RNA Folding at Millisecond Intervals by Synchrotron Hydroxyl Radical Footprinting

See allHide authors and affiliations

Science  20 Mar 1998:
Vol. 279, Issue 5358, pp. 1940-1943
DOI: 10.1126/science.279.5358.1940


Radiolysis of water with a synchrotron x-ray beam permits the hydroxyl radical–accessible surface of an RNA to be mapped with nucleotide resolution in 10 milliseconds. Application of this method to folding of the Tetrahymena ribozyme revealed that the most stable domain of the tertiary structure, P4-P6, formed cooperatively within 3 seconds. Exterior helices became protected from hydroxyl radicals in 10 seconds, whereas the catalytic center required minutes to be completely folded. The results show that rapid collapse to a partially disordered state is followed by a slow search for the active structure.

The speed of chemical reactions carried out by ribozymes is often limited by conformational changes in the RNA (1). As a result, the process by which RNA molecules fold into their native conformation has received much attention. Early investigations into the folding of tRNA established approximate time scales for the formation of RNA secondary (10 4 to 10 5 s) and tertiary interactions (10 2 to 10 1 s), with the reorganization of incorrect secondary structures occurring more slowly (0.1 to 1 s) (2). Recent work has shown that folding of large RNAs is more complex (3), involving multiple pathways (4). Individual domains of an RNA may form at rates that differ by orders of magnitude, with some transitions requiring minutes to reach completion (3-7). Identification of the paths by which large RNAs fold has been hampered by the lack of experimental methods capable of probing RNA conformation with nucleotide resolution at subsecond time scales (8). Here, we describe direct measurement of the complete folding pathway of theTetrahymena ribozyme by hydroxyl radical footprinting using a synchrotron x-ray beam.

Hydroxyl radical ribose oxidation and resulting strand cleavage are correlated with the solvent accessibility of the RNA backbone (9, 10) and are insensitive to base sequence and secondary structure (11). Generation of hydroxyl radicals by the radiolysis of water yields cleavage products that are comparable with Fe(II)-EDTA–dependent reactions (7, 12). The high flux provided by white-light x-ray beams at the National Synchrotron Light Source (NSLS) permits footprinting of the ribozyme to be accomplished with millisecond time resolution (7).

The ribozyme derived from the Tetrahymena group I intron (Fig. 1A) folds into a well-defined tertiary structure in the presence of Mg2+, and Mg2+ is required for catalytic activity (1). The ribozyme contains at least three domains of tertiary structure (13) that, when separated, can reassociate to form the active ribozyme (14). The domain containing paired regions P4-P6 (Fig. 1A) folds independently (10, 15), and formation of P4-P6 has been proposed to be the first step in the folding pathway of the ribozyme (3, 16). In earlier experiments in which RNA was manually mixed with Mg2+ before exposure to the x-ray beam, we showed that the tertiary structure of the P4-P6 domain is formed within 30 s, the initial time of the assay (7).

Figure 1

Hydroxyl radical footprinting of theTetrahymena L-21 ribozyme with a synchrotron x-ray beam. (A) Ribozyme secondary structure adapted from (13, 27). Paired (P) and joining (J) regions are numbered 5′ to 3′. Lettered bases are protected from hydroxyl radical cleavage in 10 mM Mg2+ at 42°C; the remaining sequence is outlined schematically. Rate constants for changes in protection were determined independently for each colored area as illustrated in Fig. 2 (21). Colors indicate regions with similar folding rates. Rate constants were not determined for several protected sites in P4-P6 (nucleotides 195 to 196, 223 to 227, and 252 to 260) because of high background and very weak protection. (B) Front and back views of a space-filling modelof the P4-P6 domain from (10), with nucleotides colored as in (A).

To resolve early steps in the ribozyme folding pathway, we installed a stopped-flow apparatus with an x-ray exposure chamber on NSLS beamline X-9A (17). The flux of X-9A absorbed by the sample was sufficient to cleave 20% of the RNA molecules with exposures as short as 10 ms (18). Folding reactions were begun by mixing RNA with buffer containing Mg2+, to a final concentration of 10 mM (19). Samples were irradiated at a series of times after mixing, and the hydroxyl radical cleavage products were separated by gel electrophoresis (19). The ribozyme was fully active after passage through the stopped-flow apparatus (20), verifying that the RNA had folded correctly under these experimental conditions.

We determined the folding kinetics of the ribozyme by quantitating the changes in solvent accessibility of individual sites as a function of time (Fig. 2) (7, 21). After the addition of Mg2+, specific nucleotides within the P4-P6 domain became protected from cleavage within 100 ms (22), and the extent of protection reached a plateau within several seconds (Fig. 2). Comparison of this plateau with control reactions, in which the ribozyme was preequilibrated in Mg2+ before the start of the experiment, demonstrates that folding of P4-P6 was complete within this time (Fig.2, triangles).

Figure 2

Time dependence of hydroxyl radical protection. Fractional saturation of individual protected sites,Y̅, was determined from fits to the coupled equations p = p lower + (p upperp lower) Y̅and Y̅ = 1 –e kt, where pis the apparent saturation, p lower andp upper are the lower and upper limits of the transition curve, respectively, k is the first-order rate constant, and t is time (in seconds). Data from three to six independent experiments were plotted simultaneously and fit to the first-order rate expression (solid line). Additional exponential terms were not supported by the data. Open symbols represent controls in which the ribozyme was preequilibrated with Mg2+. Details of the data analysis are described elsewhere (29). A similar plot was produced for each of the protected sites shown in Fig. 1A. (A) Protection of P5c, nucleotides 174 to 176; k = 2.7 (−1.3, +1.8) s−1. (B) A-rich bulge, nucleotides 183 to 189; k = 0.9 (±0.3) s−1. (C) P2, nucleotides 57 to 59;k = 0.20 (±0.05) s−1. (Insets) Expansion of first 3 s of time axis.

The rapidly protected sites correspond to nucleotides that were excluded from solvent by folding of the P4-P6 domain upon itself (Fig.1, A and B). These rapidly protected sites include nucleotides in P5 and P5a, an A-rich bulge, and a GAAA tetraloop that are involved in interactions that stabilize the tertiary structure of the domain (15, 23). Rate constants determined for each of these regions (Fig. 1A, orange) were the same within experimental error. Thus, the tertiary interactions within the P4-P6 domain were established in a concerted manner at a rate of about 1 s 1 at 42°C.

A subset of nucleotides in P5c (Fig. 1A, green) was protected about twice as rapidly as other regions in the P4-P6 domain. Protection of riboses in P5c results from local interactions that bury the P5c backbone (15). P5a-P5c constitutes a Mg2+-rich subdomain that folds independently of interactions with P4 and P6 (15, 24). Although the differences in the rate constants determined for these nucleotides are at the limit of the precision of the data, these results suggest that formation of a metal ion “core” in P5a-P5c (24) is one of the earliest folding transitions of the ribozyme. This region could serve as a nucleation site for additional tertiary structure.

Several groups of nucleotides are protected from hydroxyl radicals by tertiary contacts that are present in the ribozyme but not in the isolated P4-P6 domain (15). Residues in P5 (118 to 121) and joining region J5/4 (204 to 208) are protected by folding of P9.1 and P9.2 (25). The 5′ and 3′ ends of the P4-P6 domain are part of a triple helix that mediates interactions with double helices P3 and P7 (14). Both groups of nucleotides (Fig. 1A, pink) became protected more slowly (k ∼ 0.3 s 1) than the interior of P4-P6. Thus, tertiary contacts with the P3-P9 domain (P3, P7, P8, and P9) formed after the P4-P6 domain was folded. This is consistent with the observation that tertiary interactions with P4-P6 stabilize the folded structure of P3-P9 (13, 25, 26).

In agreement with this conclusion, nucleotides in P2, P2.1, and P9.1 were protected from hydroxyl radical cleavage at similar rates (k = 0.2 to 0.4 s 1) (Fig. 1A, pink). These helices bridge the two central domains of the ribozyme, stabilizing the catalytic center by base pairing between the loops L2 and L5c and L2.1 and L9.1 (27). P2-P2.1 and P9.1-P9.2 are proposed to wrap around the exterior of the folded ribozyme, and protection from hydroxyl radical cleavage results from contacts with the folded P4-P6 and P3-P9 domains (26, 27). Thus, much of the tertiary structure is formed in about 10 s.

In contrast, P3, P7, and P9 required minutes (k = 0.02 to 0.06 s 1) to become fully protected from hydroxyl radical cleavage (Fig. 1A; yellow). This result is consistent with our earlier results (7) and with oligonucleotide hybridization and chemical modification data showing that P3 and P7 are the last stems to form completely (3, 5,7). Protection of nucleotides in P6 (220 to 222) is attributed to close packing with the stacked P3 and P8 helices (27), and it appeared at a similar rate (k = 0.03 s 1) (Fig. 1A). Thus, the P3-P9 sequences remain disordered until late in the folding process. Solvent accessibility of P3-P9 could result from either a highly extended conformation or from a mixture of conformations with nonoverlapping footprints that are in slow exchange.

The ability to probe solvent-accessible regions of an RNA backbone within the first tens of milliseconds of a reaction provides a visualization of early steps in the folding pathway of theTetrahymena ribozyme. An assumption of the model depicted in Fig. 3 is that much of the secondary structure is formed under the initial conditions of our assay (no Mg2+ and 42°C). After the addition of Mg2+, the earliest evidence of tertiary structure appeared within the P5a-P5c subdomain. Subsequent collapse of the P4-P6 domain was concerted and occurred in a few seconds. Interdomain contacts with P2-P2.1 and P9.1 were established within 10 s, implying further condensation of the RNA. Organization of the catalytic core, including formation of P3 and P7, is about tenfold slower (3, 7) and could involve either local fluctuations of the RNA chain or rearrangement of alternative secondary structures (4, 5,28). Large movements of P3-P9 could be accommodated by transient opening of interactions between P2-P2.1 and P9.1.

Figure 3

A model for the early steps of the Mg2+-dependent folding of the Tetrahymenaribozyme. Residues in P5c become protected most rapidly, about twofold faster than nucleotides in the interior of the P4-P6 domain. Nucleotides that are excluded from solvent by interactions with P2-P2.1 and the P3-P9 domain are protected more slowly. Ordering of the catalytic core occurs over several minutes (3,5, 7) and may involve reorganization of alternative conformations after collapse to a partially disordered intermediate state. For simplicity, folding is depicted as a linear sequence of events, although there are likely to be multiple folding pathways with different intermediates (4). Some molecules in the population may reach the native state rapidly, whereas others fold slowly because of the presence of kinetic traps.

This study reports RNA folding kinetics in which condensation of tertiary structure occurred at rates similar to that of in vivo self-splicing (29). Folding of the P4-P6 domain of theTetrahymena ribozyme was only tenfold slower (1 s 1) than formation of tertiary interactions in tRNA (100 ms) (2). Folding of P4-P6 and interactions with P2 rapidly reduce the number of available conformations. The catalytic center, however, remains disordered until late in the folding process; final steps occur after the RNA has become compact. Because formation of the native structure requires condensation of the RNA and coordination of multiple Mg2+ ions (15,24), the fast folding events may be sensitive to solvation and divalent ion concentration.


View Abstract

Navigate This Article