simulations resulted in an upper limit of the dipolar coupling constant D_dip of ∼360 MHz, which corresponds to a Mn²⁺-Mn²⁺ distance of ∼6 Å (Eaton and Eaton, 2000).

Addition of a fifth Mn²⁺ ion yields the [ribozyme∕5 Mn²⁺] complex with a spectrum that is only slightly different compared to the spectrum of the [ribozyme∕4 Mn²⁺] complex (Fig. 2). The intensity of the two high-field resonances increases, and they are shifted back to higher field [g=1.3 (5277 G) and g=1.1 (6000 G)] as found for the monomeric Mn²⁺ spectra. Therefore, the fifth Mn²⁺ does not engage in an additional dimer but is more likely a magnetically isolated Mn²⁺ ion whose signal is masked by the intense dimer spectrum.

These low-temperature studies indicate that the five Mn²⁺ ions possess different affinities and that the obtained K_d of 0.6 μM is indeed only an upper value.

Competition experiments with Cd²⁺ ions

Detailed characterizations of metal ion binding sites are typically performed on the sites of highest affinity, while the characterization of the lower-affinity sites is hampered by the presence of the occupied high-affinity sites. To characterize the lower-affinity sites while at the same time having the high-affinity sites populated (as would be the case in the catalytically active structure) we devise here a method for the silencing of the high-affinity sites by using EPR-inactive Cd²⁺ ions.

In a first step, the ability of Cd²⁺ to replace Mn²⁺ ions was probed. The [Diels-Alder ribozyme∕5 Mn²⁺] complex was prepared, showing no cw EPR signal at room temperature. The titration of Cd²⁺ into this solution led to the reappearance of the [Mn(H₂O)₆]²⁺ cw-EPR signal with an intensity that is within the experimental error comparable to the signal intensity of a sample of 40 μM Mn²⁺ [Fig. 3a]. This demonstrates that Cd²⁺ replaces all five high-affinity Mn²⁺ ions and that roughly one Cd²⁺ ion replaces one Mn²⁺ ion from the ribozyme. The other way around, the titration of the [ribozyme∕5 Cd²⁺] complex with Mn²⁺ showed no binding of Mn²⁺ ions up to a Mn²⁺ concentration of 1 mM (data not shown). Consequently, it can be concluded that Cd²⁺ binds stronger to the high-affinity binding sites than Mn²⁺, and that Cd²⁺ cannot be substituted by Mn²⁺.

Figure 3. (a)Cd ²⁺ titration curve of a sample containing 40 μMMn ²⁺ in 30 mM Tris-HCl buffer, pH 7.3, 0.3 M NaCl (squares) and of a sample containing.

8 μM of the Diels-Alder complex [ribozyme∕5Mn²⁺] in the same buffer (open circles). (b) Schematic of the Cd²⁺∕Mn²⁺ distribution in the Diels-Alder ribozyme. Open circle: nonoccupied innersphere binding site; Open circle with cross: innersphere binding site occupied by Cd²⁺; Solid circle: innersphere binding site occupied by Mn²⁺; Open square: nonoccupied outersphere binding site; Square with cross: outersphere binding site occupied by Cd²⁺; Solid square: outersphere binding site occupied by Mn²⁺. (c) cw X-band EPR spectra of the following complexes recorded at 4.2 K. (A) [ribozyme∕1 Mn²⁺], (B) free [Mn(H₂O)₆]²⁺ ion in water, (C) [ribozyme∕3 Cd²⁺∕1 Mn²⁺], (D) [ribozyme∕1 Cd²⁺∕4 Mn²⁺], and (E) [ribozyme∕4 Mn²⁺].

This ability of Cd²⁺ was used to prepare the following ribozyme∕metal complexes: one Cd²⁺ and no Mn²⁺ (1Cd²⁺-0Mn²⁺), 1Cd²⁺-1Mn²⁺, 1Cd²⁺-2Mn²⁺, 1Cd²⁺-3Mn²⁺, and 1Cd²⁺-4Mn²⁺. Then complexes with 2Cd²⁺, 3Cd²⁺, and 4Cd²⁺ ions were made and the free metal binding sites subsequently filled with Mn²⁺ ions. In every preparation the Cd²⁺ ions were added first so that the Mn²⁺ ions can occupy only the free binding sites. The occupation of the binding sites can be followed by the schematic representation of the experiment as depicted in Fig. 3b. The cw EPR spectra of these ten [ribozyme∕x Cd²⁺∕y Mn²⁺] complexes were recorded at 4.2 K and are summarized in Fig. 3c together with the spectrum of free [Mn(H₂O)₆]²⁺.

For the [ribozyme∕x Cd²⁺∕y Mn²⁺] complexes with a total amount of metal(II) ions of no more than three (1 Mn²⁺, 2 Mn²⁺, 3 Mn²⁺, 1Cd²⁺-1Mn²⁺, 1Cd²⁺-2Mn²⁺, 2Cd²⁺-1Mn²⁺), the spectral features are identical to the spectrum of the [ribozyme∕1 Mn²⁺] complex [Fig. 3c, trace A] but are distinctively different from free [Mn(H₂O)₆]²⁺ with octahedral symmetry [Fig. 3c, trace B]. This confirms the finding that the first three binding sites have comparable chemical environments. It furthermore indicates that the addition of Cd²⁺ to one binding site does not lead to significant structural perturbations at the other binding sites. The large anisotropy of these spectra is interpreted to be due to an asymmetric innersphere coordination of these first three monomeric Mn²⁺ ions.

In order to select the last two high-affinity binding sites we occupied the first three sites with Cd²⁺ ions and added one Mn²⁺ (3Cd²⁺-1Mn²⁺). The resulting spectrum with a dominant sextet at g=2 (3400 G) is shown in Fig. 3c, trace C. This spectrum is totally different compared with the spectra of the first three monomeric innersphere bound Mn²⁺ ions and of the free [Mn(H₂O)₆]²⁺. To exclude the possibility that the spectrum is a superposition of free and bound Mn²⁺, the solution was warmed to room temperature and showed no EPR signal, thereby verifying that all Mn²⁺ is bound. The intense sextet at g=2 suggests a more symmetric octahedral coordination geometry (Antanaitis et al., 1987) and indicates that this fourth high-affinity binding site binds a manganese-hexa-aqua ion through outersphere contacts (Reed and Poyner, 2000). The same spectrum was obtained for the [ribozyme∕ 4Cd²⁺∕1 Mn²⁺] complex, suggesting that the fifth Mn²⁺ is an outersphere bound [Mn(H₂O)₆]²⁺ too. Consequently, the [ribozyme∕3 Cd²⁺∕2 Mn²⁺] complex shows the same spectrum but with roughly doubled intensity. All the other combinations of metal(II) ion ratios (1Cd²⁺-3Mn²⁺, 1Cd²⁺-4Mn²⁺, 2Cd²⁺-2Mn²⁺, 2Cd²⁺-3Mn²⁺) show only superpositions of the spectra of the innersphere and outersphere bound Mn²⁺ ions in the respective ratios [trace D, Fig. 3c].

Interestingly, the [ribozyme∕1 Cd²⁺∕3 Mn²⁺] complex in which one of the three innersphere high-affinity binding sites is occupied by Cd²⁺ shows no Mn²⁺ dimer spectrum as observed for the [ribozyme∕4 Mn²⁺] complex. Similarly, no dimer spectrum was obtained in the case of the [ribozyme∕1 Cd²⁺∕4 Mn²⁺] complex [Fig. 3c]. We also turned the experiment around in the sense that first the four Mn²⁺ ions were added yielding the dimer spectrum, followed by the addition of one Cd²⁺, which led to the loss of the dimer spectrum and the appearance of the superposition of the monomeric innersphere and outersphere Mn²⁺ spectra. In fact, none of the spectra with at least one Cd²⁺ show the Mn²⁺ dimer spectrum.

These data lead us to conclude that the first and fourth ion selectively occupy sites critical for the formation of the dimer. Different scenarios can be considered to explain the observations. One possibility is that metal ion 4 is directly involved in the dimer and couples with one of the other three, or the binding of the fourth Mn²⁺ ion induces a conformational change, and the dimer is in fact formed from any combination out of the first three ions. According to scenario 1, the highest-affinity site occupied by the first Cd²⁺ ion belongs to the dimer actually formed from metal ions 1 and 4. Thus, the metal-metal dimer is still formed, but does not give rise to an electronically coupled dimer EPR spectrum as one of its ions is diamagnetic. In case of a conformational change (scenario 2), the first Cd²⁺ still occupies one of the dimer sites but pairs after the conformational change with either ion 2 or 3. Again, the formed dimer would not be electronically coupled.

The fact that the sextet at g=2 characteristic for outersphere bound [Mn(H₂O)₆]²⁺ is not visible in the [ribozyme∕4 Mn²⁺] spectrum supports the idea that the dimer is indeed formed by sites 1 and 4, since the fourth ion would no longer be a monomeric outersphere ion but part of the dimer with the characteristic dimer features. In case of the other scenario, the fourth ion would remain the monomeric outersphere ion and should show the sextet, which is, however, not observed in the experiment. Finally, we can rule out the possibility that the binding of Cd²⁺ to site 1 inhibits the conformational change induced by metal ion binding to site 4 and thereby prevents the formation of a dimer between ions 2 and 3, because the ribozyme remains catalytically active in the presence of cadmium ions, which it should not if the structural change did not take place (see below).

Based on these data, we conclude that the binding sites possess different affinities in the following order: site 1>site 2≅site 3>site 4>site 5. The order in which the binding sites are occupied seems to be the same for Mn²⁺ and Cd²⁺, since the sequence in which the outersphere and innersphere EPR spectra appear remains unchanged [compare Figs. 23c]. Yet it should be mentioned that we cannot fully rule out the possibility that the different metal binding sites are to a certain extent also statistically occupied. However, the percentage of this statistical occupation appears to happen on a minor scale according to all EPR spectroscopic evidence outlined above.

Influence of AMDA on the Mn²⁺ binding sites

For the minimal hammerhead ribozyme the cationic organic ligand neomycin was found to inhibit the catalytic self-cleavage of this RNA (Clouet-d’Orval et al., 1995) and to displace the high-affinity metal(II) ion (Schiemann et al., 2003). In the case of the Diels-Alder ribozyme the charge-neutral product analog AMDA binds to the catalytic site and inhibits the multiple-turnover reaction (Keiper et al., 2004). Therefore, we were interested in the question whether this inhibition is also accompanied by metal(II) displacement.

The titration of the [Diels-Alder ribozyme∕5 Mn²⁺] complex with AMDA up to an excess of 100-fold at room temperature does not lead to the appearance of a signal corresponding to free Mn²⁺ ions [Fig. 4a]. This indicates that AMDA does not replace any of the five high affinity Mn²⁺ ions from the Diels-Alder ribozyme molecule.

Figure 4. (a) AMDA titration curves of the [Diels-Alder ribozyme∕5Mn ²⁺ ].

(b) Mn²⁺ titration curves of samples containing only buffer (30 mM Tris-HCl, pH 7.3, 0.3 M NaCl) (squares) and of samples containing 8 μM Diels-Alder ribozyme, 0.8 mM AMDA in the same buffer (open circles).

To investigate whether the presence of AMDA influences the binding of Mn²⁺ ions to the ribozyme, a sample containing the Diels-Alder ribozyme and AMDA in ratios of 1:1, 1:40, 1:100 was titrated with Mn²⁺ [Fig. 4b]. These titrations showed that five Mn²⁺ ions were bound per ribozyme with an upper K_d of 0.6 μM as in the titration without AMDA. Additionally, the [ribozyme∕5 Mn²⁺] complex shows the same cw X-band EPR spectrum at 4.2 K in the presence of bound AMDA (complex:AMDA=1:100) as without AMDA. Thus, AMDA has no influence on the number, affinity, and general structure of the Mn²⁺ binding sites. This agrees with the crystal structure in which no metal(II) ions are located in the catalytic pocket and which shows no changes in the Mg²⁺ binding sites with AMDA bound.

Correlation of the EPR data with crystallographic and kinetic information

The crystal structure shows a total of eight Mg²⁺ ions [Serganov et al., 2005; Fig. 1c]. Out of these eight, the metal(II) ions Mg7 (orange) and Mg8 (red) make contacts between two RNA molecules in the crystal lattice. However, these two ions should not be present in solution, since the ribozyme acts—according to all available biochemical evidence (Wombacher et al., 2006)—as a monomeric enzyme. Furthermore, Mg4 (dark green, outersphere) apparently binds to a stretch of regular A-RNA (blue helix I), rendering a high-affinity interaction in solution less likely. This leaves a total of five potential metal ion binding sites, which agrees well with the five high-affinity binding sites found EPR spectroscopically.

Of the remaining five metal(II) ions in the crystal, two ions (Mg1 and Mg2, shown in dark blue and light blue, respectively) are innersphere coordinated, involving phosphate oxygens, and are located in close proximity to each other (7.4 A distance). Mg3 (cyan) is outersphere coordinated at the interface of the catalytic pocket and helix III. Mg5 (light green)—although only outersphere coordinated—seems to stabilize the back of the catalytic pocket by intricate coordination. Mg6 (gold, outersphere) sits at the interface of helix I and one of the pseudoknot-forming minihelices. Thus, the crystal structure and the EPR experiments reveal both innersphere and outersphere bound ions. However, an assignment of the outersphere and innersphere ions found in the EPR studies to the ions in the crystal structure must be done with caution because the EPR experiments were performed with Mn²⁺ whereas Mg²⁺ was used for the crystal structure. As known from the crystal structures of the minimal hammerhead ribozyme, Mn²⁺ ions can be shifted within the binding sites compared to Mg²⁺ ions (Pley et al., 1994; Scott et al., 1995; Scott et al., 1996), which is also in agreement with the hard and soft acids and bases (HSAB) concept. Furthermore, differences could arise from the ambiguities associated with the 3.0 Å resolution of the crystal structure. Hence, it is not surprising that the amount of the respective metal(II) ion types differs, 2 innersphere and 3 outersphere in the crystal structure and 3 innersphere and 2 outersphere in solution by EPR.

According to all evidence, the electronically coupled dimer identified by EPR corresponds to Mg1-Mg2 in the crystal structure. The distance of ∼6 Å inferred from the EPR spectrum agrees well with the distance of 7.4 Å measured between Mg1 and Mg2 in the crystal, especially considering the exchange of Mg²⁺ by Mn²⁺. A close-up view of the chemical environment [Fig. 5a] shows a shared binding pocket in which two neighboring phosphates, connected by one ribose, provide the innersphere ligands and various N- and O-ligands provide outersphere coordination at distances between 3.9 and 4.4 A. These two metal ions are obviously important for folding into the catalytically active structure and in particular for packing one of the pseudoknot minihelices tightly into the major grove of helix II. Furthermore, in the crystal, Mg2 stabilizes a crucial reversed Hoogsteen base pair. This is, to our knowledge, the first demonstration of an electronically coupled metal ion dimer in RNA.

Figure 5. Kinetic and structural correlations.

(a) Close-up view of the structure of the metal ion dimer formed by Mg1 (dark blue, left) and Mg2 (light blue, right) and its binding site as observed in the crystal structures (PDB code 1YLS). (b) Dependence of the initial rate on the metal(II) ion concentration. Highlighted spots (circled) represent metal ion mixtures.

In order to correlate the EPR data with catalytic activity, the initial rate of the ribozyme-catalyzed multiple-turnover Diels-Alder reaction was studied as a function of the concentration of Mg²⁺, Mn²⁺, and Cd²⁺ ions [Fig. 5b]. With Mg²⁺, we observe a sigmoidal dependence with a K_1∕2 of 150 μM, corresponding to 21 Mg ions per RNA molecule. With Mn²⁺, the catalytic performance is reduced by about half, and K_1∕2 increases slightly to 200 μM. These figures indicate that high-affinity binding alone is not sufficient for full catalytic activity, and further RNA-metal ion interactions are required.

Cd²⁺ ions alone do not induce catalytic activity. However, the addition of Cd²⁺ to a catalytically active ribozyme mixture containing either Mg²⁺ or Mn²⁺ had only minor effects on the catalytic activity [the highlighted spots in Fig. 5b]. These latter experiments were performed under conditions identical to those in the respective EPR measurements, indicating that the high-affinity binding sites are occupied by Cd²⁺ in these enzyme assays too. The high-catalytic activity of the ribozyme in Mn²⁺∕Cd²⁺ mixtures confirms that the use of Cd²⁺ in the EPR experiments did not lead to the formation of catalytically incompetent RNA structures and that cadmium-induced EPR silencing is apparently a valid method for the investigation of multiple metal ion-RNA interactions.

CONCLUSIONS

To conclude, the Diels-Alder ribozyme in solution contains five high-affinity Mn²⁺ binding sites with an upper K_d of 0.6±0.2 μM irrespective of the monovalent ion concentration. This renders Mg²⁺ ions 4, 7, and 8 found in the crystal structure as likely being due to solid state interactions. Out of the five metal(II) ions, two are outersphere and three are innersphere bound, where the latter ones possess higher affinities than the two outersphere bound ions. The competition experiments with Cd²⁺ showed that the affinities are also different within the two classes, allowing the five sites to be occupied consecutively. Two Mn²⁺ ions were found to form an electronically coupled dimer with a metal-to-metal distance of ∼6 Å, which is proposed to correspond to Mg1 and Mg2 in the crystal structure with a measured intermetal distance of 7.4 Å. The EPR measurements in the presence of the product analog inhibitor AMDA provide further support for the concept of a preformed catalytic pocket (Keiper et al., 2004) and suggests furthermore that the high-affinity metal(II) ions are not directly involved in the catalytic carbon-carbon bond formation but are involved in the stabilization of the RNA fold.

To obtain more detailed structural information about the different binding sites, temperature-dependent high-field EPR measurements and pulsed EPR experiments combined with site-directed spin labeling are underway. However, the methodology of cadmium-induced EPR silencing developed within this project was already found to be a useful tool for the investigation of RNA molecules interacting with multiple metal ions of one sort.

EXPERIMENTAL SECTION

RNA oligonucleotides and chemicals

The 49-nt Diels-Alder ribozyme (for sequence, secondary, and tertiary structure, see Fig. 1) was purchased from CSS Chemical Synthesis Service (Craigavon, UK). For the EPR experiments, the RNA was dissolved in 30 mM Tris-HCl buffer solution (pH 7.3) containing 0.1–4.3 M NaCl (>99.999%). MnCl₂ and CdCl₂ (>99.99%) were obtained from Sigma-Aldrich and were dissolved in the same buffer to a final concentration of 10 mM. The product analog inhibitor AMDA was synthesized as described previously (Stuhlmann and Jäschke, 2002), and 10 mM solutions of AMDA were prepared in Tris-HCl buffer with 10% acetonitrile and 10% ethanol as organic modifiers. Anthracene hexaethylene glycol (AHEG) for activity measurements was synthesized according to Fiammengo et al. (2005). All other chemicals were obtained from Sigma-Aldrich, Fluka, and Carl Roth.

EPR experiments

Cw X-band EPR spectra were recorded on a Bruker ESP 500e EPR spectrometer, equipped with an ER 4103TM cylindrical mode resonator for room temperature measurements in aqueous solutions. The titrations of the ribozyme with Mn²⁺ or ADMA were performed at 295 K using a sterilized flat cell with a volume of 150 μl and the following spectrometer parameter: center field 3480, sweep width 1000 G, 1024 points, microwave power 2 mW, modulation amplitude 12 G, modulation frequency 100 kHz, conversion time 41 ms, time constant 41 ms, 20 scans. The concentration of the ribozyme was in each case 8 μM. The resulting Mn²⁺ EPR signals were base-line corrected, doubly integrated, and the magnitude of the integral was plotted against the concentration of added Mn²⁺. Binding isotherms for Mn²⁺ were constructed by plotting the concentration of bound Mn²⁺ divided by the concentration of ribozyme in solution ($\left[{\mathrm{Mn}}_{\text{bound}}^{2+}\right]$∕[ribozyme]) versus the concentration offree Mn²⁺$\left(\right[{\mathrm{Mn}}_{\text{free}}^{2+}\left]\right)$. The concentration of bound Mn²⁺ was determined by comparing the Mn²⁺-signal intensity of a Mn²⁺-ribozyme sample with a Mn²⁺ standard sample containing 8 μM of free Mn²⁺. The difference in intensity between both is assigned to the concentration of bound Mn²⁺. The binding isotherms were then fitted according to Eq. 1 assuming j classes of n independent noninteracting binding sites to determine the dissociation constants K_d:

$$\frac{\left[{\mathrm{Mn}}_{\mathrm{bound}}^{2+}\right]}{\left[\mathrm{ribozyme}\right]}=\sum _{i=1}^j\frac{n_i\left[{\mathrm{Mn}}_{\mathrm{free}}^{2+}\right]}{K_{d\left(i\right)}+\left[{\mathrm{Mn}}_{\mathrm{free}}^{2+}\right]}.$$ (1)

Each experiment was repeated at least three times and the results were in each case reproducible (the error bars are shown on the graphs).

Low-temperature experiments at 4.2 K were performed on samples containing 0.2 mM Diels-Alder ribozyme in Tris-HCl buffer with 0.3 M NaCl and 20% sucrose. The cw-EPR spectra were recorded at this temperature with an ER 4102ST rectangular resonator and a cryostat from Oxford using the following parameters: center field 3350, sweep width 6500 G, 2048 points, microwave power 0.2 mW, modulation amplitude 23 G, modulation frequency 100 kHz, conversion time 82 ms, time constant 82 ms, 10 scans. The simulations of cw-EPR spectra were performed using XSophe software (University of Queensland, Brisbane, Australia and Bruker Biospin GmbH).

Ribozyme activity measurements

All substances were dissolved in water, except for N-pentyl maleimide (NPM) which was dissolved in ethanol (10 mM stock solution). The assay was conducted in a 7 μl cuvette (Hellma) at room temperature in a Cary 50 UV spectrometer (Varian). Components were added in the following order, with the final concentrations in brackets: Tris-HCl buffer pH 7.4 (30 mM), NaCl (300 mM), ribozyme (7 μM), divalent ions (Mg²⁺, Mn²⁺, Cd²⁺) (range of 0–80000 μM), AHEG (100 μM). Reactions were started by adding NPM (1 mM). The decrease in anthracene absorbance at 365 nm was recorded over 10 min.

For the determination of the initial rate V_ini [μM∕min], only the first 5% decrease of absorbance was analyzed by linear regression. The second-order rate constant was calculated from Eq. 2:

$$k\left[{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}\right]=\frac{V[\mathrm{M}∕\mathrm{s}]}{{\mathrm{Ant}}_0\left[\mathrm{M}\right]\bullet {\mathrm{NPM}}_0\left[\mathrm{M}\right]}.$$ (2)