Remarkable stability of hairpins containing 2',5'-linked RNA loops.

We report here the results of a comparative study of hairpin loops that differ in the connectivity of phosphodiester linkages (3',5'- versus 2',5'-linkages). In addition, we have studied the effect of changing the stem composition on the thermodynamic stability of hairpin loops. Specifically, we constructed hairpins containing one of six stem duplex combinations, i.e., DNA:DNA ("DD"), RNA:RNA ("RR"), DNA:RNA ("DR"), 2',5'-RNA:RNA ("RR"), 2',5'-RNA:DNA ("RD"), and 2',5'-RNA:2',5'-RNA ("RR"), and one of three tetraloop compositions, i.e., 2',5'-RNA ("R"), RNA ("R"), and DNA ("D"). All hairpins contained the conserved and well-studied loop sequence 5'-...C(UUCG)G...-3' [Cheong et al. Nature 1990, 346, 680-682]. We show that the 2',5'-linked loop C(UUCG)G, i.e.,...C(3'p5')U(2'p5')U(2'p5')C(2'p5')G(2'p5')G(3'p5')..., like its "normal" RNA counterpart, forms an unusually stable tetraloop structure. We also show that the stability imparted by 2',5'-RNA loops is dependent on base sequence, a property that is shared with the regioisomeric 3',5'-RNA loops. Remarkably, we find that the stability of the UUCG tetraloop is virtually independent of the hairpin stem composition (DD, RR, RR, etc.), whereas the native RNA tetraloop exerts extra stability only when the stem is duplex RNA (R:R). As a result, the relative stabilities of hairpins with a 2',5'-linked tetraloop, e.g. ggac(UUCG)gtcc (T(m) = 61.4 degrees C), are often superior to those with RNA tetraloops, e.g. ggac(UUCG)gtcc (T(m) = 54.6 degrees C). In fact, it has been possible to observe the formation of a 2',5'-RNA:DNA hybrid duplex by linking the hybrid's strands to a (UUCG) loop. These duplexes (RD), which are not stable enough to form in an intermolecular complex [Wasner et al. Biochemistry 1998, 37, 7478-7486], were stable at room temperature (T(m) approximately 50 degrees C). Thus, 2',5'-loops have potentially important implications in the study of nucleic acid complexes where structural data are not yet available. Furthermore, they may be particularly useful as structural motifs for synthetic ribozymes and nucleic acid "aptamers".     

Molecular dynamics simulation of the structure, dynamics, and thermostability of the RNA hairpins uCACGg and cUUCGg

Classical replica-exchange molecular dynamics simulations are performed to study structure, dynamics and thermostability of the 14-mer RNA hairpins uCACGg and cUUCGg. Despite of the different sequence and closing base pair of the two systems, recent NMR studies have shown that the tetraloop CACG is strikingly similar in overall geometry and hydrogen bonding to the canonical UUCG tetraloop. On the other hand, the two systems differ significantly in their functionality and thermostability. The simulations confirm the structural similarities of the two RNA hairpins at room temperature but also reveal that the UUCG loop is more flexible than the CACG loop. Concerning the functionality, the CACG loop shows a stronger attitude to donate hydrogens than the UUCG loop, although their global solvent accessible surface is quite similar. The simulations qualitatively reproduce the experimentally found difference in melting temperatures (20 K). In the case of the uCACGg hairpin, the thermal unfolding occurs cooperatively in an all-or-none fashion, while the cUUCGg hairpin shows less cooperativity but exhibits intermediate states during the unfolding process.     

Probing RNA hairpins with cobalt(III)hexammine and electrospray ionization mass spectrometry

In this work, electrospray ionization mass spectrometry (ESI MS) was employed to study the interactions of cobalt(III) hexammine, Co(NH3)6(3+), with five RNA hairpins representing the 790 loop of 16S ribosomal RNA and 1920 loop of 23S ribosomal RNA. The RNAs varied in mismatch identity (G.U versus A.C) and level of base modification (pseudouridine versus uridine). Co(NH3)6(3+) binding was observed with the four RNA hairpins that contained a G.U wobble pair in the stem region. ESI MS revealed 1:1 and 1:2 complex formation with all RNAs. Weaker binding was observed with the fifth RNA hairpin that contained an A.C wobble pair in the stem region. The effects of pH on Co(NH3)6(3+) binding were also examined.     

Restricting the conformational heterogeneity of RNA by specific incorporation of 8-bromoguanosine

In an effort to reduce the conformational heterogeneity of RNA, the modified nucleobase 8-bromoguanosine (8BrG) was introduced into oligonucleotides having the hairpin tetraloop motif YNMG (Y = U or C and M = C or A). Purine nucleobases with bromine at position eight are known to preferentially adopt the syn conformation as nucleosides. The hairpin tetraloop motif YNMG was chosen as a model system because it has a syn guanosine at position four of the loop that is essential for thermodynamic stability. Thermodynamic and structural characterization of modified oligonucleotides with the hairpin sequences UUCG, CGCG, and CGAG by UV-melting and NMR spectroscopy revealed that 8BrG substitution has a small effect upon the hairpin conformation, while the duplex conformation is strongly destabilized (DeltaDeltaG degrees 37 approximately +4.7 kcal mol-1), thus inhibiting dimerization. These results support a model in which 8BrG substitution shifts the hairpin-duplex equilibrium constant toward the hairpin conformation by destabilizing the duplex. This methodology should be useful for limiting conformational heterogeneity in large RNAs, with potential applications in structural biology and enzymology.     

Measurement of the effect of monovalent cations on RNA hairpin stability

Using optical tweezers, we have measured the effect of monovalent cation concentration and species on the folding free energy of five large (49-124 nt) RNA hairpins, including HIV-1 TAR and molecules approximating A.U and G.C homopolymers. RNA secondary structure thermodynamics are accurately described by a model consisting of nearest-neighbor interactions and additive loop and bulge terms. Melting of small (<15 bp) duplexes and hairpins in 1 M NaCl has been used to determine the parameters of this model, which is now used extensively to predict structure and folding dynamics. Few systematic measurements have been made in other ionic conditions or for larger structures. By applying mechanical force, we measured the work required to fold and unfold single hairpins at room temperature over a range of cation concentrations from 50 to 1000 mM. Free energies were then determined using the Crooks fluctuation theorem. We observed the following: (1) In most cases, the nearest-neighbor model accurately predicted the free energy of folding at 1 M NaCl. (2) Free energy was proportional to the logarithm of salt concentration. (3) Substituting potassium ions for sodium slightly decreased hairpin stability. The TAR hairpin also misfolded nearly twice as often in KCl, indicating a differential kinetic response. (4) Monovalent cation concentration affects RNA stability in a sequence-dependent manner. G.C helices were unaffected by changing salt concentration, A.U helices were modestly affected, and the hairpin loop was very sensitive. Surprisingly, the U.C.U bulge of TAR was found to be equally stable in all conditions tested. We also report a new estimate for the elastic parameters of single-stranded RNA.     

Synthesis,structure, and photochemistry of exceptionally stable synthetic DNA hairpins with stilbene diether linkers

The structure and properties of 18 hairpin-forming bis(oligonucleotide) conjugates possessing stilbene diether linkers are reported. Conjugates possessing bis(2-hydroxyethyl)stilbene 4,4'-diether linkers form the most stable DNA hairpins reported to date. Hairpins with as few as two T:A base pairs or four noncanonical G:G base pairs are stable at room temperature. Increasing the length of the hydroxyalkyl groups results in a decrease in hairpin thermal stability. On the basis of the investigation of their circular dichroism spectra, all of the hairpins investigated adopt B-DNA structures, except for a hairpin with a short poly(G:C) stem which forms a Z-DNA structure. Both the strong fluorescence of the stilbene diether linkers and their trans-cis photoisomerization are totally quenched in hairpins possessing neighboring T:A and G:C base pairs. Quenching is attributed to an electron-transfer mechanism in which the singlet stilbene serves as an electron donor and T or C serves as an electron acceptor. In contrast, in denatured hairpins and hairpins possessing neighboring G:G base pairs the stilbene diether linkers undergo efficient photoisomerization.     

Thermodynamic coupling of the loop and stem in unusually stable DNA hairpins closed by CG base pairs

For certain DNA hairpin loops, a CG closing base pair has enhanced stability over other closing base pairs, which cannot be explained by the current nearest-neighbor model. We report the use of three-carbon (C3) spacers to investigate the expandability of DNA hairpin loops and the coupling between the loop and closing base pair. Inserting the C3-spacers at most positions in these model loops produced only a modest stabilization or destabilization except for insertion between the 5' end of the loop and the CG closing base pair, which gave a large destabilization. Further investigation on tetraloops and triloops with other closing base pairs established that this destabilization is specific to the unusually stable CG closing base pair. Studies with the nucleotide analogues 2-aminopurine and 2,6-diaminopurine indicated that this stabilization may be due to coupling between functional groups on the first base of the loop and the CG closing base pair. The C3-spacers provide a simple way to interrupt potential interactions and thereby probe loop/stem coupling.     

Template-directed synthesis using the heterogeneous templates produced by montmorillonite catalysis. A possible bridge between the prebiotic and RNA worlds

The synthesis of oligoguanylates [oligo(G)s] is catalyzed by a template of oligocytidylates [oligo(C)s] containing 2',5'- and 3',5'-linked phosphodiester bonds with and without incorporated C5'ppC groupings. An oligo(C) template containing exclusively 2',5'-phosphodiester bonds also serves as a template for the synthesis of complementary oligo(G)s. The oligo(C) template was prepared by the condensation of the 5'-phosphorimidazolide of cytidine on montmorillonite clay. These studies establish that RNA oligomers prepared by mineral catalysis, or other routes on the primitive earth, did not have to be exclusively 3',5'-linked to catalyze template-directed synthesis, since oligo(C)s containing a variety of linkage isomers serve as templates for the formation of complementary oligo(G)s. These findings support the postulate that origin of the RNA world was initiated by the RNA oligomers produced by polymerization of activated monomers formed by prebiotic processes.     

Face-to-face and edge-to-face pi-pi interactions in a synthetic DNA hairpin with a stilbenediether linker

Synthetic conjugates possessing bis(2-hydroxyethyl)stilbene-4,4'-diether linkers (Sd2) form the most stable DNA hairpins reported to date. Factors that affect stability are length and flexibility of the linkers and pi-stacking of the stilbene moiety on the adjacent base pair. The crystal structure of the hairpin d(GT(4)G)-Sd2-d(CA(4)C) was determined at 1.5 A resolution. The conformations of the two molecules in the asymmetric unit differ both in the linker and the stem portions. One of them shows a planar stilbene that is stacked on the adjacent G:C base pair. The other displays considerable rotation between the phenyl rings and an unprecedented edge-to-face orientation of stilbene and base pair. The observation of considerable variations in the conformation of the Sd moiety in the crystal structure allows us to exclude restriction of motion as the reason for the absence of Sd photoisomerization in the hairpins. Conformational differences in the stem portion of the two hairpin molecules go along with different Mg(2+) binding modes. Most remarkable among them is the sequence-specific coordination of a metal ion in the narrow A-tract minor groove. The crystal structure provides unequivocal evidence that a fully hydrated Mg(2+) ion can penetrate the narrow A-tract minor groove, causing the groove to further contract. Overall, the structural data provide a better understanding of the origins of hairpin stability and their photochemical behavior in solution.     

Sequestering of Eu(III) by a GAAA RNA tetraloop

The site-specific binding of metal ions maintains an important role in the structure, thermal stability, and function of folded RNA structures. RNA tetraloops of the "GNRA" family (where N = any base and R = any purine), which owe their unusual stability to base stacking and an extensive hydrogen bonding network, have been observed to bind metal ions having different chemical and geometric properties. We have used laser-induced lanthanide luminescence and isothermal titration calorimetry (ITC) to examine the metal-binding properties of an RNA stem loop of the GNRA family. Previous research has shown that a single Eu(III) ion binds the stem loop fragment in a highly dehydrated site with a K(d) of approximately 12 microM. Curve-fitting analysis of the broad luminescence excitation spectrum of Eu(III) upon complexation with the tetraloop fragment indicates the possibility of two microenvironments that do not differ in hydration number. Binding of Eu(III) to the loop was accompanied by positive enthalpic changes, consistent with energetic cost of removal of water molecules and suggesting that the binding is entropically driven. By comparison, binding of Mg(II) or Mn(II) to the RNA loop, or Eu(III) to the DNA analogue of the loop, was associated with exothermic changes, consistent with predominantly outer-sphere coordination. These results suggest specific binding, most probably involving ligands on the 5' side of the loop.     

Spectroscopic and structural impact of a stem base-pair change in DNA hairpins: GTTC-ACA-GAAC versus GTAC-ACA-GTAC

Successive investigations over the last decade have revealed and confirmed a stable loop closure in a family of d-[GTAC-5Pur6N7N-GTAC] hairpins, where 5Pur6N7N is a AAA, GAG and AXC loop (X being any nucleotide). The trinucleotide loop is characterized by a well defined 5Pur-7N mispairing mode, and by upfield chemical shifts for three sugar protons of the apical nucleotide 6N. The GTTC-ACA-GAAC DNA hairpin, of interest for its likely involvement in Vibrio cholerae genome mutations, has now been investigated. The GTAC-ACA-GTAC DNA hairpin has also been studied because it is intermediate between the other structures, as it contains the loop of the hairpin under consideration and the stem of the above family. The two hairpins with the ACA loop are stable. They show the same mispairing mode and similar upfield shifts as the previous family, but GTTC-ACA-GAAC seems to be slightly less compact than any other. GTTC-ACA-GAAC is remarkable in that it exhibits a B(II) character for the phosphate-ester conformation at 8Gp9A, together with a swing of the upper hairpin into the major groove that, in particular, brings 6CH1' roughly as close to 7AH2 as to 6CH6. These unexpected structural features are qualitatively deduced from (1)H and (31)P NMR spectra, and confirmed by Raman spectroscopy. This comparative study shows that not only the loop sequence but also the stem sequence may control hairpin structures.     

One-step, regioselective synthesis of up to 50-mers of RNA oligomers by montmorillonite catalysis

5'-Nucleotides of A and U with the phosphate activated with 1-methyladenine generate RNA oligomers containing 40-50 monomers in 1 day in reactions catalyzed by montmorillonite. The corresponding monomers of C give oligomers that are 20-25-mers in length after a 9-day reaction. It was not possible to determine the chain lengths of the oligomers of G since they did not give well-defined bands on gel electrophoresis. Co-oligomers of A and U as well as A, U, G, and C were also prepared. The oligo(A)s formed were separated by gel electrophoresis, and the bands of the 7-39-mers were isolated, the 3',5'-phosphodiester bonds were cleaved by RNase T(2), and the terminal phosphate groups were cleaved with alkaline phosphatase. HPLC analysis revealed that the proportions of A(5)'pp(5)'A, A, A(2)'pA, and A(2)'pA(2)'pA formed were almost the same for the long and shorter oligomers. A similar structure analysis performed on the oligo(U)s established that the proportions of U(5)'pp(5)'U, U, U(2)'pU, U(2)'pU(2)'pU, U(2)'pU(2)'pU(2)'pU, and U(2)'pU(2)'pU(2)'pU(2)'pU did not vary with chain length. The structural analysis of the oligomers of A revealed that 74% of the phosphodiester bonds were 3',5'-linked a value slightly greater than 67% observed when imidazole was the activating group. 61% of the bonds in the U oligomers were 3',5'-linked, which is almost 3 times greater than the 20% measured when imidazole was the activating group. The potential significance of these data to the origin and early evolution of life is discussed.     

Folding of a stable DNA motif involves a highly cooperative network of interactions

Hairpins are structural elements that play important roles in the folding and function of RNA and DNA. The extent of cooperativity in folding is an important aspect of the RNA folding problem. We reasoned that an investigation into the origin of cooperativity might be best carried out on a stable nucleic acid system with a limited number of interactions, such as a stable DNA hairpin loop. The stable d(cGNAg) hairpin loop motif (closing base pair in lower case; loop in upper case; N = A, C, G, or T) is stabilized through only three interactions: two loop-loop hydrogen bonds in a sheared GA base pair and a loop-closing base pair interaction. Herein, we investigate this network of interactions and test whether the loop-loop and loop-closing base pair interactions communicate. Thermodynamic measurements of nucleotide analogue substituted oligonucleotides were used to probe the additivity of the interactions. On the basis of double mutant cycles, all interactions were found to be nonadditive and interdependent, suggesting that loop-loop and loop-closing base pair interactions form in a highly cooperative manner. When double mutant cycles were repeated in the absence of the other interaction, nonadditivity was significantly reduced suggesting that coupling is indirect and requires all three interactions in order to be optimal. A cooperative network of interactions helps explain the structural and energetic bases of stability in certain DNA hairpins and paves the way for similar studies in more complex nucleic acid systems.     

Structure and electronic spectra of DNA mini-hairpins with G(n):C(n) stems

The solution structure of a synthetic DNA mini-hairpin possessing a stilbenediether linker and three G:C base pairs has been obtained using (1)H NMR spectral data and constrained torsion angle molecular dynamics. Notable features of this structure include a compact hairpin loop having a short stilbene-guanine plane-to-plane distance and approximate B-DNA geometry for the three base pairs. Comparison of the electronic spectra of mini-hairpins having one-to-four G:C base pairs and stilbenediether or hexamethyleneglycol linkers reveals the presence of features in the UV and CD spectra of the stilbene-linked hairpins that are not observed for the ethyleneglycol-linked hairpins. Investigation of the electronic structure of a stilbene-linked hairpin having a single G:C base pair by means of time-dependent density functional theory shows that the highest occupied molecular orbital, but not the lowest unoccupied molecular orbital, is delocalized over the stilbene and adjacent guanine. The calculated UV and CD spectra are highly dependent upon hairpin conformation, but reproduce the major features of the experimental spectra. These results illustrate the utility of an integrated experimental and theoretical approach to understanding the complex electronic spectra of pi-stacked chromophores.     

Exploring the energy landscape of a small RNA hairpin

The energy landscape of a small RNA tetraloop hairpin is explored by temperature jump kinetics and base-substitution. The folding kinetics are single-exponential near the folding transition midpoint T(m). An additional fast phase appears below the midpoint, and an additional slow phase appears above the midpoint. Stem mutation affects the high-temperature phase, while loop mutation affects the low-temperature phase. An adjusted 2-D lattice model reproduces the temperature-dependent phases, although it oversimplifies the structural interpretation. A four-state free energy landscape model is generated based on the lattice model. This model explains the thermodynamics and multiphase kinetics over the full temperature range of the experiments. An analysis of three variants shows that one of the intermediate RNA structures is a stacking-related trap affected by stem but not loop modification, while the other is an early intermediate that forms some stem and loop structure. Even a very fast-folding 8-mer RNA with an ideal tetraloop sequence has a rugged energy landscape, ideal for testing analytical and computational models.     

Phenanthrene‐Derived DNA Hairpin Mimics

Self‐complementary oligodeoxynucleotides containing 3,6‐disubstituted phenanthrenes adopt highly stable, hairpin‐like structures. The thermodynamic stability of the hairpin mimics depends on the overall length of the phenanthrene building block. Hairpin loops composed of a phenanthrene‐3,6‐dicarboxamide and ethylene linkers were found to be optimal. The hairpin mimics are more stable than the analogous hairpins containing either a dT₄ or dA₄ tetraloop. Model studies indicate that the thermodynamic stability of the hairpin mimics is primarily due to aromatic stacking of the phenanthrene‐3,6‐dicarboxamide onto the adjoining base pair of the DNA duplex.     

Design of an adenosine analogue that selectively improves the affinity of a mutant U1A protein for RNA

The RNA recognition motif (RRM), one of the most common RNA binding domains, contains three highly conserved aromatic amino acids that participate in stacking interactions with RNA bases. We have investigated the contribution of these highly conserved aromatic amino acids to the affinity of the complex formed between the N-terminal RRM of the U1A protein and stem loop 2 of U1 snRNA. Previously, we found that substitution of one of these conserved aromatic amino acids, Phe56, with Ala resulted in a large destabilization of the complex. Here, we have modified A6, the base in stem loop 2 RNA that stacks with Phe56, to compensate for a portion of the destabilization caused by the Phe56Ala mutation. We have designed two modified adenosines, A-3CPh and A-4CPh, in which a phenyl group is linked to the adenosine such that it may replace the phenyl group that is eliminated by the Phe56Ala mutation in the complex. We have found that incorporation of A-3CPh into stem loop 2 RNA stabilizes the complex formed with Phe56Ala by 0.6 kcal/mol, while incorporation of A-4CPh into stem loop 2 RNA stabilizes this complex by 1.8 kcal/mol. Either base modification destabilizes the wild-type complex by 0.8-0.9 kcal/mol. Experiments with other U1A mutant proteins suggest that the stabilization of the complex between the Phe56Ala U1A protein and stem loop 2 RNA is due to a specific interaction between the Phe56Ala U1A protein and A6-4CPh stem loop 2 RNA.     

General combinatorics of RNA hairpins and cloverleaves

For an abstract single-strand RNA, a combinatorial analysis is given for two important structures, hairpins and cloverleaves. The total number of hairpins and cloverleaves of a given length with minimal hairpin loop length m(m > 0) and with minimal stack length l(l > 0) is computed, under the assumption that all base pairs can occur.     

NMR studies of DNA single strands and DNA:RNA hybrids with and without 1-propynylation at C5 of oligopyrimidines

The 1-propynylation at C5 of consecutive pyrimidines in DNA can enhance DNA:RNA hybrid stability at 37 degrees C by over 1 kcal/mol of substitution [Barnes, T. W., III; Turner, D. H. J. Am. Chem. Soc.2001, 123, 4107-4118]. To provide information on the structural consequences of propynylation, two-dimensional NMR spectroscopy was used to study the structures of several oligonucleotides. Intraresidue nuclear Overhauser effect spectroscopy cross peaks were observed at 30 degrees C and a 200 ms mixing time in the H6-H1' region for 5'(dC(P)C(P)U(P)C(P)C(P)U(P)U(P)) (ssPrODN) but not for 5'(dCCUCCUU) (ssODN), suggesting preorganization of the propynylated single strand. NMR structures of the duplexes 5'(dC(P)C(P)U(P)C(P)C(P)U(P)U(P))3':3'(rGAGGAGGAAAU)5' (PrODN:RNA), 5'(dCC(P)U(P)C(P)C(P)U(P)U(P))3':3'(rGAGGAGGAAAU)5' (sPrODN1:RNA), and 5'(dCCUCCUU)3':3'(rGAGGAGGAAAU)5' (ODN:RNA) indicate that their global structures are almost identical. The NMR data, however, suggest that the 5'-end of sPrODN1:RNA is more dynamic than that of PrODN:RNA. In the propynylated duplexes, the propyne group stacks on the aromatic ring of the 5'-base and extends into the major groove. The results suggest that the increased stability of the propynylated duplexes is caused by preorganization of the propynylated single strand and different interactions in the double strand. The propynyl group provides volume exclusion, enhanced stacking, and possibly different solvation.     

Chemical Approach for the Study of the ‘Kissing Complex’ of Moloney murine leukaemia Virus

The replication of Moloney murine leukaemia virus relies on the formation of a stable homodimeric ‘kissing complex’ of a GACG tetraloop interacting through only two C?G base pairs flanked of 5′‐adjacent unpaired adenosines A9. Previous NMR investigations of a model stem loop 1 has not permitted to reveal the origin of this interaction. Therefore, with the aim of deeper comprehension of the phenomena, the model sequence 10 was prepared where position 9 has been substituted for a nucleoside offering a wider π‐stacking. In this context, the wyosine phosphoramidite building block 2 was prepared and incorporated by adapting the conditions of the automated synthesis and developing original templated enzymatic ligation. However, no ‘kissing interaction’ has been observed for this model sequence 10 due to steric hindrance as confirmed by computational simulation. Consequently, several other model sequences, 18, 23 – 26 , containing modified nucleosides were prepared. Finally, the importance of the cross‐loop H‐bond between G8 and G11 nucleobases was revealed by preparing a 18mer RNA hairpin 27 , where the guanosine G8 has been substituted for inosine. The latter, which does not possess a C³ amino function compared to guanosine, is unable to form any ‘kissing complex’ demonstrating the importance of this secondary interaction in the formation of the complex.