The NMR and X-ray structures of the Saccharomyces cerevisiae Vts1 SAM domain define a surface for the recognition of RNA hairpins

Amodiaquine Nonspecifically Binds Double Stranded and Three-Way Junction DNA Structures

The 4-aminoquinoline class of compounds includes the important antimalarial compounds amodiaquine and chloroquine. Despite their medicinal importance, the mode of action of these compounds is poorly understood. In a previous study we observed these compounds, as well as quinine and mefloquine, tightly bind the DNA cocaine-binding aptamer. Here, we further explore the range of nucleic acid structures bound by these compounds. To gauge a wide range of binding affinities, we used isothermal titration calorimetry to explore high affinity binding (nM to tens of μM) and NMR spectroscopy to assay weak binding biding in the hundreds of micromolar range. We find that amodiaquine tightly binds all double stranded DNA structures explored. Mefloquine binds double stranded DNA duplex molecules tightly and weakly associates with a three-way junction DNA construct. Quinine and chloroquine only weakly bind duplex DNA but do not tightly bind any of the DNA constructs explored. A simulation of the free energy of binding of these ligands to the Dickerson-Drew dodecamer resulted in an excellent agreement between the simulated and experimental free energy. These results provide new insight into the DNA binding of clinically important antimalarial compounds and may play a role in future development of new antimalarials.

ResidualBind: Uncovering Sequence-Structure Preferences of RNA-Binding Proteins with Deep Neural Networks

Deep neural networks have demonstrated improved performance at predicting sequence specificities of DNA- and RNA-binding proteins. However, it remains unclear why they perform better than previous methods that rely on k-mers and position weight matrices. Here, we highlight a recent deep learning-based software package, called ResidualBind, that analyzes RNA-protein interactions using only RNA sequence as an input feature and performs global importance analysis for model interpretability. We discuss practical considerations for model interpretability to uncover learned sequence motifs and their secondary structure preferences.

DNA binding by the antimalarial compound artemisinin

Artemisinin (ART) is a vital medicinal compound that is used alone or as part of a combination therapy against malaria. ART is thought to function by attaching to heme covalently and alkylating a range of proteins. Using a combination of biophysical methods, we demonstrate that ART is bound by three-way junction and duplex containing DNA molecules. Binding of ART by DNA is first shown for the cocaine-binding DNA aptamer and extensively studied using this DNA molecule. Isothermal titration calorimetry methods show that the binding of ART is both entropically and enthalpically driven at physiological NaCl concentration. Native mass spectrometry methods confirm DNA binding and show that a non-covalent complex is formed. Nuclear magnetic resonance spectroscopy shows that ART binds at the three-way junction of the cocaine-binding aptamer, and that binding results in the folding of the structure-switching variant of this aptamer. This structure-switching ability was exploited using the photochrome aptamer switch assay to demonstrate that ART can be detected using this biosensing assay. This study is the first to demonstrate the DNA binding ability of ART and should lay the foundation for further work to study implications of DNA binding for the antimalarial activity of ART.

Global importance analysis: An interpretability method to quantify importance of genomic features in deep neural networks

Deep neural networks have demonstrated improved performance at predicting the sequence specificities of DNA- and RNA-binding proteins compared to previous methods that rely on k-mers and position weight matrices. To gain insights into why a DNN makes a given prediction, model interpretability methods, such as attribution methods, can be employed to identify motif-like representations along a given sequence. Because explanations are given on an individual sequence basis and can vary substantially across sequences, deducing generalizable trends across the dataset and quantifying their effect size remains a challenge. Here we introduce global importance analysis (GIA), a model interpretability method that quantifies the population-level effect size that putative patterns have on model predictions. GIA provides an avenue to quantitatively test hypotheses of putative patterns and their interactions with other patterns, as well as map out specific functions the network has learned. As a case study, we demonstrate the utility of GIA on the computational task of predicting RNA-protein interactions from sequence. We first introduce a convolutional network, we call ResidualBind, and benchmark its performance against previous methods on RNAcompete data. Using GIA, we then demonstrate that in addition to sequence motifs, ResidualBind learns a model that considers the number of motifs, their spacing, and sequence context, such as RNA secondary structure and GC-bias.

A Non-amyloid Prion Particle that Activates a Heritable Gene Expression Program

Spatiotemporal gene regulation is often driven by RNA-binding proteins that harbor long intrinsically disordered regions in addition to folded RNA-binding domains. We report that the disordered region of the evolutionarily ancient developmental regulator Vts1/Smaug drives self-assembly into gel-like condensates. These proteinaceous particles are not composed of amyloid, yet they are infectious, allowing them to act as a protein-based epigenetic element: a prion [SMAUG⁺]. In contrast to many amyloid prions, condensation of Vts1 enhances its function in mRNA decay, and its self-assembly properties are conserved over large evolutionary distances. Yeast cells harboring [SMAUG⁺] downregulate a coherent network of mRNAs and exhibit improved growth under nutrient limitation. Vts1 condensates formed from purified protein can transform naive cells to acquire [SMAUG⁺]. Our data establish that non-amyloid self-assembly of RNA-binding proteins can drive a form of epigenetics beyond the chromosome, instilling adaptive gene expression programs that are heritable over long biological timescales.

What is the best reference state for building statistical potentials in RNA 3D structure evaluation?

Knowledge-based statistical potentials have been shown to be efficient in protein structure evaluation/prediction, and the core difference between various statistical potentials is attributed to the choice of reference states. However, for RNA 3D structure evaluation, a comprehensive examination on reference states is still lacking. In this work, we built six statistical potentials based on six reference states widely used in protein structure evaluation, including averaging, quasi-chemical approximation, atom-shuffled, finite-ideal-gas, spherical-noninteracting, and random-walk-chain reference states, and we examined the six reference states against three RNA test sets including six subsets. Our extensive examinations show that, overall, for identifying native structures and ranking decoy structures, the finite-ideal-gas and random-walk-chain reference states are slightly superior to others, while for identifying near-native structures, there is only a slight difference between these reference states. Our further analyses show that the performance of a statistical potential is apparently dependent on the quality of the training set. Furthermore, we found that the performance of a statistical potential is closely related to the origin of test sets, and for the three realistic test subsets, the six statistical potentials have overall unsatisfactory performance. This work presents a comprehensive examination on the existing reference states and statistical potentials for RNA 3D structure evaluation.

Comprehensive and quantitative mapping of RNA-protein interactions across a transcribed eukaryotic genome

RNA-binding proteins (RBPs) control the fate of nearly every transcript in a cell. However, no existing approach for studying these posttranscriptional gene regulators combines transcriptome-wide throughput and biophysical precision. Here, we describe an assay that accomplishes this. Using commonly available hardware, we built a customizable, open-source platform that leverages the inherent throughput of Illumina technology for direct biophysical measurements. We used the platform to quantitatively measure the binding affinity of the prototypical RBP Vts1 for every transcript in the Saccharomyces cerevisiae genome. The scale and precision of these measurements revealed many previously unknown features of this well-studied RBP. Our transcribed genome array (TGA) assayed both rare and abundant transcripts with equivalent proficiency, revealing hundreds of low-abundance targets missed by previous approaches. These targets regulated diverse biological processes including nutrient sensing and the DNA damage response, and implicated Vts1 in de novo gene "birth." TGA provided single-nucleotide resolution for each binding site and delineated a highly specific sequence and structure motif for Vts1 binding. Changes in transcript levels in vts1Δ cells established the regulatory function of these binding sites. The impact of Vts1 on transcript abundance was largely independent of where it bound within an mRNA, challenging prevailing assumptions about how this RBP drives RNA degradation. TGA thus enables a quantitative description of the relationship between variant RNA structures, affinity, and in vivo phenotype on a transcriptome-wide scale. We anticipate that TGA will provide similarly comprehensive and quantitative insights into the function of virtually any RBP.

Increase in backbone mobility of the VTS1p-SAM domain on binding to SRE-RNA

The sterile alpha motif (SAM) domain of VTS1p, a posttranscriptional gene regulator, belongs to a family of SAM domains conserved from yeast to humans. Even though SAM domains were originally classified as protein-protein interaction domains, recently, it was shown that the yeast VTS1p-SAM and the SAM domain of its Drosophila homolog Smaug can specifically recognize RNA hairpins termed Smaug recognition element (SRE). Structural studies of the SRE-RNA complex of VTS1p-SAM revealed that the SAM domain primarily recognizes the shape of the RNA fold induced by the Watson-Crick base-pairing in the RNA pentaloop. Only the central G nucleotide is specifically recognized. The VTS1p-SAM domain recognizes SRE-RNAs with a CNGGN pentaloop where N is any nucleotide. The C1-G4 base pair in the wild type can be replaced by any pair of nucleotides that can form base pairs even though the binding affinity is greatest with a pyrimidine in position 1 and a purine in position 4. The interaction thus combines elements of sequence-specific and non-sequence-specific recognitions. The lack of structural rearrangements in either partner following binding is rather intriguing, suggesting that molecular dynamics may play an important role in imparting relaxed specificity with respect to the exact combination of nucleotides in the loop, except for the central nucleotide. In this work, we extend our previous studies of SRE-RNA interaction with VTS1p, by comparing the dynamics of the VTS1p-SAM domain both in its free form and when bound to SRE-RNA. The 15N relaxation studies of backbone dynamics suggest the presence of a dynamic interaction interface, with residues associated with specific G3 recognition becoming more rigid on RNA binding while other regions attain increased flexibility. The results parallel the observations from our studies of dynamics changes in SRE-RNA upon binding to VTS1p-SAM and shows that molecular dynamics could play a crucial role in modulating binding affinity and possibly contribute to the free energy of the interaction through an entropy-driven mechanism.

Involvement of Vts1, a structure-specific RNA-binding protein, in Okazaki fragment processing in yeast

The non-essential VTS1 gene of Saccharomyces cerevisiae is highly conserved in eukaryotes and encodes a sequence- and structure-specific RNA-binding protein. The Vts1 protein has been implicated in post-transcriptional regulation of a specific set of mRNAs that contains its-binding site at their 3'-untranslated region. In this study, we identified VTS1 as a multi-copy suppressor of dna2-K1080E, a lethal mutant allele of DNA2 that lacks DNA helicase activity. The suppression was allele-specific, since overexpression of Vts1 did not suppress the temperature-dependent growth defects of dna2Delta405N devoid of the N-terminal 405-amino-acid residues. Purified recombinant Vts1 stimulated the endonuclease activity of wild-type Dna2, but not the endonuclease activity of Dna2Delta405N, indicating that the activation requires the N-terminal domain of Dna2. Stimulation of Dna2 endonuclease activity by Vts1 appeared to be the direct cause of suppression, since the multi-copy expression of Dna2-K1080E suppressed the lethality observed with its single-copy expression. We found that vts1Delta dna2Delta405N and vts1Deltadna2-7 double mutant cells displayed synergistic growth defects, in support of a functional interaction between two genes. Our results provide both in vivo and in vitro evidence that Vts1 is involved in lagging strand synthesis by modulating the Dna2 endonuclease activity that plays an essential role in Okazaki fragment processing.

Mouse hepatitis virus stem-loop 2 adopts a uYNMG(U)a-like tetraloop structure that is highly functionally tolerant of base substitutions

Stem-loop 2 (SL2) of the 5'-untranslated region of the mouse hepatitis virus (MHV) contains a highly conserved pentaloop (C47-U48-U49-G50-U51) stacked on a 5-bp stem. Solution nuclear magnetic resonance experiments are consistent with a 5'-uYNMG(U)a or uCUYG(U)a tetraloop conformation characterized by an anti-C47-syn-G50 base-pairing interaction, with U51 flipped out into solution and G50 stacked on A52. Previous studies showed that U48C and U48A substitutions in MHV SL2 were lethal, while a U48G substitution was viable. Here, we characterize viruses harboring all remaining single-nucleotide substitutions in the pentaloop of MHV SL2 and also investigate the degree to which the sequence context of key pentaloop point mutations influences the MHV replication phenotype. U49 or U51 substitution mutants all are viable; C47 substitution mutants also are viable but produce slightly smaller plaques than wild-type virus. In contrast, G50A and G50C viruses are severely crippled and form much smaller plaques. Virus could not be recovered from G50U-containing mutants; rather, only true wild-type revertants or a virus, G50U/C47A, containing a second site mutation were recovered. These functional data suggest that the Watson-Crick edges of C47 and G50 (or A47 and U50 in the G50U/C47A mutant) are in close enough proximity to a hydrogen bond with U51 flipped out of the hairpin. Remarkably, increasing the helical stem stability rescues the previously lethal mutants U48C and G50U. These studies suggest that SL2 functions as an important, but rather plastic, structural element in stimulating subgenomic RNA synthesis in coronaviruses.

Discovering structural cis-regulatory elements by modeling the behaviors of mRNAs

Gene expression is regulated at each step from chromatin remodeling through translation and degradation. Several known RNA-binding regulatory proteins interact with specific RNA secondary structures in addition to specific nucleotides. To provide a more comprehensive understanding of the regulation of gene expression, we developed an integrative computational approach that leverages functional genomics data and nucleotide sequences to discover RNA secondary structure-defined cis-regulatory elements (SCREs). We applied our structural cis-regulatory element detector (StructRED) to microarray and mRNA sequence data from Saccharomyces cerevisiae, Drosophila melanogaster, and Homo sapiens. We recovered the known specificities of Vts1p in yeast and Smaug in flies. In addition, we discovered six putative SCREs in flies and three in humans. We characterized the SCREs based on their condition-specific regulatory influences, the annotation of the transcripts that contain them, and their locations within transcripts. Overall, we show that modeling functional genomics data in terms of combined RNA structure and sequence motifs is an effective method for discovering the specificities and regulatory roles of RNA-binding proteins.

Solution structure of stem-loop alpha of the hepatitis B virus post-transcriptional regulatory element

Chronic hepatitis B virus (HBV) infections may lead to severe diseases like liver cirrhosis or hepatocellular carcinoma (HCC). The HBV post-transcriptional regulatory element (HPRE) facilitates the nuclear export of unspliced viral mRNAs, contains a splicing regulatory element and resides in the 3′-region of all viral transcripts. The HPRE consists of three sub-elements α (nucleotides 1151–1346), β1 (nucleotides 1347–1457) and β2 (nucleotides 1458–1582), which confer together full export competence. Here, we present the NMR solution structure (pdb 2JYM) of the stem-loop α (SLα, nucleotides 1292–1321) located in the sub-element α. The SLα contains a CAGGC pentaloop highly conserved in hepatoviruses, which essentially adopts a CUNG-like tetraloop conformation. Furthermore, the SLα harbours a single bulged G residue flanked by A-helical regions. The structure is highly suggestive of serving two functions in the context of export of unspliced viral RNA: binding sterile alpha motif (SAM-) domain containing proteins and/or preventing the utilization of a 3′-splice site contained within SLα.

RNA recognition by the Vts1p SAM domain

The putative yeast post-transcriptional regulator Vts1p and its related protein Smaug, from Drosophila melanogaster, each use a sterile alpha motif (SAM) domain to bind an RNA hairpin termed the Smaug recognition element (SRE). Here, we present the NMR structures of the Vts1p-SRE complex and the free SRE. Structural highlights include the direct recognition of a guanine base and the formation or stabilization of a base pair in the SRE loop.

Sequence-specific recognition of RNA hairpins by the SAM domain of Vts1p

The SAM domain of the Saccharomyces cerevisiae post-transcriptional regulator Vts1p epitomizes a subfamily of SAM domains conserved from yeast to humans that function as sequence-specific RNA-binding domains. Here we report the 2.0-A X-ray structure of the Vts1p SAM domain bound to a high-affinity RNA ligand. Specificity of RNA binding arises from the association of a guanosine loop base with a shallow pocket on the SAM domain and from multiple SAM domain contacts to the unique backbone structure of the loop, defined in part by a nonplanar base pair within the loop. We have validated NNF1 as an endogenous target of Vts1p among 79 transcripts that copurify with Vts1p. Bioinformatic analysis of these mRNAs demonstrates that the RNA-binding specificity of Vts1p in vivo is probably more stringent than that of the isolated SAM domain in vitro.