The Apo Riboswitch as a Molecular Hydra

Summary

Riboswitches ‘sense’ metabolites but knowledge is sparse for structures without bound ligand. Stoddard et al. (2010) determined an apo riboswitch structure ‘closed’ to metabolite binding. Further SAXS, biochemical and computational analyses support ensemble behavior with interconverting open and closed conformations.

It has been said that two heads are better than one, but not if you’re Hercules trying to slay the multi-headed Hydra. In Stoddard et al. (2010), Batey and co-workers come one step closer to hero status for their detailed molecular description and ensemble analysis of a metabolite-free variant of the S-adenosyl methionine (SAM)-I riboswitch. To use a mythological analogy, the riboswitch aptamer adopts multiple conformations in solution reminiscent of the Hydra’s many serpentine heads, which Hercules battled one at a time. In reality, researchers aspire to learn the factors that influence ligand binding and specificity, which is fundamental to understanding how riboswitches mediate gene regulation. In their quest, the Batey team compared their new ligand-free SAM-I aptamer to a prior crystal structure of a SAM-bound state (Montange and Batey, 2006). The results revealed surprisingly few changes in the global fold with significant differences on a local level. In the absence of metabolite, the SAM binding pocket rearranges by a 7.5 Å, placing Ade46 in a location that blocks recognition of the SAM adenine moiety (Figure 1). This “closed” conformation raises the question, how does the aptamer detect free SAM if the binding pocket is plugged? Insight into this question came in part from prior crystal structures of the lysine riboswitch determined in the metabolite-bound and free states (Garst et al., 2008; Serganov et al., 2008). Although lysine is buried in a solvent-sequestered pocket, the cavity remained closed even when the metabolite is absent. A plausible explanation is that the ligand-free structure is not wholly representative of the solution population but represents a stable conformation suited to crystallization. As such, the Batey team undertook the Herculean “labor” of finding evidence for ensemble behavior that accounts for both solution and crystallographic observations.

Figure 1.

Comparison of the T. tengcongensis metF-H2 SAM-I riboswitch aptamer structures determined in the presence and absence of metabolite.

Prior analyses of riboswitches in the metabolite free and bound states utilized small-angle x-ray scattering (SAXS) [reviewed in (Baird et al., 2010)]. This approach can provide sensitive measurements of global conformational differences in the presence and absence of ligands. Transformation of the scattering data into a real-space paired-distance distribution plot can provide information on the most probable scattering pair (R_G) and the maximum intramolecular distance (D_Max). These restraints, along with the observed SAXS profiles, can guide the restoration of molecular envelopes representative of the most probable solution conformation (Svergun, 1999). The envelopes in turn can direct rigid-body conformational changes to bring high-resolution structures into agreement with solution measurements. The resulting models can be compared directly to the experimental data for agreement – analogous to crystallographic R_factor calculations. Variations of this approach have been reported for the bound- and free-states of the respective thiamine pyrophosphate (TPP) and cyclic-di-GMP riboswitches (Ali et al., 2010; Kulshina et al., 2009).

The unique strategy of the Batey team was to incorporate ensemble optimization into their SAXS methodology to allow sampling of conformational states for the ligand-free SAM-I aptamer. Chemical modification and melting temperature (T_M) analyses were performed first to test the hypothesis that the apo riboswitch truly exhibits ensemble behavior. The authors used selective 2′-hydroxyl acylation analyzed by primer extension or SHAPE to probe the folding dependence in response to Mg²⁺ and SAM. In the presence of Mg²⁺ alone, multiple folding transitions were observed as a function of metal concentration. Significantly, at 10 mM Mg²⁺ there was evidence for residual unfolding. In contrast, nearly complete folding was observed over a narrow range of 0.5–1.0 mM Mg²⁺ in the presence of SAM. This result was also reflected in T_M measurements that demonstrated a SAM-dependent stabilization of the global fold. Changes in T_M monitored at each residue showed position-dependent effects in which tertiary structure regions, such as those in the SAM-binding pocket, exhibited greater metabolite-dependent melting differences than helical regions. Nucleotide analog interference mapping (NAIM) showed nucleotide variants that interfere with ligand binding map both inside and outside the SAM binding pocket, including locations like the kink turn. These results further support the requirement for a stable three-dimensional structure for SAM binding. Stoddard et al. (2010) interpreted the overall effects as ensemble behavior since many of the probing results showed a continuum of folding states ranging from unfolded to folded despite the absence of metabolite.

Not surprisingly poor agreement was obtained from efforts to fit the ligand-free SAM-I SAXS profiles with calculated scattering curves derived from the ligand-free crystal structure. To overcome this disparity, the Batey team generated a SAM-I riboswitch conformational-diversity library that was filtered by a genetic algorithm (Bernado et al., 2007). The resulting ensemble comprised thirteen structures with eleven “open” conformers that allow metabolite access to the pocket, and two “closed” conformers. The results provide an initial shape landscape that accounts for major, co-existing populations of aptamer-open and aptamer-closed conformations in which helices P1 and P3 move to the open state with a “scissor” action (Figure 2). Thus, instead of attacking each of the Hydra’s heads separately like Hercules, Stoddard et al. (2010) tackled their conformationally diverse “beasts” simultaneously.

Figure 2.

Cartoon diagram of the SAM-I riboswitch ensemble derived from SAXS in the absence of metabolite (colored tubes) overlayed on the crystal structure of the SAM-bound crystal structure (dark green ribbon). SAM is depicted as a red CPK model. Adapted from (Stoddard et al., 2010).

However, questions still remained such as, what pathways exist for interconversion of the SAM-I aptamer’s open and closed states? And what is the energetic barrier and timescale for interconversion of these states? To address these questions, the authors used replica-exchange molecular dynamics, which allows characterization of equilibrium thermodynamics. The results revealed that the open and closed states described in Figure 1 represent local energy minima with a barrier of 2–3 kcal mole⁻¹ for interconversion. Although the closed state is 1–1.5 kcal mol⁻¹ more stable than the open one, both conformations were sampled rapidly during the 1.7 μsec simulation providing a plausible pathway to produce metabolite-receptive states from unreceptive local folds.

Overall the work of the Batey team represents a laudable effort. Although SAXS ensemble analysis has been applied to proteins (Bernado et al., 2007) adaptation of these methods to the SAM-I riboswitch should open the door for related investigations. A broader take-home message may be that the SAM-I aptamer from T. tengcongensis maintains substantial global folding in the presence of Mg²⁺ alone as a means to rapidly select and respond to changing metabolite concentrations in the cell. Conformational heterogeneity plays an important role in this function and can be fine-tuned by minor sequence variations. This complexity is illustrated by the observation that riboswitches of the same class – but from different species – behave differently with respect to Mg²⁺-dependent folding in the absence of ligand (Baird et al., 2010). As such, it would appear that modern-day riboswitch heroes face many more “labors”, whereas Hercules was able to retire after only twelve.