Figure 5. Representative pictures of crystals of SAM-II and preQ1 riboswitch-ligand complexes.
Figure 6. Representative diffraction images of preQ1 riboswitch-ligand crystals showing the improving diffraction limits as optimization of conditions continues. Arrows indicate diffraction limits. (a, b) Diffraction images of preQ1d crystals. (c) Diffraction image of crystal generated from a complex of methylated preQ1c and its ligand.
Each macromolecule, however, requires unique crystallization solutions that are unpredictable, thereby necessitating large scale trial-and-error for crystallization. For initial screening of crystallization conditions, riboswitch-ligand complexes with different concentrations were tested at two temperatures and with different crystallization solutions from commercial kits. Approximately 600-1,800 different conditions for each riboswitch-ligand complex were tested for crystal formation, summing up to over 6,000 different conditions. Initial crystallization hits were found for all tested complexes: 24 for the SAM-II and 102 for the preQ1 constructs. Most of the hits were crystalline-like material; nevertheless, promising crystals were found for all cases (Figure 5). Crystallization of the preQ1a-ligand complex resulted in the highest success rate with 3.8 % of tested conditions producing crystals, while other riboswitch-ligand
complexes had a success rate of 1-2 %. Optimization of crystallization conditions is necessary to grow larger and better diffracting crystals. Careful optimization of initial hit crystal conditions, such as varying the reagent concentrations and pH, yielded eleven conditions with moderate quality crystals: one each for SAM-IIa, SAM-IIb, and preQ1a, and four each for the preQ1c and preQ1d riboswitches. The best crystals generated from these optimizations showed the following diffraction limits: SAM-IIb, 8.0 Å; preQ1a, 6.6 Å; preQ1c, 5.8 Å; and preQ1d, 6.5 Å (Fig. 6a, b as examples of diffraction images). Simultaneously during the traditional optimization of crystal conditions, an alternative approach was pursued for the preQ1c-ligand complex, the most promising construct for the generation of high quality crystals. During chemical synthesis of preQ1c RNA, a hydrogen atom in the 2´-hydroxyl group was replaced with a methyl group in one of the nucleotides of
Figure 7. Modified uridine monophosphate used for crystallization of preQ1c. Green shading shows 2’-O-methylation Potential Se substitution is indicated in red.
the helix (Figure 7). It was thought that a single methyl group would not affect helix formation, but would introduce a hydrophobic patch in the RNA backbone and could promote formation of novel packing interactions during crystallization. Four variants of the preQ1c RNA were prepared with single substitutions at nucleotides A-3, U-4, A-19, or U-20 (Figure 3e). However, crystals for the modified RNA-ligand complexes were not reproducible in previously optimized conditions, requiring new crystal screenings to find initial hits. Constructs with methylated A-3 and U-4 failed to produce crystals in 1056 tested conditions, while the constructs with modifications at A-19 and U-20 produced crystals in 29 of 672 tested conditions. X-ray testing of a crystal produced from modified U-20 preQ1c showed a resolution limit of 3.1 Å (Figure 6c), the best obtained results of the project. Furthermore, resolution limits better than 3.5 Å are typically considered sufficient
The Stony Brook Young Investigators Review, Fall 2011