Computational identification of RNA functional determinants by three-dimensional quantitative structure-activity relationships

Abstract

Anti-infection drugs target vital functions of infectious agents, including their ribosome and other essential non-coding RNAs. One of the reasons infectious agents become resistant to drugs is due to mutations that eliminate drug-binding affinity while maintaining vital elements. Identifying these elements is based on the determination of viable and lethal mutants and associated structures. However, determining the structure of enough mutants at high resolution is not always possible. Here, we introduce a new computational method, MC-3DQSAR, to determine the vital elements of target RNA structure from mutagenesis and available high-resolution data. We applied the method to further characterize the structural determinants of the bacterial 23S ribosomal RNA sarcin-ricin loop (SRL), as well as those of the lead-activated and hammerhead ribozymes. The method was accurate in confirming experimentally determined essential structural elements and predicting the viability of new SRL variants, which were either observed in bacteria or validated in bacterial growth assays. Our results indicate that MC-3DQSAR could be used systematically to evaluate the drug-target potentials of any RNA sites using current high-resolution structural data.

Figure 1.

MC-3DQSAR steps. A set of 3D models is built for each sequence of the learning set by using atomic superimposition of substituted bases in a template of a reference high-resolution structure, or by 3D modeling. The models constitute the training set. The exposed area of each atomic group is computed using pymol and a probe water of radius 1.4 Å. The solvent accessible groups are considered potential determinants if they are present in all positive examples of the training set and absent in at least one negative example. Determining the activity of a new sequence variant consists in building its 3D structure, computing the solvent accessibility of its chemical groups and comparing its profile to the activity profile.

Figure 2.

E. coli 23S rRNA SRL. (A) Stereo view of the SRL 3D structure (PDB ID 2AWB). The base and phosphodiester linkage (cylinder) of each nucleotide are shown in dark gray. Activity determinants identified by MC-3DQSAR are shown with spheres, where the donor groups are shown in black and bold labels and acceptors in light gray and regular labels. Neutral determinants are not shown. (B) Secondary structure and NCMs. The NCMs are numbered 1–7. The backbone is shown using bold lines; the base pairing interactions are shown using the Leontis and Westhof nomenclature; and the base stacking interactions are shown using the Major and Thibault nomenclature. The activity determinants identified by MC-3DQSAR are shown in boxes. The identity of the atomic groups from the seed sequence is shown in bold for donor, italic for neutral and regular for acceptor determinants.

Figure 3.

Structural determinants of viable bacterial SRLs. (A) Electrostatic surface of the sarcin–ricin/EF-G interaction. Surface potentials are presented in kT/e. The red color indicates negatively charged areas (EF-G from −1; SRL from −8) and blue positively charged areas (EF-G to 1; SRL to 0). The SRL/EF-G recognition put in close contact two negatively charged areas, made possible by the presence of two Mg2+ ions (green spheres). (B) Plausible Mg²⁺ coordination between the SRL and EF-G from crystal structures (PDB IDs 4KIX and 4KIY). The green dashed lines indicate possible direct coordination. One of the Mg²⁺ ions (green sphere on the right) is located near atoms Gln120:OE1 and Asp20:OD2 of the EF-G, and A2657:O2P, G2659:O6, G2659:N7 and G2661:O1P of the SRL. The second Mg²⁺ ion (green sphere on the left) is near atoms Arg59:O and Asp20:OD1 of the EF-G, and G2655:O2′, G2662:O1P, G2662:O2P, G2663:N7 and G2664:O6 of the SRL. The EF-G residues and SRL nucleotides involved in Mg²⁺ coordination are shown with colored sticks; others in gray. The bases G2655, G2659, G2663 and G2664 stack and are shown with blue carbons. (C) The SRL structural determinants represented as sequence and base interaction constraints. The SRL is composed of the GNRA and G-bulge motifs linked by the Y2658:N2663, but C2658:C2663, junction base pair.

Figure 4.

Leadzyme structure. (A) Stereo view of the leadzyme 3D structure (PDB ID 1NUJ). Graphical details as in Figure 2A. (B) Secondary structure and NCMs (numbered 1–8). Base pairing and stacking nomenclature, and activity determinants are shown as in Figure 2A.

Figure 5.

Hammerhead ribozyme structure. (A) Stereo view of the hammerhead ribozyme 3D structure (PDB ID 2OEU). Graphical details as in Figure 2A. (B) Secondary structure and NCMs (numbered 1–7). Base pairing and stacking nomenclature, and activity determinants are shown as in Figure 2A.