Steve L. Bonilla

Functional RNA structures are typically far more dynamic than most proteins solved by cryo-electron microscopy (cryo-EM). Flexible junctions and weak tertiary contacts produce broad conformational ensembles, and cryo-EM particle classification analyses discard unaligned particles, which inherently contain the information needed to quantify such broad ensembles. Here, we infer compositional and conformational distributions from unclassified particle images by Bayesian ensemble reweighting. We tested our approach on the P4-P6 domain of the Tetrahymena thermophila group I intron, a classic model system of Mg²⁺-dependent RNA folding. While stable monomeric and dimeric states yielded high-resolution reconstructions, analyses of abundant, flexible conformations required Bayesian ensemble inference. Our approach quantitatively captured the Mg²⁺-dependent folding transition, indicating that cryo-EM can report the abundance of coexisting RNA states in agreement with bulk solution measurements. These results showcase how quantitative studies of RNA folding can be achieved by combining cryo-EM measurements with statistical ensemble-inference tools. SUMMARY: RNAs are remarkably versatile macromolecules that fold into dynamic three-dimensional (3D) structures to perform critical functions in cells and viruses, making them increasingly attractive target for therapeutics and as tools for biotechnology. Yet our understanding of RNA structure–function relationships remains limited, in part because structural biology tools are poorly suited to probing the dynamic ensembles that RNAs occupy. The raw 2D images collected in single-particle cryo-EM, however, inherently contain the information needed to visualize and quantify such ensembles. Here, we use the P4-P6 domain of the Tetrahymena thermophila group I intron, an archetypal model of Mg²⁺-dependent RNA folding, to leverage this information. While stable monomeric and dimeric states yielded high-resolution structures, flexible conformations escaped conventional 3D reconstructions—even though they dominate at physiological Mg²⁺ concentrations. To overcome this limitation, we integrated single-particle cryo-EM with a Bayesian reweighting framework that quantifies RNA ensemble distributions directly from raw 2D images. By statistically reweighting computational ensembles against experimental images via probabilistic modeling and deep learning, our approach quantitatively captures Mg²⁺-dependent population shifts between extended, compact, monomeric, and dimeric species, in agreement with fundamental principles of RNA folding and bulk solution measurements. Our framework supports the sampling and visualization of conformations within the ensemble, allows the quantification of errors, and demonstrates that the learned continuous ensemble fits the data significantly better than any set of discrete structures. These results showcase how quantitative RNA folding can be achieved by combining high-resolution structural biology measurements with statistical tools that treat RNAs as conformational ensembles.

Orthoflavivirus RNA genomes resist host 5′–3′ exoribonucleases to produce subgenomic flaviviral RNAs (sfRNAs). This resistance is conferred by exoribonuclease-resistant RNA (xrRNA) structures within the viral 3′ untranslated region that often occur in tandem, and whose function can be coupled. In dengue virus serotype 2 (DENV2), this coupling results in changing patterns of sfRNA identity and abundance associated with the ability of the virus to adapt to host vs. vector infections. The physical basis of this coupling was unknown. Using a combination of virology, biochemistry, bioinformatics, structural biology, and biophysics, we explored the structural and sequence determinants of tandem xrRNA coupling in DENV2. We discovered that the spatial proximity, order, and structural integrity of the tandem xrRNAs are all important for coupling. Furthermore, an unpaired A-rich linker that lies between the two xrRNAs is essential in stabilizing a specific structure that correlates to coupling. This A-rich sequence likely forms tertiary contacts with an adjacent stem-loop structure to form a physical bridge between the two xrRNAs, a finding that is supported by a mid-resolution cryo-electron microscopy (cryo-EM) map of the DENV2 tandem xrRNAs. Disruption of the structure of this bridge by mutation changes the relative orientation or spacing between the tandem xrRNAs, which is correlated to their functional coupling. These findings help provide an explanation for the coupling between tandem xrRNAs, suggesting a new mechanistic hypothesis in which the two tandem xrRNAs can simultaneously encounter Xrn1. IMPORTANCE: Dengue virus (DENV) generates non-coding subgenomic flaviviral RNAs (sfRNAs) that affect several cellular pathways and are important for successful infection. These sfRNAs are formed by structured RNA elements in the viral genome called exoribonuclease-resistant RNAs (xrRNAs), which fold into a distinct three-dimensional topology to block degradation by host cell exoribonucleases and often occur in tandem. Specific patterns of sfRNAs made during infection are important for host vs. vector fitness, and in DENV2, this pattern depends on functional coupling between tandem xrRNAs. However, the source of this functional coupling was unknown. We determined that an unpaired A-rich linker between the tandem xrRNAs is necessary for creating a structural bridge between the tandem xrRNAs. This bridge appears to favor a specific orientation between the tandem xrRNAs that is correlated to coupling and therefore to the patterns and relative abundance of sfRNAs produced during infection.

Viral RNA genomes contain structured elements that recruit the machinery necessary for their replication. However, these mechanisms remain poorly understood because of RNA structure’s dynamic nature. Using cryo-EM and single-molecule Förster resonance energy transfer, we show that the flaviviral stem-loop A (SLA)—which recruits the viral polymerase NS5 to initiate negative-strand synthesis—populates a conserved, multi-state conformational ensemble governed by a dynamic three-dimensional (3D) motif. Structures of Zika and dengue SLA–NS5 complexes reveal that NS5 engages a preorganized SLA conformation. Notably, destabilizing this conformation impairs negative-strand synthesis without affecting NS5 binding. Thus, the conserved SLA 3D ensemble guides a post-binding conformational search that converts an initial encounter complex into a productive replication-initiation complex—revealing how RNA conformational ensembles, rather than static structures, dictate function. SUMMARY: Viral genomes encode dynamic RNA structures that are specifically recognized by proteins to regulate critical steps of the viral life cycle. Although these RNAs exist as multi-state conformational ensembles, structural studies have largely captured single conformations, limiting our mechanistic understanding of RNA-protein recognition and function. Here, using single-particle cryo-electron microscopy (cryo-EM), single-molecule Förster resonance energy transfer, and biochemical assays, we show how flaviviral genomes use structural dynamics encoded in their stem-loop A (SLA) to guide a conserved, multi-step mechanism of negative-strand synthesis initiation by non-structural protein 5 (NS5). Cryo-EM ensembles of SLAs from dengue, Zika, West Nile, and yellow fever viruses reveal a conserved conformational landscape dictated by alternative stacking configurations of a central three-way junction (3WJ). Cryo-EM structures of SLA-NS5 complexes from dengue and Zika show that NS5 engages a specific state within SLA 3D ensemble, one in which the 3WJ adopts a GRR/A tetraloop-like motif that we identify across diverse structural RNAs. Strikingly, mutations that substantially destabilize this state impair negative-strand synthesis without affecting NS5 binding—decoupling binding from function and ruling out a simple conformational-capture mechanism. Instead, the SLA’s intrinsic structural dynamics direct a post-binding conformational search that converts an initial, conformationally heterogeneous encounter complex into a productive replication-initiation complex. These results demonstrate a mechanism of RNA-protein recognition in which the 3D conformational ensemble of an RNA guides formation of the functional state downstream of binding, and identify a ubiquitous, intrinsically dynamic 3D motif within the SLA’s 3WJ as a critical determinant of flavivirus negative-strand synthesis.

Orthoflaviviruses use programmed resistance to host 5′ to 3′ exoribonucleases to produce subgenomic flaviviral RNAs (sfRNAs) during infection. This resistance is conferred by exoribonuclease-resistant RNA (xrRNA) structures that often occur in tandem and whose function can be coupled. In dengue virus serotype 2 (DENV2) this coupling results in changing patterns of sfRNA identity and abundance linked to the ability of the virus to adapt to host vs. vector infections. The physical basis of this coupling was unknown. Using a combination of virology, biochemistry, bioinformatics, structural biology, and biophysics, we explored the structural and sequence determinants of tandem xrRNA coupling in DENV2. We discovered that the spatial proximity, order, and structural integrity of the tandem xrRNAs are all important for coupling. Furthermore, an unpaired A-rich linker that lies between the two xrRNAs is essential in stabilizing a specific structure that correlates to coupling. This A-rich sequence likely forms tertiary contacts with an adjacent stem-loop structure to form a physical bridge between the two xrRNAs, a finding that is supported by a mid-resolution cryoEM map of the DENV2 tandem xrRNAs. Disruption of the structure of this bridge by mutation changes the relative orientation or spacing between the tandem xrRNAs, which is correlated to their functional coupling. These findings provide an explanation for the coupling between tandem xrRNAs and suggests new mechanistic hypotheses. IMPORTANCE: Dengue virus (DENV) generates non-coding subgenomic flaviviral RNAs (sfRNAs) that affect several cellular pathways and are important for successful infection. These sfRNAs are formed by structured RNA elements in the viral genome called exoribonuclease-resistant RNAs (xrRNAs), which fold into a distinct three-dimensional topology to block degradation by host cell exoribonucleases and often occur in tandem. Specific patterns of sfRNAs made during infection are important for host vs. vector fitness, and in DENV2 this pattern depends on functional coupling between tandem xrRNAs. However, the source of this functional coupling was unknown. We determined that an unpaired A-rich linker between the tandem xrRNAs is necessary for creating a structural bridge between the tandem xrRNAs. This bridge appears to favor a specific orientation between the tandem xrRNAs that is correlated to coupling and therefore to the patterns and relative abundance of sfRNAs produced during infection.

Folded RNAs contain tertiary contact motifs whose structures and energetics are conserved across different RNAs. The transferable properties of RNA motifs simplify the RNA folding problem, but measuring energetic and conformational properties of many motifs remains a challenge. Here, we use a high-throughput thermodynamic approach to investigate how sequence changes alter the binding properties of naturally occurring motifs, the GAAA tetraloop • tetraloop receptor (TLR) interactions. We measured the binding energies and conformational preferences of TLR sequences that span mutational pathways from the canonical 11ntR to two other natural TLRs, the IC3R and Vc2R. While the IC3R and Vc2R share highly similar energetic and conformational properties, the landscapes that map the sequence changes for their conversion from the 11ntR to changes in these properties differ dramatically. Differences in the energetic landscapes stem from the mutations needed to convert the 11ntR to the IC3R and Vc2R rather than a difference in the intrinsic energetic architectures of these TLRs. The conformational landscapes feature several nonnative TLR variants with conformational preferences that differ from both the initial and final TLRs; these species represent potential branching points along the multidimensional sequence space to sequences with greater fitness in other RNA contexts with alternative conformational preferences. Our high-throughput, quantitative approach reveals the complex nature of sequence-fitness landscapes and leads to models for their molecular origins. Systematic and quantitative molecular approaches provide critical insights into understanding the evolution of natural RNAs as they traverse complex landscapes in response to selective pressures.

Folded RNAs contain tertiary contact motifs whose structures and energetics are conserved across different RNAs. The transferable properties of RNA motifs simplify the RNA folding problem, but measuring energetic and conformational properties of many motifs remains a challenge. Here, we use a high-throughput thermodynamic approach to investigate how sequence changes alter the binding properties of naturally-occurring motifs, the GAAA tetraloop • tetraloop receptor (TLR) interactions. We measured the binding energies and conformational preferences of TLR sequences that span mutational pathways from the canonical 11ntR to two other natural TLRs, the IC3R and Vc2R. While the IC3R and Vc2R share highly similar energetic and conformational properties, the landscapes that map the sequence changes for their conversion from the 11ntR to changes in these properties differ dramatically. Differences in the energetic landscapes stem from the mutations needed to convert the 11ntR to the IC3R and Vc2R rather than a difference in the intrinsic energetic architectures of these TLRs. The conformational landscapes feature several non-native TLR variants with conformational preferences that differ from both the initial and final TLRs; these species represent potential branching points along the multidimensional sequence space to sequences with greater fitness in other RNA contexts with alternative conformational preferences. Our high-throughput, quantitative approach reveals the complex nature of sequence-fitness landscapes and leads to models for their molecular origins. Systematic and quantitative molecular approaches provide critical insights into understanding the evolution of natural RNAs as they traverse complex landscapes in response to selective pressures.

Structured RNAs and RNA/protein complexes perform critical cellular functions. They often contain structurally conserved tertiary contact “motifs,” whose occurrence simplifies the RNA folding landscape. Prior studies have focused on the conformational and energetic modularity of intact motifs. Here, we turn to the dissection of one common motif, the 11nt receptor (11ntR), using quantitative analysis of RNA on a massively parallel array to measure the binding of all single and double 11ntR mutants to GAAA and GUAA tetraloops, thereby probing the energetic architecture of the motif. While the 11ntR behaves as a motif, its cooperativity is not absolute. Instead, we uncovered a gradient from high cooperativity amongst base-paired and neighboring residues to additivity between distant residues. As expected, substitutions at residues in direct contact with the GAAA tetraloop resulted in the largest decreases to binding, and energetic penalties of mutations were substantially smaller for binding to the alternate GUAA tetraloop, which lacks tertiary contacts present with the canonical GAAA tetraloop. However, we found that the energetic consequences of base partner substitutions are not, in general, simply described by base pair type or isostericity. We also found exceptions to the previously established stability–abundance relationship for 11ntR sequence variants. These findings of “exceptions to the rule” highlight the power of systematic high-throughput approaches to uncover novel variants for future study in addition to providing an energetic map of a functional RNA.

Structured RNAs and RNA/protein complexes perform critical cellular functions. They often contain structurally conserved tertiary contact “motifs,” whose occurrence simplifies the RNA folding landscape. Prior studies have focused on the conformational and energetic modularity of intact motifs. Here we turn to the dissection of one common motif, the 11nt receptor (11ntR), using quantitative analysis of RNA on a massively parallel array (RNA-MaP) to measure the binding of all single and double 11ntR mutants to GAAA and GUAA tetraloops, thereby probing the energetic architecture of the motif. While the 11ntR behaves as a motif, its cooperativity is not absolute. Instead, we uncovered a gradient from high cooperativity amongst base paired and neighboring residues to additivity between distant residues. As expected, substitutions at residues in direct contact with the GAAA tetraloop resulted in the largest decreases to binding, and energetic penalties of mutations were substantially smaller for binding to the alternate GUAA tetraloop, which lacks tertiary contacts present with the canonical GAAA tetraloop. However, we found that the energetic consequences of base partner substitutions are not, in general, simply described by base pair type or isostericity. We also found exceptions to the previously established stability-abundance relationship for 11ntR sequence variants. These findings of “exceptions to the rule” highlight the power of systematic high-throughput approaches to uncover novel variants for future study in addition to providing an energetic map of a functional RNA. Significance Statement Properly folded RNAs perform essential biological processes. Many RNAs contain tertiary contact “motifs” whose structural and energetic properties are conserved across different RNAs. This study delves into the nature of an RNA motif. We determined the effects of mutations on the energetic properties of a common tertiary contact motif, referred to as the 11-nucleotide receptor. As deep study of the energetic architecture of this and other RNAs requires thermodynamic measurements for many sequence variants, we used RNA-MaP, a quantitative, high-throughput method, to obtain binding free energies for all single and double mutants of the motif. Our results revealed the energetic architecture of this motif and identified rare variants with unexpected properties.