Mg(II) binding by bovine prothrombin fragment 1 via equilibrium dialysis and the relative roles of Mg(II) and Ca(II) in blood coagulation

The first direct equilibrium dialysis titration of the blood coagulation protein bovine prothrombin fragment 1 with Mg(II) is presented. Fragment 1 has fewer thermodynamic binding sites for Mg(II) than Ca(II), less overall binding affinity, and significantly less cooperativity. Several nonlinear curve fitting models were tested for describing the binding of fragment 1 with Mg(II), Ca(II), and mixed metal binding data. The Mg(II) data is represented by essentially five equivalent, noninteracting sites; for Ca(II), a model with three tight, cooperative sites and four "loose", equal affinity, noninteracting sites provides the best model. Based on the reported equilibrium dialysis data and in conjunction with other experimental data, a model for the binding of Ca(II) and Mg(II) to bovine prothrombin fragment 1 is proposed. The key difference between the binding of these divalent ions is that Ca(II) apparently causes a specific conformational change reflected by the cooperativity observed in the Ca(II) titration. The binding of Ca(II) ions to the three tight, cooperative sites establishes a conformation that is essential for phospholipid X Ca(II) X protein binding. The filling of the loose sites with Ca(II) ions leads to charge reduction and subsequent phospholipid X Ca(II) X protein complex interaction. Binding of Mg(II) to bovine prothrombin fragment 1 does not yield a complex with the necessary phospholipid-binding conformation. However, Mg(II) is apparently capable of stabilizing the Ca(II) conformation as is observed in the mixed metal ion binding data and the synergism in thrombin formation.     

Structural and biochemical characterization of a new Mg(2+) binding site near Tyr94 in the restriction endonuclease PvuII

We have determined the crystal structure of the PvuII endonuclease in the presence of Mg(2+). According to the structural data, divalent metal ion binding in the PvuII subunits is highly asymmetric. The PvuII-Mg(2+) complex has two distinct metal ion binding sites, one in each monomer. One site is formed by the catalytic residues Asp58 and Glu68, and has extensive similarities to a catalytically important site found in all structurally examined restriction endonucleases. The other binding site is located in the other monomer, in the immediate vicinity of the hydroxyl group of Tyr94; it has no analogy to metal ion binding sites found so far in restriction endonucleases. To assign the number of metal ions involved and to better understand the role of Mg(2+) binding to Tyr94 for the function of PvuII, we have exchanged Tyr94 by Phe and characterized the metal ion dependence of DNA cleavage of wild-type PvuII and the Y94F variant. Wild-type PvuII cleaves both strands of the DNA in a concerted reaction. Mg(2+) binding, as measured by the Mg(2+) dependence of DNA cleavage, occurs with a Hill coefficient of 4, meaning that at least two metal ions are bound to each subunit in a cooperative fashion upon formation of the active complex. Quenched-flow experiments show that DNA cleavage occurs about tenfold faster if Mg(2+) is pre-incubated with enzyme or DNA than if preformed enzyme-DNA complexes are mixed with Mg(2+). These results show that Mg(2+) cannot easily enter the active center of the preformed enzyme-DNA complex, but that for fast cleavage the metal ions must already be bound to the apoenzyme and carried with the enzyme into the enzyme-DNA complex. The Y94F variant, in contrast to wild-type PvuII, does not cleave DNA in a concerted manner and metal ion binding occurs with a Hill coefficient of 1. These results indicate that removal of the Mg(2+) binding site at Tyr94 completely disrupts the cooperativity in DNA cleavage. Moreover, in quenched-flow experiments Y94F cleaves DNA about ten times more slowly than wild-type PvuII, regardless of the order of mixing. From these results we conclude that wild-type PvuII cleaves DNA in a fast and concerted reaction, because the Mg(2+) required for catalysis are already bound at the enzyme, one of them at Tyr94. We suggest that this Mg(2+) is shifted to the active center during binding of a specific DNA substrate. These results, for the first time, shed light on the pathway by which metal ions as essential cofactors enter the catalytic center of restriction endonucleases.     

Protein interactions with immobilized transition metal ions: quantitative evaluations of variations in affinity and binding capacity

The interaction of proteins with immobilized transition-metal ions proceeds via mechanisms influenced by metal type and degree of coordination, variations in mobile phase constituents, and protein surface architecture at or near the metal binding site(s). The contributions each of these variables make toward the affinity of protein surfaces for immobilized metal ions remain empirical. We have used equilibrium binding analyses to evaluate the influence of pH and competitive binding reagents on the apparent equilibrium dissociation constant (Kd) and binding capacity of immobilized Cu(II) and Ni(II) ions for several model proteins of known three-dimensional structure. Linear Scatchard plots suggested that 8/13 of the proteins evaluated interacted with immobilized metal ions via a single class of operational (Kd = 10-700 microM) binding sites. Those proteins with the highest affinities for the immobilized Cu(II) ions (5/13) showed evidence of multiple, non-identical or nonindependent binding sites. The effects of altered metal type, pH, and concentration of competitive affinity reagents (e.g., imidazole, free metal ions) on the apparent Kd and binding capacity varied in magnitude for individual proteins. The presence of free Cu(II) ions did not detectably alter either the affinity or binding capacity of the proteins for immobilized Cu(II) ions. The expected relationship between the relative chromatographic elution sequence and calculated affinity constants was not entirely evident by evaluation under only one set of conditions. Our results demonstrate the utility of nonchromatographic equilibrium binding analyses for the quantitative evaluation of experimental variables affecting the relative affinity and capacity of immobilized metal ions for proteins. This approach affords the opportunity to improve understanding and to vary the contribution of interaction mechanisms involved.     

Uncoupling metallonuclease metal ion binding sites via nudge mutagenesis

The hydrolysis of phosphodiester bonds by nucleases is critical to nucleic acid processing. Many nucleases utilize metal ion cofactors, and for a number of these enzymes two active-site metal ions have been detected. Testing proposed mechanistic roles for individual bound metal ions has been hampered by the similarity between the sites and cooperative behavior. In the homodimeric PvuII restriction endonuclease, the metal ion dependence of DNA binding is sigmoidal and consistent with two classes of coupled metal ion binding sites. We reasoned that a conservative active-site mutation would perturb the ligand field sufficiently to observe the titration of individual metal ion binding sites without significantly disturbing enzyme function. Indeed, mutation of a Tyr residue 5.5 A from both metal ions in the enzyme-substrate crystal structure (Y94F) renders the metal ion dependence of DNA binding biphasic: two classes of metal ion binding sites become distinct in the presence of DNA. The perturbation in metal ion coordination is supported by 1H-15N heteronuclear single quantum coherence spectra of enzyme-Ca(II) and enzyme-Ca(II)-DNA complexes. Metal ion binding by free Y94F is basically unperturbed: through multiple experiments with different metal ions, the data are consistent with two alkaline earth metal ion binding sites per subunit of low millimolar affinity, behavior which is very similar to that of the wild type. The results presented here indicate a role for the hydroxyl group of Tyr94 in the coupling of metal ion binding sites in the presence of DNA. Its removal causes the affinities for the two metal ion binding sites to be resolved in the presence of substrate. Such tuning of metal ion affinities will be invaluable to efforts to ascertain the contributions of individual bound metal ions to metallonuclease function.     

Activation of hypoxanthine/guanine phosphoribosyltransferase from yeast by divalent zinc and nickel ions

We have observed previously that the reactions catalyzed by hypoxanthine/guanine phosphoribosyltransferase (HGPRTase) are activated by Mg(II), Mn(II), and Co(II), and we have defined the mechanism by which these activations proceed [Biochemistry 22, 3419-3424 (1983)]. A more extensive survey of the kinds of metal ions that will activate the HGPRTase catalysis now has been completed through the use of an HPLC assay procedure. Although Fe(II) and Ca(II) are unable to activate this reaction, a significant activation was achieved with the addition of spectroscopically pure Zn(II) to the assay solution. In addition some IMP synthesis resulted from the addition of Ni(II) to the assay mixture. Both the Zn(II) and Ni(II) kinetic effects on HGPRTase over a limited metal ion concentration range have been analyzed through the use of curve-fitting exercises. These results, in addition to the similar pH profiles for the activations by Mg(II), Mn(II), Co(II), and Zn(II), suggest that all of these metal ions activate the HGPRTase-catalyzed synthesis of IMP by way of the same mechanism [model II as defined by London and Steck, Biochemistry 8, 1767-1779 (1969)], during which two divalent ions bind to the HGPRTase active site per molecule of PRibPP.     

The metallomics approach: use of Fe(II) and Cu(II) footprinting to examine metal binding sites on serum albumins

Metal binding to serum albumins is examined by oxidative protein-cleavage chemistry, and relative affinities of multiple metal ions to particular sites on these proteins were identified using a fast and reliable chemical footprinting approach. Fe(ii) and Cu(ii), for example, mediate protein cleavage at their respective binding sites on serum albumins, in the presence of hydrogen peroxide and ascorbate. This metal-mediated protein-cleavge reaction is used to evaluate the binding of metal ions, Na(+), Mg(2+), Ca(2+), Al(3+), Cr(3+), Mn(2+), Co(2+), Ni(2+), Zn(2+), Cd(2+), Hg(2+), Pb(2+), and Ce(3+) to albumins, and the relative affinities (selectivities) of the metal ions are rapidly evaluated by examining the extent of inhibition of protein cleavage. Four distinct systems Fe(II)/BSA, Cu(II)/BSA, Fe(II)/HSA and Cu(II)/HSA are examined using the above strategy. This metallomics approach is novel, even though the cleavage of serum albumins by Fe(II)/Cu(II) has been reported previously by this laboratory and many others. The protein cleavage products were analyzed by SDS PAGE, and the intensities of the product bands quantified to evaluate the extent of inhibition of the cleavage and thereby evaluate the relative binding affinities of specific metal ions to particular sites on albumins. The data show that Co(II) and Cr(III) showed the highest degree of inhibition, across the table, followed by Mn(II) and Ce(III). Alakali metal ions and alkaline earth metal ions showed very poor affinity for these metal sites on albumins. Thus, metal binding profiles for particular sites on proteins can be obtained quickly and accurately, using the metallomics approach.     

Investigation of restriction enzyme cofactor requirements: a relationship between metal ion properties and sequence specificity

Restriction enzymes are important model systems for understanding the mechanistic contributions of metal ions to nuclease activity. These systems are unique in that they combine distinct functions which have been shown to depend on metal ions: high-affinity DNA binding, sequence-specific recognition of DNA, and Mg(II)-dependent phosphodiester cleavage. While Ca(II) and Mn(II) are commonly used to promote DNA binding and cleavage, respectively, the metal ion properties that are critical to the support of these functions are not clear. To address this question, we assessed the abilities of a series of metal ions to promote DNA binding, sequence specificity, and cleavage in the representative PvuII endonuclease. Among the metal ions tested [Ca(II), Sr(II), Ba(II), Eu(III), Tb(III), Cd(II), Mn(II), Co(II), and Zn(II)], only Mn(II) and Co(II) were similar enough to Mg(II) to support detectable cleavage activity. Interestingly, cofactor requirements for the support of DNA binding are much more permissive; the survey of DNA binding cofactors indicated that Cd(II) and the heavier and larger alkaline earth metal ions Sr(II) and Ba(II) were effective cofactors, stimulating DNA binding affinity 20-200-fold. Impressively, the trivalent lanthanides Tb(III) and Eu(III) promoted DNA binding as efficiently as Ca(II), corresponding to an increase in affinity over 1000-fold higher than that observed under metal-free conditions. The trend for DNA binding affinity supported by these ions suggests that ionic radius and charge are not critical to the promotion of DNA binding. To examine the role of metal ions in sequence discrimination, we determined specificity factors [K(a)(specific)/K(a)(nonspecific)] in the presence of Cd(II), Ba(II), and Tb(III). Most interestingly, all of these ions compromised sequence specificity to some degree compared to Ca(II), by either increased affinity for a noncognate sequence, decreased affinity for the cognate sequence, or both. These results suggest that while amino acid-base contacts are important for specificity, the properties of metal ion cofactors at the catalytic site are also critical for sequence discrimination. This insight is invaluable to our efforts to understand and subsequently design sequence-specific nucleases.     

Proton, calcium, and magnesium binding by peptides containing gamma-carboxyglutamic acid.

gamma-Carboxyglutamic acid (Gla) is believed to bind Ca [II] ions and Mg [II] ions in prothrombin and other coagulation proteins. Binding constants for H+, Ca [II] ions, and Mg [II] ions to Gla-containing peptides are determined using pH and ion selective electrode titrations. The binding constants for peptides containing a single Gla residue are similar to the constants for malonic acid. Peptides containing two Gla residues in sequence (di-Gla peptides) bind Ca [II] ions and Mg [II] ions more strongly. KMgL for the di-Gla peptides is similar to the site-binding constant for Ca [II] ions in denatured BF1. These di-Gla peptides may be useful analogs for metal binding by the disordered Gla domain in BF1.     

Mutational analysis of primosome assembly sites. I. Distinct classes of mutants in the pBR322 Escherichia coli factor Y DNA effector sequences

The assembly of the primosome, a multienzyme complex responsible for priming of lagging-strand DNA synthesis in Escherichia coli, occurs on defined regions of DNA. These primosome assembly sites are on the order of 70 nucleotides in length, yet they share little DNA sequence homology. In order to understand the interaction of the primosomal proteins with these sites, the isolation of single-base substitution mutants of the wild-type sequences has been undertaken. The response of 32 of these mutated primosome assembly sites to increasing concentrations of monovalent and divalent cations when they were used as DNA effectors for E. coli replication factor Y-catalyzed ATP hydrolysis and their efficiency as primosome-dependent DNA replication templates have revealed the existence of four distinct classes of mutations in primosome assembly sites. Class I mutations have essentially no effect on the activities elicited by the DNA site; thus, it is likely that they define nonessential or spacer nucleotide residues. Class II mutated DNAs require higher Mg2+ concentrations than the wild-type DNA to be fully activated as factor Y ATPase effectors and cannot be stimulated in the ATPase reaction by monovalent salt at suboptimal levels of Mg2+. The implication of this mutant phenotype on the role of secondary and tertiary DNA structure in determining an active site is examined in the accompanying article (Soeller, W., Abarzúa, P., and Marians, K. J. (1984) J. Biol. Chem. 259, 14293-14300). Class III mutations coinactivate both the ATPase effector and DNA replication template activity of the site, indicating that they probably represent essential contact points between factor Y and the DNA. Class IV mutated DNAs behave in a manner similar to class II mutated DNAs in the ATPase reaction, but have a replication template activity intermediate between that of the class III and class II mutant DNAs. It is possible that these mutant DNAs are deficient in their ability to catalyze, during primosome assembly, a step subsequent to that of factor Y binding.(ABSTRACT TRUNCATED AT 400 WORDS)     

Identification of the magnesium ion binding site in the catalytic center of Escherichia coli primase by iron cleavage

Magnesium is essential for the catalysis reaction of Escherichia coli primase, the enzyme synthesizing primer RNA chains for initiation of DNA replication. To map the Mg(2+) binding site in the catalytic center of primase, we have employed the iron cleavage method in which the native bound Mg(2+) ions were replaced with Fe(2+) ions and the protein was then cleaved in the vicinity of the metal binding site by adding DTT which generated free hydroxyl radicals from the bound iron. Three Fe(2+) cleavages were generated at sites designated I, II, and III. Adding Mg(2+) or Mn(2+) ions to the reaction strongly inhibited Fe(2+) cleavage; however, adding Ca(2+) or Ba(2+) ions had much less effect. Mapping by chemical cleavage and subsequent site-directed mutagensis demonstrated that three acidic residues, Asp345 and Asp347 of a conserved DPD sequence and Asp269 of a conserved EGYMD sequence, were the amino acid residues that chelated Mg(2+) ions in the catalytic center of primase. Cleavage data suggested that binding to D345 is significantly stronger than to D347 and somewhat stronger than to D269.     

DNA binding to mica correlates with cationic radius: assay by atomic force microscopy.

In buffers containing selected transition metal salts, DNA binds to mica tightly enough to be directly imaged in the buffer in the atomic force microscope (AFM, also known as scanning force microscope). The binding of DNA to mica, as measured by AFM-imaging, is correlated with the radius of the transition metal cation. The transition metal cations that effectively bind DNA to mica are Ni(II), Co(II), and Zn(II), which have ionic radii from 0.69 to 0.74 A. In Mn(II), ionic radius 0.82 A, DNA binds weakly to mica. In Cd(II) and Hg(II), respective ionic radii of 0.97 and 1.1 A, DNA does not bind to mica well enough to be imaged with the AFM. These results may to relate to how large a cation can fit into the cavities above the recessed hydroxyl groups in the mica lattice, although hypotheses based on hydrated ionic radii cannot be ruled out. The dependence of DNA binding on the concentrations of the cations Ni(II), Co(II), or Zn(II) shows maximal DNA binding at approximately 1-mM cation. Mg(II) does not bind DNA tightly enough to mica for AFM imaging. Mg(II) is a Group 2 cation with an ionic radius similar to that of Ni(II). Ni(II), Co(II), and Zn(II) have anomalously high enthalpies of hydration that may relate to their ability to bind DNA to mica. This AFM assay for DNA binding to mica has potential applications for assaying the binding of other polymers to mica and other flat surfaces.     

Metal ion binding sites in a group II intron core

Group II introns are catalytic RNA molecules that require divalent metal ions for folding, substrate binding, and chemical catalysis. Metal ion binding sites in the group II core have now been elucidated by monitoring the site-specific RNA hydrolysis patterns of bound ions such as Tb(3+) and Mg(2+). Major sites are localized near active site elements such as domain 5 and its surrounding tertiary interaction partners. Numerous sites are also observed at intron substructures that are involved in binding and potentially activating the splice sites. These results highlight the locations of specific metal ions that are likely to play a role in ribozyme catalysis.     

Studying metal ion binding properties of a three-way junction RNA by heteronuclear NMR

Self-splicing group II introns are highly structured RNA molecules, containing a characteristic secondary and catalytically active tertiary structure, which is formed only in the presence of Mg(II). Mg(II) initiates the first folding step governed by the κζ element within domain 1 (D1κζ). We recently solved the NMR structure of D1κζ derived from the mitochondrial group II intron ribozyme Sc.ai5γ and demonstrated that Mg(II) is essential for its stabilization. Here, we performed a detailed multinuclear NMR study of metal ion interactions with D1κζ, using Cd(II) and cobalt(III)hexammine to probe inner- and outer-sphere coordination of Mg(II) and thus to better characterize its binding sites. Accordingly, we mapped ¹H, ¹⁵N, ¹³C, and ³¹P spectral changes upon addition of different amounts of the metal ions. Our NMR data reveal a Cd(II)-assisted macrochelate formation at the 5′-end triphosphate, a preferential Cd(II) binding to guanines in a helical context, an electrostatic interaction in the ζ tetraloop receptor and various metal ion interactions in the GAAA tetraloop and κ element. These results together with our recently published data on Mg(II) interaction provide a much better understanding of Mg(II) binding to D1κζ, and reveal how intricate and complex metal ion interactions can be.     

Selective adsorption of apoferritin on immobilized Fe(III): demonstration of Fe(III) binding sites

Immobilized metal ion affinity chromatography has been used to demonstrate and partially characterize Fe(III) binding sites on apoferritin. Binding of Fe(III) to these sites is influenced by pH, but not affected by high ionic strength. These results suggest that both ionic and coordinate covalent interactions are important in the formation of the Fe(III): apoferritin complex. This is, to our knowledge, the first demonstration of direct Fe(III) binding to apoferritin. Other immobilized metal ions, including Zn(II), Ni(II), Cu(II), Cr(III), Co(II), and Tb(III), displayed little or no adsorption of apoferritin. The analytical technique of immobilized metal ion affinity chromatography also shows great promise in the purification of apoferritin, ferritin, and other iron-binding proteins.     

Mapping divalent metal ion binding sites in a group II intron by Mn(2+)- and Zn(2+)-induced site-specific RNA cleavage.

The function of group II introns depends on positively charged divalent metal ions that stabilize the ribozyme structure and may be directly involved in catalysis. We investigated Mn2+- and Zn2+-induced site-specific RNA cleavage to identify metal ions that fit into binding pockets within the structurally conserved bI1 group II intron domains (DI-DVI), which might fulfill essential roles in intron function. Ten cleavage sites were identified in DI, two sites in DIII and two in DVI. All cleavage sites are located in the center or close to single-stranded and flexible RNA structures. Strand scissions mediated by Mn2+/Zn2+ are competed for by Mg2+, indicating the existence of Mg2+ binding pockets in physical proximity to the observed Mn2+-/Zn2+-induced cleavage positions. To distinguish between metal ions with a role in structure stabilization and those that play a more specific and critical role in the catalytic process of intron splicing, we combined structural and functional assays, comparing wild-type precursor and multiple splicing-deficient mutants. We identified six regions with binding pockets for Mg2+ ions presumably playing an important role in bI1 structure stabilization. Remarkably, assays with DI deletions and branch point mutants revealed the existence of one Mg2+ binding pocket near the branching A, which is involved in first-step catalysis. This pocket formation depends on precise interaction between the branching nucleotide and the 5' splice site, but does not require exon-binding site 1/intron binding site 1 interaction. This Mg2+ ion might support the correct placing of the branching A into the 'first-step active site'.     

Insight into the catalytic mechanism of DNA polymerase beta: structures of intermediate complexes

The catalytic reaction mediated by DNA polymerases is known to require two Mg(II) ions, one associated with dNTP binding and the other involved in metal ion catalysis of the chemical step. Here we report a functional intermediate structure of a DNA polymerase with only one metal ion bound, the DNA polymerase beta-DNA template-primer-chromium(III).2'-deoxythymidine 5'-beta,gamma-methylenetriphosphate [Cr(III).dTMPPCP] complex, at 2.6 A resolution. The complex is distinct from the structures of other polymerase-DNA-ddNTP complexes in that the 3'-terminus of the primer has a free hydroxyl group. Hence, this structure represents a fully functional intermediate state. Support for this contention is provided by the observation of turnover in biochemical assays of crystallized protein as well as from the determination that soaking Pol beta crystals with Mn(II) ions leads to formation of the product complex, Pol beta-DNA-Cr(III).PCP, whose structure is also reported. An important feature of both structures is that the fingers subdomain is closed, similar to structures of other ternary complexes in which both metal ion sites are occupied. These results suggest that closing of the fingers subdomain is induced specifically by binding of the metal-dNTP complex prior to binding of the catalytic Mg(2+) ion. This has led us to reevaluate our previous evidence regarding the existence of a rate-limiting conformational change in Pol beta's reaction pathway. The results of stopped-flow studies suggest that there is no detectable rate-limiting conformational change step.     

Distinct roles of two Mg2+ binding sites in regulation of murine flap endonuclease-1 activities

Removal of flap DNA intermediates in DNA replication and repair by flap endonuclease-1 (FEN-1) is essential for mammalian genome integrity. Divalent metal ions, Mg(2+) or Mn(2+), are required for the active center of FEN-1 nucleases. However, it remains unclear as to how Mg(2+) stimulates enzymatic activity. In the present study, we systemically characterize the interaction between Mg(2+) and murine FEN-1 (mFEN-1). We demonstrate that Mg(2+) stimulates mFEN-1 activity at physiological levels but inhibits the activity at concentrations higher than 20 mM. Our data suggest that mFEN-1 exists as a metalloenzyme in physiological conditions and that each enzyme molecule binds two Mg(2+) ions. Binding of Mg(2+) to the M1 binding site coordinated by the D86 residue cluster enhances mFEN-1's capability of substrate binding, while binding of the metal to the M2 binding site coordinated by the D181 residue cluster induces conformational changes. Both of these steps are needed for catalysis. Weak, nonspecific Mg(2+) binding is likely responsible for the enzyme inhibition at high concentrations of the cation. Taken together, our results suggest distinct roles for two Mg(2+) binding sites in the regulation of mFEN-1 nuclease activities in a mode different from the "two-metal mechanism".     

Binding of europium(III) ions to RNA stem loops: role of the primary hydration sphere in complex formation

Understanding the process by which RNA molecules fold into stable structures includes study of the role of site-bound metal ions. Because the alkaline earth metal ions typically associated with RNA structure [most often Mg(II)] do not provide convenient spectroscopic signals, replacement with metal ions having spectroscopically useful properties has been a valuable approach. The luminescence properties of the lanthanide(III) series, in particular europium(III), have made them useful in the study of complexation with biomolecules. We review the physical, chemical, and spectroscopic characteristics of Eu(III) that contribute to its value as a probe of RNA-metal ion interactions, and examples of information obtained from studies of Eu(III) bound to small RNA stem loops. Although Eu(III) has similar site preference to Mg(II), luminescence and isothermal titration calorimetry measurements indicate that Ln(III) loses water molecules from the inner hydration sphere more readily than does Mg(II), resulting in more direct coordination between RNA and the metal ion and very different energetics of binding. In some cases, e.g., a GAAA tetraloop, binding appears to occur by a lock and key process; in the same base sequence containing certain deoxynucleoside substitutions that alter loop structure, binding appears to occur by an induced fit process.