Use of the spliceosomal protein U1A to facilitate crystallization and structure determination of complex RNAs

Abstract

The structure determination of complex RNA molecules such as ribozymes, riboswitches and aptamers by X-ray crystallography hinges on the preparation of well-ordered crystals. Success usually results from molecular engineering to facilitate crystallization. An approach that has resulted in 10 new RNA structures in the past decade is the use of the U1A crystallization module. In this approach, the cognate site for the U1A spliceosomal protein is introduced into a functionally dispensable location in the RNA of interest, and the RNA is cocrystallized with the basic RNA-binding protein. In addition to facilitating crystallization, the presence of U1A can be useful for de novo phase determination. In this paper, some general considerations for the use of this approach to RNA crystallization are presented, and specifics of the application of the U1A module to the crystallization of the hairpin ribozyme and the tetracycline aptamer are reviewed.

Figure 1

Summary of the strategy employed for generating candidate crystallization constructs for the hairpin ribozyme. A. Schematic of the four-helix junction form of the hairpin ribozyme embedded in the negative strand of the circular satellite RNA tobacco ringspot virus. Red arrow denotes site of self-cleavage B. Prototype minimized ribozyme split into an in vitro transcribed strand (black) and a synthetic oligonucleotide (gold) containing a single 2′-methoxy substitution (circle) that blocks the cleavage reaction. C. The U1A binding site was introduced at the distal ends of either stem B (bottom) or stem C. Varying numbers of spacer base pairs separated the U1A binding loop from the core of the RNA. D. Additional molecular diversity was generated by extending the ends of stems A and D by varying the lengths of both the transcript and the synthetic oligonucleotide.

Figure 2

Summary of the strategy employed for generating candidate crystallization constructs for the tetracycline aptamer. A. The minimal tetracycline aptamer (black lines) is bracketed by a functionally dispensable loop L2 and distal portion of P1 (gray). B. A series of crystallization constructs was generated by splitting the RNA into a short synthetic oligonucleotide (red) and a longer in vitro transcript (black). Diversity was generated by varying the length of the distal ends of P1 and P2. C. Another series of constructs contained a GAAA tetraloop in place of L2. Diversity obtained from varying the number of spacer base pairs connecting the tetraloop to P2, and by varying the length of the distal end of P1. D. One set of crystallization constructs contained the U1A binding site at the distal end of P2. Diversity was generated in the same manner as in (C). E. Another set of crystallization constructs was circularly permuted relative to the parental sequence (A), by introducing the U1A binding site at the distal end of P1.

Figure 3

Plot of the density of RNA crystals grown employing the U1A crystallization module vs. resolution limit. The Matthews number [Vm, (73)] is plotted as a function of the maximum resolution for all structures in Table 1, except the Azoarcus group I intron, which appears to be an outlier (Vm ~ 4.4). Although the linear regression is based on a small number of data points, it suggests counterintuitively that higher resolution correlates with higher solvent content (R = 0.77).

Figure 4

Comparison of crystal packing of two different constructs of the tfoX c-di-GMP riboswitch from Vibrio cholereae crystallized with the aid of the U1A module. A. Detail of crystal packing of the crystals of Kulshina et al. (50). Here, the distal end of P1a of one RNA molecule (white) stacks on the equivalent helical end of another molecule (black). The 3′ overhanging G of each molecule (* and †) is extruded from the intermolecular helical stack. B. Detail of crystal packing of the crystals of Smith et al. (54). This RNA construct differs from that of (A), inter alia, in having a 3 nt 5′ extension. These nucleotides form three Watson-Crick base pairs (*) with three nucleotides of the U1A binding site (black) of another molecule in the crystal. U1A-RBD is shown as a ribbon figure (74).

Figure 5

Packing of a T-loop tetraloop against the U1A crystallization module in crystals of the flexizyme (48). The same three nucleotides highlighted in Figure 4B form Watson-Crick base pairs (*) with three nucleotides from the T-loop of a symmetry related molecule (black) in this crystal form.