R EPORTS stacked (undocked state), or it can form a bent conformation with tertiary interactions Correlating Structural Dynamics between loops A and B (docked state) (12, 13). The structures of the isolated domains and Function in Single and that of the docked ribozyme have been studied by biochemical methods (14, 15), Ribozyme Molecules nuclear magnetic resonance (NMR) (16, 17), and x-ray crystallography (18). Xiaowei Zhuang,1* Harold Kim,1 Miguel J. B. Pereira,2 The hairpin ribozyme reversibly cleaves Hazen P. Babcock,1* Nils G. Walter,2† Steven Chu1† its substrate (S) into two products (3ЈP and 5ЈP). Previous studies (12) have suggested a We have studied the correlation between structural dynamics and function of multistep reaction pathway (Fig. 1B): (i) Sub- the hairpin ribozyme. The enzyme-substrate complex exists in either docked strate S binds to the ribozyme, resulting in the (active) or undocked (inactive) conformations. Using single-molecule fluores- undocked conformation; (ii) the complex cence methods, we found complex structural dynamics with four docked states folds into the docked state; (iii) cleavage of distinct stabilities and a strong memory effect where each molecule rarely occurs; and (iv) the complex undocks and the switches between different docked states. We also found substrate cleavage to cleavage products dissociate, or vice versa. In be rate-limited by a combination of conformational transitions and reversible this work, the proposed reaction pathway has chemistry equilibrium. The complex structural dynamics quantitatively explain been largely verified and many of the kinetic the heterogeneous cleavage kinetics common to many catalytic RNAs. The rate constants have been measured. With intimate coupling of structural dynamics and function is likely a general phe- these rate constants, the rate-limiting steps of nomenon for RNA. the overall cleavage reaction (19, 20) have been determined and the heterogeneity in RNA plays a central role in cellular pro- relationship that is still poorly understood. cleavage reaction kinetics (21, 22) can now cesses such as splicing and translation (1). Small ribozymes provide critical model be understood. In addition to a number of protein-indepen- systems for structural dynamics–function Figure 1A shows the ribozyme construct dent RNA enzymes (ribozymes), recent in- studies (6), but despite their apparent sim- used in our experiments (23). The dye mole- sights into the structure and function of plicity, little is known about the structural cules Cy3 (donor) and Cy5 (acceptor) are large ribonucleoproteins suggest that ribo- dynamics of these enzymes. Here, we used attached to the 3Ј and 5Ј ends of the RzA somes, and perhaps spliceosomes, are also fluorescence resonance energy transfer strand, respectively, and serve as the FRET “ribozymes” in which the RNA constituents (FRET) to study the structural dynamics of pair. Biotin attached to the 5Ј end of strand play the major catalytic role (2–5). The cat- the hairpin ribozyme at the single-molecule RzB allows the molecule to be immobilized alytic activities of ribozymes depend critical- level (7–10). Our results show that the struc- on a surface via the streptavidin-biotin inter- ly on their folded structures, and the complex tural dynamics of RNA can be very complex action (7). The donor and acceptor fluores- structural dynamics of RNA may impose and is intimately coupled to its function. cence of single ribozyme molecules was consequences on their catalytic functions, a The hairpin ribozyme is derived from the detected using a total internal reflection or autocatalytic negative strand of the tobacco scanning confocal microscope (7). 1Department of Physics, Stanford University, Stan- ringspot virus satellite RNA (11). The mini- First, we measured the cleavage activity ford, CA 94305, USA. 2Department of Chemistry, mal active form of the ribozyme of ϳ50 of the surface-immobilized ribozyme by University of Michigan, Ann Arbor, MI 48109, USA. nucleotides consists of the two independently monitoring FRET between Cy3 and Cy5. The *Present address: Department of Chemistry and Chemical folding helix-loop-helix domains A and B FRET value is defined as I /(I ϩ I ), where Biology, Harvard University, Cambridge, MA 02138, USA. A A D †To whom correspondence should be addressed. E- (Fig. 1A). This RNA can form an extended ID and IA are the fluorescence intensities of mail: [email protected], [email protected] conformation with the two domains coaxially the donor and acceptor, respectively. Low Fig. 1. Structural dynamics and function of the hairpin ribozyme. (A) The two-strand (RzA, RzB) hairpin ribozyme (SV5 EH4) used in this study binds substrate (orange; arrow, cleavage site) to form domain A, comprising helices H1 and H2 (short lines, Watson-Crick base pairs) and the symmetric internal loop A. Domain A is connected by a flexible hinge to domain B of the ribozyme, comprising helices H3 and H4 and the asymmetric internal loop B. Noncanonical base pairs are indicated as dashed lines. Color code for tertiary hydrogen bonds in the docked state: red, gϩ1-C25 Watson-Crick base pair; blue, ribose zipper; purple, U42 binding pocket (18). For our studies, biotin and the fluorophores Cy3 and Cy5 were attached as indicated. (B) The putative reaction pathway of the hairpin ribozyme. The rate constants measured in this work are given. [Details of the experiments determining these rate constants are described in the text and (25).] www.sciencemag.org SCIENCE VOL 296 24 MAY 2002 1473 R EPORTS FRET occurs when the enzyme is in the reaction kinetics were found free in solution docked conformation. Then the molecule undocked state, with the distance between the for the same ribozyme construct (Fig. 2) and switches stochastically between the undocked 3Ј and 5Ј ends of the RzA strand estimated to for the unlabeled ribozyme (21). Thus, the and docked states. be ϳ8nm(24). High FRET occurs in the functional properties of the hairpin ribozyme We found direct evidence that cleavage docked state (ϳ3 nm), and intermediate are not compromised by surface immobiliza- occurs only in the docked state, as previously FRET is seen when S is not bound (ϳ6 nm) tion and dye labeling. suggested (12). Figure 3B shows a typical (24). A FRET distribution of ribozyme mol- Next, we studied the conformational FRET time trace of cleavable substrate S ecules during the cleavage reaction is shown changes of the ribozyme upon S binding. bound to a single ribozyme. The final FRET in Fig. 2. The population of S-free ribozyme After S binding, 98 to 99% of the single- jump to an intermediate level signals the grows with time as a result of cleavage and molecule FRET time trajectories (Fig. 3A) release of cleavage products. Among the sev- product release. The reaction time course show a FRET decrease; this finding indicates eral hundred traces examined, 90 to 95% shows heterogeneous reaction kinetics with that the molecule enters the undocked state show transitions to the S-free state from the two distinct cleavage rate constants. Identical before attaining the catalytically active docked state. The fraction (5 to 10%) show- ing the transition from the low-FRET state is consistent with the contributions from short Fig. 2. Single-molecule and bulk solution mea- docked events (which were missed because surements of enzymat- of our 2-s time resolution) and from slow ic activities. The open substrate dissociation (11). Thus, cleavage symbols show the reac- only occurs in the docked state. tion time course of sur- The docking and undocking kinetics can face-immobilized ri- be deduced from the dwell times of the bozyme. For initiation of cleavage, a buffer molecules in the undocked and docked containing 200 nM S states, respectively (Fig. 3A) (7). The dwell was added to the im- times in the undocked state can be de- mobilized ribozymes for scribed by a single rate constant for dock- 30 s to allow S binding; ϭ Ϫ1 ing, kdock 0.008 s . The dwell times in S was then removed the docked state show a complex behavior, from the buffer. During the reaction, the FRET with at least four rate constants for undock- ϭ Ϫ1 ϭ distribution showed ing: kundock,1 0.005 s , kundock,2 0.06 Ϫ1 ϭ Ϫ1 ϭ three distinct ribozyme s , kundock,3 0.5 s , and kundock,4 3 populations: undocked sϪ1 (Fig. 4) (25). Fewer rate constants or a (FRET ϳ0.15), docked ϳ Gaussian distribution of rate constants do ( 0.81), and S-free ri- not sufficiently describe the data. This find- bozymes (ϳ0.38). The peak at FRET ϳ0 was due to inactive Cy5 (7). The S-free fraction is plotted against time. In a control ing suggests the existence of four distinct experiment with noncleavable S [with a 2Ј-O-methyl modification at the cleavage site, S(2ЈOMeA-1)], docked states. The docking and slowest the S-free fraction accumulated with a rate constant slower than 4 ϫ 10Ϫ5 sϪ1, indicating that S undocking rate constants compare well dissociation is much slower than cleavage. The solid symbols show the reaction time course for the same with previously estimated rate constants for ribozyme free in solution, as determined by gel electrophoresis and autoradiography (21). The data docking and undocking from ensemble so- cannot be fit by a single-exponential function, indicating heterogeneous reaction kinetics. The solid curve lution measurements (12). The three less is a numerical fit using the rate equations given in (25). stably docked states were not detected pre- viously. Their fast undocking means that Fig. 3. Following the these states are not significantly populated structural dynamics at any time, making them practically im- and function of single ribozyme molecules. possible to detect by ensemble methods. (A) Typical fluores- The time trajectories of individual mol- cence time trace of a ecules reveal a pronounced “memory” ef- single ribozyme upon fect (26–29).