Increased Ribozyme Activity in Crowded Solutions*

Background: RNAs have evolved to function in the crowded environment of the cell. Results: Large crowder molecules increase the catalytic activity of a group I ribozyme at physiological Mg2+, whereas small neutral molecules do not. Conclusion: Excluded volume due to crowding stabilizes the native active state of RNAs. Significance: Crowding significantly improves RNA folding and function. Noncoding RNAs must function in the crowded environment of the cell. Previous small-angle x-ray scattering experiments showed that molecular crowders stabilize the structure of the Azoarcus group I ribozyme, allowing the ribozyme to fold at low physiological Mg2+ concentrations. Here, we used an RNA cleavage assay to show that the PEG and Ficoll crowder molecules increased the biochemical activity of the ribozyme, whereas sucrose did not. Crowding lowered the Mg2+ threshold at which activity was detected and increased total RNA cleavage at high Mg2+ concentrations sufficient to fold the RNA in crowded or dilute solution. After correcting for solution viscosity, the observed reaction rate was proportional to the fraction of active ribozyme. We conclude that molecular crowders stabilize the native ribozyme and favor the active structure relative to compact inactive folding intermediates.

Folded RNA molecules have evolved to function in the intracellular environment, which is crowded with other cellular macromolecules that are thought to occupy up to 30% of the total cell volume (1,2). RNA is therefore excluded from onethird of the cell's volume. These excluded volume effects are well described by theories based on simple molecular shapes (3,4), but whether excluded volume contributes substantially to the energetics of RNA folding in vivo remains controversial.
Recent small-angle x-ray scattering (SAXS) 2 experiments demonstrated that PEG crowder molecules stabilize compactly folded RNA structures (5,6). The RNA not only folded at lower Mg 2ϩ concentration in the presence of PEG, but the folded RNA was more compact in the crowded solution, as judged by changes to the solution scattering. This stabilizing effect was not observed in ethylene glycol and could not be explained by changes to Mg 2ϩ ion or water activity (5,6). We therefore concluded that PEG stabilizes the folded RNA mainly through excluded volume effects.
The SAXS experiments could not reveal, however, whether molecular crowding improves (or hinders) the biochemical function of the RNA. Here, we show that macromolecular crowding increases the catalytic activity of the group I ribozyme from the bacterium Azoarcus (7,8), demonstrating that crowders not only favor collapse of the RNA helices into folded structures but also populate the true native (functional) state.
In our activity assays, we used the L-3 Azoarcus ribozyme, which is similar to the L-9 ribozyme used for SAXS in crowded solutions (5) but with an additional 6 nucleotides at the 5Ј-end to facilitate binding of oligonucleotide substrates that mimic the natural self-splicing of the Azoarcus pre-tRNA ( Fig. 1). Base pairing of the substrate RNA to a guide sequence in the ribozyme produces the P1 splice site helix, which docks into the active site via tertiary interactions with the core of the folded ribozyme (9,10). An exogenous G nucleotide binds a pocket in P7, from which its 3Ј-OH attacks the splice site phosphodiester, cleaving the RNA substrate into two shorter fragments (9 -12). Two coordinated Mg 2ϩ ions organize the active site and stabilize the transition state for phosphoryl transfer (12,13).
Previous work on the Tetrahymena and Azoarcus group I ribozymes showed that substrate binding determines the observed cleavage rate for single-turnover reaction conditions (14,15). Therefore, at high RNA concentrations, when substrate binding is fast, the amount of product formed in the first turnover reflects the fraction of ribozyme-substrate complexes competent to react (14,15). This relationship between activity and the fraction of native ribozyme has been used to study the assembly of core helices (9,16,17), the association of Mg 2ϩ ions (18) and guanosine (19) with the core, and the effects of mutations (20) and protein chaperones (15) on formation of the native state.
PEG was previously shown to accelerate the activity of the hammerhead ribozyme (21). By comparing the cleavage rates of two 58-and 86-nucleotide ribozymes and the thermal stability of individual helices and hairpins, it was reasoned that the main influence of PEG as a crowder was to destabilize Watson-Crick base pairs. This destabilization allowed incorrectly folded RNA to refold more easily to the native state, increasing the overall activity. Tellingly, hammerhead self-cleavage was not accelerated when using Ficoll as a crowder, and from this, the authors inferred that the excluded volume effects were minimal. PEG was also reported to directly alter DNA quadruplex conformation (22). By contrast, compartmentalization increased selfcleavage of a two-part hammerhead ribozyme, showing that co-localization can increase RNA interactions (23).
The aim of our work was to understand how crowders perturb the formation of the native state of the Azoarcus ribozyme. We addressed whether the same native state is formed in crowded and dilute solutions, whether crowding alters the intrinsic maximum activity of the ribozyme, and whether this is most attributable to excluded volume (steric) effects or surface interactions.

EXPERIMENTAL PROCEDURES
RNA Preparation-The L-3 Azoarcus ribozyme (201 nucleotides) was transcribed in vitro from PCR templates as described previously (11,17). The RNA stock was incubated at 50°C for 5 min before use. The RNA solutions were 0.4 mg/ml in 20 mM Tris-HCl (pH 7.5) with the desired co-solute concentration. These conditions were used for previous SAXS measurements. Prior to each measurement, the final RNA solution was incubated for 5 min at 50°C and then for 30 min at 37°C. This treatment was found in control reactions and native gels to minimize the presence of a misfolded species (24) that forms at low temperatures (15).
Activity Assays-RNA substrate cleavage assays were performed as described (17) by adding trace 32 P-labeled 5Ј-rCAUAUCGCC and 0.5 mM GTP to ribozyme pre-equilibrated at the desired PEG and MgCl 2 concentrations. Reactions were stopped between 25 s and 30 min. Because the ϳ1 nM substrate was 1000 times less concentrated than the ribozyme (6 M), these conditions measured the single-turnover rate for substrate cleavage, which we used to describe the native state fraction. Size exclusion chromatography assays confirmed that the substrate was at least 80% bound with or without PEG under these reaction conditions. Gel mobility shift assays confirmed that substrate binding occurred within the shortest time measured (ϳ15 s).

RESULTS AND DISCUSSION
Ribozyme Activity Assays Measure Folding-We observed previously that the cleavage activity of the L-3 Azoarcus ribozyme increased between 1 and 10 mM MgCl 2 , as the native state of the ribozyme became more populated. The increase in activity correlated with folding of the active site over the same range of MgCl 2 concentrations, as reported by a change in the hydroxyl radical footprinting pattern and cleavage in the presence of Tb 3ϩ (11,18).
To determine whether PEG crowding alters the ribozyme activity, we compared the reaction rates in dilute solution and at various concentrations of PEG. Fig. 2 shows the fraction of oligomers that were cleaved as a function of time in solutions containing 8% (panel a) and 16% (panel b) (w/v) PEG 1000 and a range of MgCl 2 concentrations.
Under our single-turnover reaction conditions (37°C), the rate constant and amplitude of the initial reaction phase  . a, fraction of oligomer cleaved over time. 32 P-Labeled 9-nucleotide substrate was added to the prefolded L-3 Azoarcus ribozyme in 8% PEG 1000 solutions with varying Mg 2ϩ concentrations as indicated. The amplitude of the initial phase is assumed to be proportional to the fraction of the native ribozyme. Data at 1.1 mM MgCl 2 were fit to a simple exponential for the initial phase (dashed line) and to a stretched exponential for the initial phase (solid line). Systematic deviations from the data at short times indicate that a simple exponential does not fully describe the data. We saw similar systematic differences under other solution conditions. No such systematic differences were observed for the stretched exponential fits (all other MgCl 2 concentrations). b, same as a but with 16% PEG 1000. Error bars were calculated as the S.D. of values from three repeats of four experimental conditions. Because these errors were found to be similar, the mean S.D. was assumed to represent the measurement error for all experiments.
increased with increasing MgCl 2 concentration (Fig. 2), corresponding to a greater fraction of folded ribozyme at higher MgCl 2 concentrations (11). The rate of cleavage in this initial phase also increased with MgCl 2 concentration, from 0.1 min Ϫ1 in 1 mM MgCl 2 to Ն3 min Ϫ1 in 5 mM MgCl 2 . At the highest concentrations of Mg 2ϩ tested (5-20 mM MgCl 2 ), the product was formed within 15-20 s at 37°C (17,20), comparable to a self-splicing rate of Ն5 min Ϫ1 for the full pre-tRNA at 32°C (25). At low Mg 2ϩ concentrations, the rate of cleavage appeared to be limited by the formation of the active ribozymesubstrate complex, as reported previously (15). To accommodate this, the data were fit to a biphasic partially stretched exponential model (Equation 1), allowing us to extract the amplitude of the initial phase (A), its observed rate constant (k 1 ), and a rate constant for a slower reaction phase (k 2 ). The fraction of cleaved product approached the asymptotic limit of ϳ0.5, reflecting the internal equilibrium between cleavage of the substrate and religation of the products, which is ϳ1 (15). In this analysis, we assume that the internal equilibrium is unaffected by the presence of crowders.
The stretching parameter (n) was required to adequately fit the initial phase. To illustrate this, fits are shown in Fig. 2a to the data at 8% PEG 1000 and 1.1 mM MgCl 2 (black circles) with both the stretched (solid line) and simple (dashed line) exponential fits. Similar differences in goodness of fit were found for the other salt and PEG concentrations. Our fits give values 1 Ͻ n Ͻ 2 and are therefore compressed rather than stretched exponentials.
At low MgCl 2 concentrations, the initial phase of cleavage was more difficult to discern because the fraction of native ribozyme dropped below our detection limit. The slow buildup of product at very low MgCl 2 was presumably due to multiple turnovers of this tiny fraction of active ribozymes.
Crowding Favors the Active Ribozyme Structure-The first main observation from this work is that the ribozyme was still enzymatically active when folded in solutions crowded with PEG. From this, we infer that crowding stabilizes the true native ensemble of structures. Between ϳ1.5 and 5 mM MgCl 2 , the activity measured in crowded solutions was consistently higher than that in non-crowded solutions (Fig. 3a). In this [Mg 2ϩ ] range, a large fraction of the ribozyme population is folded with or without PEG (5, 6). We therefore conclude that the crowders additionally favor the active structure with respect to inactive compact structures. The measured R g of the folded RNA is lower in PEG (5), and we speculate that this additional compression of the native ensemble reflects the increased occupancy of the active conformation.
Crowding Increases Ribozyme Activity at Low Mg 2ϩ Concentrations-The second important observation from our data is that crowders increased the amount of active ribozyme at physiological (Յ1 mM) MgCl 2 concentrations. To estimate the fraction of active RNA, we plotted the amplitudes of the first turnover (A) from Equation 1 as a function of MgCl 2 concentration for different PEG 1000 solutions (Fig. 3a).
When the burst amplitudes were fit to a three-state model that empirically describes the ribozyme folding transitions in Mg 2ϩ (20,26), we observed a steep transition from inactivity to activity at ϳ1 mM MgCl 2 (Fig. 3a). A second smaller increase in activity occurred at ϳ2 mM MgCl 2 . 16% PEG 1000 lowered the midpoint of the main transition to 0.79 Ϯ 0.01 mM MgCl 2 from 1.08 Ϯ 0.03 in buffer.
One possibility is that PEG simply increases the activity of the Mg 2ϩ ions. Previous measurements showed that the change in Mg 2ϩ activity is too small to account for stabilization of the RNA in crowded solutions (5). When the solution volume occupied by PEG is accounted for, the transition midpoint in 16% PEG 1000 is 0.91 millimolal, still well below that in buffer.
Crowding Slows the Ribozyme Reaction Rate-The third result is that the observed rate of cleavage became slower when the ribozyme was stabilized in a crowded solution (Fig. 2, a and  b). Both diffusion of the substrate and intramolecular rearrangements are likely to be slower in crowded solutions due to their increased viscosity. For a first-order correction to this  27 and 28). Representative burst rates are 1.8, 2.1, 2.7, and 1.5 min Ϫ1 for 0%, 4%, 8%, and 16% PEG 1000, respectively, at 2 mM MgCl 2 . We omitted the data from experiments in 0% PEG 1000 at 3 and 5 mM MgCl 2 and in 8% PEG 1000 at 5 mM MgCl 2 as the reactions were too rapid to accurately measure the initial rate. Error bars represent errors propagated through the fits. effect, we normalized the rate constant for the initial phase to the inverse of the viscosity (27,28), which is proportional to the diffusion coefficient in the Stokes-Einstein equation. The normalized rates were then plotted as a function of the initial phase amplitude in Fig. 3b. Fig. 3b shows that the reduced rate constants in crowded solutions can be accounted for by the viscosity of the solution, as would be expected in a diffusion ratelimited process. Once viscosity is accounted for, the rate constant is proportional to the amount of folded RNA.
Physical Interpretation of the Compressed Exponential-Even after accounting for the solution viscosity, the compressed exponential equation required to fit the cleavage kinetics means that the reaction mechanism is more complex than a simple first-order reaction. The compressed exponential term on the right hand side of Equation 1 can be empirically expressed as the sum of simple exponential terms with shifted time origins (Equation 2), with A ϭ 0.5. H(x) is the Heaviside function. Thus, our data are consistent with the superposition of physical processes that contribute to the overall cleavage reaction, with varied lag times. These lags could come from substrate-or GTP-induced refolding of the RNA. In addition, the RNA ensemble may contain different populations that react at different rates owing to their different probabilities of success for substrate cleavage (29).
Comparison of Activity and Folding-More Mg 2ϩ was needed to activate the ribozyme than to form compact folding intermediates (Fig. 4), for which the main folding transition observed by SAXS had a midpoint of 0.45 mM MgCl 2 without PEG when fit to a three-state folding model (5). A discrepancy between folding transitions measured by SAXS and by activity assays has been observed before for the Azoarcus ribozyme and a series of mutants (16,20) and is due to restructuring of the active site at higher Mg 2 ϩ (18).
The native ribozyme at high Mg 2ϩ could react in Ͻ15 s. When the fraction of oligomer cleaved after 25 s was plotted at a range of MgCl 2 concentrations in 0% PEG (Fig. 4), the transition to active ribozyme shifted even higher, to 2.7 Ϯ 0.1 mM MgCl 2 . These results are consistent with the data at 20 s from our previous study (20), which are also plotted in Fig. 4 for comparison. By contrast, our low MgCl 2 reactions had initial phases of cleavage that took up to ϳ5 min to complete.
Taken together, our data indicate that crowders increase the activity of the ribozyme at low Mg 2ϩ , but the resulting population is heterogeneous. At low MgCl 2 , the whole RNA population is compact (as determined by SAXS) but may not necessarily occupy the active (native) conformation (as indicated by slow reaction rates and compressed exponentials). Therefore, we view the Mg 2ϩ -dependent increase in the amplitude and rate constant of the initial phase as reflecting the formation of folded molecules that still retain some flexibility and that transiently occupy "native" states that are competent to cleave the substrate (Fig. 1b). These fluctuations could represent docking of the P1 splice site helix or recruitment of Mg 2ϩ ions into the active site, for example (16). As the reaction amplitude is reduced at low [Mg 2ϩ ], we infer that some non-native intermediates refold to the active state more slowly than the initial reaction rate.
Excluded Volume Effects of Large Crowders-We next measured the ribozyme activity in 8% PEG 8000 to determine whether the size of the crowder and thus its excluded volume are important for stabilizing the native structure. When the burst amplitudes (A) from Equation 1 fit to these data were compared with those from solutions with 0% PEG and 8% PEG 1000, the fitted midpoints of the folding transitions in PEG 1000 and PEG 8000 were indistinguishable within error (0.938 Ϯ 0.007 mM compared with 0.91 Ϯ 0.03 mM) (Fig. 5). In previous SAXS experiments, we also observed little difference in the midpoints for the collapse transition (0.44 Ϯ 0.02 versus 0.45 Ϯ 0.01 mM MgCl 2 for 8% PEG 1000 and PEG 8000, respectively) (6). We conclude therefore that the PEG molecular weight does not appreciably change the folding of this particular RNA.
To further address whether PEG acts through surface interactions with the RNA or through steric (excluded volume) effects, we also compared the ribozyme activity in Ficoll and sucrose. Ficoll 70, a highly branched polysaccharide (M r ϭ 70,000) of sucrose, is often used for molecular crowding studies  because the volume it occupies is roughly spherical. Surface interactions are expected to be larger for sucrose, which has a larger surface area than Ficoll for a given density. As shown in Fig. 6a, the ribozyme was more active in solutions containing 5% Ficoll than in 5% sucrose, suggesting that surface interactions are less important than excluded volume effects in stabilizing the native RNA.
When the initial phase amplitudes in Ficoll and sucrose as crowders were plotted as a function of MgCl 2 concentration in Fig. 6b, there was no significant difference between RNA in 5% sucrose and in buffer alone. By contrast, introducing Ficoll shifted the folding transition to lower MgCl 2 concentrations. This was generally consistent with the very slight stabilization of the folded ribozyme by ethylene glycol compared with PEG 8000 (6). This very important observation adds further evidence that volume excluded by high molecular weight solutes strongly stabilizes biologically active RNA structures.
Conclusions-We have demonstrated that the enzyme activity of the Azoarcus ribozyme is intact when its native structure is stabilized with a combination of MgCl 2 and PEG crowder molecules. Furthermore, increased burst amplitudes suggest that the proportion of molecules in the native state, as opposed to a compact but unreactive intermediate state, is actually greater in the crowded solution. We found that increasing the molecular weight of the synthetic crowder PEG from 1000 to 8000 has little effect on RNA folding. Ficoll 70 stabilizes the native structure more efficiently than its monomer unit sucrose, which we interpret as evidence that this stabilization is driven primarily by excluded volume (and therefore entropic) effects. Finally, the ribozyme reaction is slowed down in crowded solutions; these differences are accounted for when the rates are normalized by the inverse of viscosity. FIGURE 6. Comparison of Ficoll and sucrose. a, fraction of oligomer cleaved as a function of time for the prefolded L-3 Azoarcus ribozyme in 5% Ficoll or 5% sucrose solutions. The lines and error bars are as described for Fig. 2. b, initial phase amplitude versus MgCl 2 concentration for 5% Ficoll and sucrose. For comparison, data in 0% and 4% PEG 1000 are also shown. The lines are fits to a three-state Hill equation.