Recently, we found a small Ca(2+)-dependent deoxyribozyme (unmodified), d(GCCTGGCAG(1)G(2)C(3)T(4)A(5)C(6)A(7)A(8)C(9)G(10)A(11)GTCCCT), with cleavage activity for its RNA substrate, r(AGGGACA downward arrow UGCCAGGC) ( downward arrow denotes the RNA cleavage site), in the presence of Ca(2+) and developed a functional SPR sensor chip with this deoxyribozyme [Okumoto, Y., Ohmichi, T., and Sugimoto, N. (2002) Biochemistry 41, 2769-2773]. In the study presented here, to clarify the factors contributing to the efficient catalytic activity of the unmodified deoxyribozyme, RNA cleavage reactions were carried out using 24 mutant deoxyribozymes containing one unnatural DNA nucleotide, such as dI (2'-deoxyinosine), 7-deaza-dG, 2-aminopurine, 7-deaza-dA, 2-amino-dA, dm(5)C (5-methyl-2'-deoxycytosine), or d(P)C (5-propynyl-2'-deoxycytosine). The K(m) values (Michaelis constants) with the mutants that lacked N7 and O6 of G(1) and O6 of G(2) were 4.5 and 6.6 times that of the unmodified one, respectively. The k(cat) value (cleavage rate constant) with the mutants that lacked O6 of G(10) was 0.025 times that of the unmodified one. The results of UV melting curves, SPR kinetics, and CD spectra supported the quantitative idea that the catalytic activity of the unmodified form was achieved using Ca(2+). On the basis of these results, a preliminary model for two G(1) x A(8) and G(2) x A(7) mismatched base pairs such as G(anti) x A(anti) formed in the catalytic loop is proposed. The factor of 10 increase in the k(cat)/K(m) value of the mutant deoxyribozyme, which has C(9) substituted with d(P)C, suggests that the base stacking interaction between the substituted propynyl group in dC and the nearest-neighbor base grew stronger. Thus, substituting d(P)C for dC in the catalytic loop would be one of the best ways to increase the catalytic activity of the deoxyribozyme.

Two tiny hairpin DNAs, CORE (dAGGCTTCGGCCT) and AP2 (dAGGCTXCGGCCT; X: abasic nucleotide), fold into almost the same tetraloop hairpin structure with one exception, that is, the sixth thymine (T6) of CORE is exposed to the solvent water (Kawakami, J. et al., Chem. Lett. 2001, 258-259). In the present study, we selected small peptides that bind to CORE or AP2 from a combinatorial pentapeptide library with 2.5 x 10(6) variants. On the basis of the structural information, the selected peptide sequences should indicate the essential qualifications for recognition of the hairpin loop DNA with and without a flipped base. In the selected DNA binding peptides, aromatic amino acids such as histidine for CORE and glutamine/aspartic acid for AP2 were found to be abundant amino acids. This amino acid preference suggests that CORE-binding peptides use pi-pi stacking to recognize the target while hydrogen bonding is dominant for AP2-binding peptides. To investigate the binding properties of the selected peptide to the target, surface plasmon resonance was used. The binding constant of the interaction between CORE and a CORE-binding peptide (HWHHE) was about 1.1 x 10(6) M(-1) at 25 degrees C and the resulting binding free energy change at 25 degrees C (DeltaG degrees (25)) was -8.2 kcal mol(-1). The binding of the peptide to AP2 was also analyzed and the resulting binding constant and DeltaG degrees (25) were about 4.2 x 10(4) M(-1) and -6.3 kcal mol(-1), respectively. The difference in the binding free energy changes (DeltaDeltaG degrees (25)) of 1.9 kcal mol(-1) was comparable to the values reported in other systems and was considered a consequence of the loss of pi-pi stacking. Moreover, the stabilization effect by stacking affected the dissociation step as well as the association step. Our results suggest that the existence of an aromatic ring (T6 base) produces new dominant interactions between peptides and nucleic acids, although hydrogen bonding is the preferable mode of interaction in the absence of the flipping base. These findings regarding CORE and AP2 recognition are expected to give useful information in the design of novel artificial DNA binding peptides.

To achieve a novel specific peptide-nucleic acid binding model, we designed an in vitro selection procedure to decrease the energetic contribution of the electrostatic interaction in the total binding energy and to increase the contribution of hydrogen bonding and pi-pi stacking. After the selection of hairpin-loop RNAs that specifically bound to a model peptide of lambda N protein (N peptide), a new thermostable pentaloop RNA motif (N binding thermostable RNA hairpin: NTS RNA) was revealed. The obtained NTS RNA was able to bind to the N peptide with superior specificity to the boxB RNA, which is the naturally occurring partner of the lambda N protein.

Artificial binding model systems with sufficient specificity are indispensable for study and understanding of living systems. Here, we report a small hairpinloop RNA that interacts with a basic peptide derived from bacteriophage lambda N protein, as a new protein-RNA binding model. The RNA was selected in vitro from a hairpinloop RNA library and recognized the target N peptide more specifically than the boxB RNA, the naturally occurring N protein binder.