The solution structure of d3'-EBS1*

The structure determination of d3'-EBS1* was performed by including 514 conformationally restrictive NOE distance restraints collected from 2D [1H,1H]- NOESY spectra in 100% D2O as well as in 90%

H2O/10% D2O and 28 RDC constraints (Table 9,

appendices 6 and 7, for calculation details see Materials and Methods). d3'-EBS1* adopts a stable hairpin structure closed by the 11 nucleotide long loop including EBS1*, which is in the absence of IBS1* rather flexible. The relatively low number of NOEs per residue of 17.72

Figure 47 Lowest energy structure of d3'-

EBS1* out of 200 calculated.

Table 9 NMR restraints and structural statistics for the d3'-EBS1* structure.a

With RDCs Without RDCs NOE-derived distance restraints

Intranucleotide

Internucleotide (|i - j| = 1) Long-range (|i – n| ≥ 2) Repulsive

NOE restraints per residue Total

Helix (1-9, 21-29) Loop (10-20) Dihedral restraints Hydrogen bond restraints Dipolar coupling restraints

514 178 252 84 0 17.72 20.17 13.72 180 45 28 514 178 252 84 0 17.72 20.17 13.72 180 45 0 r.m.s.d. (for all heavy atoms to the best structure (Å))

Overall Helix (1-9, 21-29) Loop (10-20) 2.06 ± 0.86 0.41 ± 0.10 2.59 ± 1.23 2.50 ± 0.65 0.81 ± 0.35 2.73 ± 1.03 NOE violations > 0.2 Å Dihedral violations > 5° 0 0 0 0 a

Results and Discussion 67

derives from the unstructured loop region, for which only 13.72 NOEs per residue were observed in comparison to the helix where 20.16 were found. As already shown by the base pairing situation (see Section 2.2.5), the loop does not adopt a rigid structure, thus being rather flexible and free to bind IBS1*. The overall r.m.s. deviation of all heavy atoms from the 20 lowest energy structures with the inclusion of RDCs is with 2.06 ± 0.86 Å rather low compared to the r.m.s. deviation for the 20 lowest energy structures without RDCs with 2.50 ± 0.65 Å (Table 9, Figure 48). However, independent superposition of the helical region (nucleotides 1-9 and 21-29) results in much lower r.m.s. deviations in both cases, whereas the independent superposition of the loop (nucleotides 10-20) gives much higher r.m.s. deviations (Table 9, Figure 48B, C, E, and F, and Figure 48).

The loop is the least well defined part of d3'-EBS1*. The reasons for the relatively poor definition of the loop are twofold. First, as already mentioned, distance constraints for the amino and imino protons could not be obtained due to the fast exchange rate of these protons with the solvent (see Section 2.2.5). Second, the majority of the assigned NOEs to the loop

Figure 48 Solution structure of d3'-EBS1* as determined by NMR. The upper panel shows the 20 lowest

energy structures of d3'-EBS1* with residual dipolar couplings, the lower panel the ones without residual dipolar couplings. (A and D) Overall superposition of all heavy atoms in d3'-EBS1*. (B and E) Superposition of all heavy atoms in the stem (nucleotides 1-9, and 21-29) of d3'-EBS1* of the 20 lowest energy structures. (C and F) Superposition of all heavy atoms in the loop (nucleotides 10-20) of d3'-EBS1* of the 20 lowest energy structures.

are either intranucleotide or between sequential nucleotides, and therefore are not useful in defining tertiary structural features.

However, apart from the overall unpaired situation in the loop, some special structural features can be discovered in the loop within the 20 lowest energy structures. The stem leading to the loop is closed by the U9-G21 wobble pair. The bases of A10 and A20, being the first and the last nucleotide of the loop, are pointing towards each other on the inside of the loop (Figure 49), but they are not completely buried as it has been already shown by the chemical shift pattern of H2 resonances described in Section 2.2.5. A10 remains partly stacked onto U9 defined by five NOE cross peaks associated with non- exchangeable protons as well as A20 onto G21, which is defined by seven NOEs. In the 20 lowest energy structures these stacking interactions are pronounced differently illustrating the loop flexibility. The arrangement of A10 and A20 is also confirmed by the observation of crosspeaks between A10H2 and A20H2 as well as A10H2 and G21H1' in the 2D [1H,1H]- NOESY spectrum in 100% D2O (see Appendix 6). All other nucleotides in the loop show no

specific arrangements. The bases belonging to EBS1* are in majority protruded being therefore solvent exposed. This also explains the absence of observable imino proton resonances for the loop region, because they are in fast exchange with the solvent and thus not detectable on the NMR timescale. These structural observations have immediate biological implications. The major function of the EBS1* is to recognize the substrate, thus building tertiary contacts. Therefore the loop is in a favoured position to bind IBS1* when it comprises a certain flexibility to expose the bases to the outside.

The only known structure of a full group II intron ribozyme and thus also of the hairpin including the exon binding site 1 was just published recently.(249) The crystal structure of an intact, self-spliced group II intron from Oceanobacillus iheyensis was solved at 3.1 Å resolution (see also Introduction Section 1.3.2.). The exon binding site in this system comprises only four nucleotides (AUAA) and is located in a loop of eight nucleotides (Figure 50). The stem of the hairpin consists only of four base pairs starting with a GC base pair and followed by three AU base pairs. The last four nucleotides in the loop, of which two belong to EBS1, show partial stacking interactions. In contrast to this the first four nucleotides are completely unstructured, of which U182 is missing due to lacking electron density. A

Figure 49 Section of d3'-EBS1* of the lowest

energy structure showing the first and the last nucleotide of the loop A10 and A20 pointing into the inside of the loop, a possible hydrogen bond is visualized.

Results and Discussion 69

comparison of the crystal structure with d3'-EBS1* is shown in Figure 50. Both structures show a stable stem but for the loop only some partial stacking interactions. Unfortunately no conclusion about the relevance of the partial stacking can be drawn from this structures as they differ from each other not only in the stacking situation of the loop but also in their size.