Patent Application
20220259590
The present disclosure relates to ribonucleic acid (RNA) aptamers, and in particular methods and systems to design RNA aptamers for increased stability and/or function.
A computer readable form of the sequence listing, “06060.PRO Construct Sequences_ST25.txt”, submitted via EFS-WEB, is herein incorporated by reference in its entirety.
RNA-based nanotechnology is an emerging field that harnesses RNA's unique structural properties to create novel nanostructures and machines. Perhaps more so than for other biomolecules, RNA tertiary structure is composed of discrete and recurring components known as tertiary ‘motifs’. Along with the helices that they interconnect, many of these structural motifs appear highly modular; that is, each motif folds into a well-defined three-dimensional (3D) structure in a broad range of contexts. By exploiting symmetry, motif repetition, and expert modeling, these motifs have been assembled into novel polyhedra, sheets, and cargo-carrying nanoparticles for biomedical use. Despite these advances, current methods still rely on human intuition in conjunction with simple visualization tools and the field is far from generating RNAs as sophisticated as natural RNA machines, which are asymmetric, too large to be solved by 3D RNA structure prediction methods, and composed of vast repertoires of distinct interacting motifs, most of which are not yet well characterized. (See Guo, P. (2010) The emerging field of RNA nanotechnology. Nat. Nanotechnol. 5, 833-842; Grabow, W. W., and Jaeger, L. (2014) RNA self-assembly and RNA nanotechnology. Acc. Chem. Res. 47, 1871-1880; Leontis, N. B., et al. (2006) The building blocks and motifs of RNA architecture. Curr. Opin. Struct. Biol. 16, 279-287; Jaeger, L., and Chworos, A. (2006) The architectonics of programmable RNA and DNA nanostructures. Curr. Opin. Struct. Biol. 16, 531-543; Jaeger, L., and Leontis, N. B. (2000) Tecto-RNA: One-Dimensional Self-Assembly through Tertiary Interactions. Angew. Chem. Int. Ed. Engl. 39, 2521-2524; Zhang, H., et al. (2013) Crystal structure of 3WJ core revealing divalent ion-promoted thermostability and assembly of the Phi29 hexameric motor pRNA. RNA 19, 1226-1237; Weizmann, Y., and Andersen, E. S. (2017) RNA nanotechnology—The knots and folds of RNA nanoparticle engineering. MRS Bull. 42, 930-935; Jasinski, D., et al. (2017) Advancement of the emerging field of RNA nanotechnology. ACS Nano 11, 1142-1164; Bindewald, E., et al. (2008) Computational strategies for the automated design of RNA nanoscale structures from building blocks using NanoTiler. J Mol Graph Model 27, 299-308; Jossinet, F., et al. (2010) Assemble: an interactive graphical tool to analyze and build RNA architectures at the 2D and 3D levels. Bioinformatics 26, 2057-2059; Wimberly, B. T., et al. (2000) Structure of the 30S ribosomal subunit. Nature 407, 327-339; Nguyen, T. H. D., et al. (2015) The architecture of the spliceosomal U4/U6.U5 tri-snRNP. Nature 523, 47-52; and Miao, Z., et al. (2017) RNA-Puzzles Round III: 3D RNA structure prediction of five riboswitches and one ribozyme. RNA 23, 655-672; the disclosures of which are incorporated herein by reference in their entirety.)
Additionally, aptamer selection suffers from two critical limitations that prevent its use in engineering scaffolds that do not require target protein reengineering. First, selection experiments are limited by the number of sequences that can be tested, which results in many cases where high quality aptamers cannot be selected. (See e.g., Wang, J. P., et al., Influence of Target Concentration and Background Binding on In Vitro Selection of Affinity Reagents. Plos One, 2012. 7(8); and Gold, L., et al., Aptamer-Based Multiplexed Proteomic Technology for Biomarker Discovery. Plos One, 2010. 5(12); the disclosures of which are incorporated by reference herein in their entireties.) Second, the structure of the aptamer cannot be explicitly controlled, which is undesirable when the goal is to generate an aptamer that can be used to precisely orient proteins relative to each other.
This summary is meant to provide examples and is not intended to be limiting of the scope of the invention in any way. For example, any feature included in an example of this summary is not required by the claims, unless the claims explicitly recite the feature. Also, the features described can be combined in a variety of ways. Various features and steps as described elsewhere in this disclosure can be included in the examples summarized here.
In one embodiment, a method of designing an RNA nanostructure, includes generating a motif library describing a plurality of structural motifs, and designing a candidate path between two points of RNA using individual motifs from the motif library.
In a further embodiment, the motif library includes canonical motifs and noncanonical motifs.
In another embodiment, the canonical motifs are double stranded RNA helix motifs of variable length.
In a still further embodiment, the canonical motifs range in size from 1-22 bp.
In still another embodiment, the noncanonical motifs include one or more of the group consisting of two-way junctions, higher-order junctions, variable-length hairpins, tertiary contacts, and multi-way junctions.
In a yet further embodiment, the designing step includes integrating an aptamer into the candidate path.
In yet another embodiment, the designing step is performed in a depth-first manner.
In a further embodiment again, the candidate path is based on motif structure.
In another embodiment again, the method further includes filling in the candidate path with sequences that best match a target secondary structure.
In a further additional embodiment, the filling in step uses sequences that minimize alternative secondary structures.
In another additional embodiment, the designing step generates a plurality of candidate paths.
In a still yet further embodiment, the method further includes filtering the plurality of candidate paths based on at least one limitation.
In still yet another embodiment, the at least one limitation is selected from the group consisting of minimum number of motifs, maximum number of motifs, minimum number of residues, maximum number of residues, minimum stability, and maximum stability.
In a still further embodiment again, the method further includes synthesizing an oligonucleotide covering the design of the candidate path.
In still another embodiment again, an RNA nanostructure comprises a plurality of RNA motifs aligned end to end forming a chain, where the plurality of RNA motifs are selected from the group consisting of canonical RNA motifs and noncanonical RNA motifs.
In a still further additional embodiment, the plurality of RNA motifs alternate between canonical RNA motifs and noncanonical RNA motifs.
In still another additional embodiment, the RNA nanostructure further includes an anchor structure connected to one end of the chain.
In a yet further embodiment again, the RNA nanostructure further includes two anchor structures, where one anchor structure is connected to one end of the chain, and the other anchor structure is connected to the other end of the chain.
In yet another embodiment again, the two anchor structures are a tetraloop and a tetraloop receptor.
In a yet further additional embodiment, the RNA nanostructure further includes an anchor structure, wherein the plurality of RNA motifs are connected to one end of the anchor structure, and at least one more RNA motif is connected to the other end of the anchor structure.
In yet another additional embodiment, the anchor structure is an aptamer.
In a further additional embodiment again, the canonical RNA motifs are double stranded RNA helix motifs.
In another additional embodiment again, the canonical RNA motifs range in size from 1 base pair to 100 base pairs.
In a still yet further embodiment again, the canonical RNA motifs range in size from 1 base pair to 22 base pairs.
In still yet another embodiment again, the noncanonical RNA motifs are selected from the group consisting of: two-way junctions, higher-order junctions, variable-length hairpins, tertiary contacts, and multi-way junctions.
The foregoing and other objects, features, and advantages of the disclosed technology will become more apparent from the following detailed description, which proceeds with reference to the accompanying figures.
Turning now to the drawings and data, embodiments herein represent a novel approach to 3D RNA design, based on the recognition that numerous recurring problems in the field can be cast into a ‘pathfinding’ problem. (See
First, a founding problem of RNA nanotechnology involves designing a compact nanostructure that aligns the two parts of the tetraloop/tetraloop-receptor (TTR) so that they can form a tertiary contact upon RNA chain folding (
A second problem is highly analogous to the TTR stabilization problem but is more difficult. Efforts to select engineered ribosomes with mRNA decoding, polypeptide synthesis, and protein excretion functions optimized for new substrates might be dramatically accelerated through the design of integrated ribosomes. An important step towards this goal involves tethering the two 23S and 16S rRNAs of the ribosome into a single RNA strand that supports E. coli growth. (See Fried, S. D., et al. (2015) Ribosome subunit stapling for orthogonal translation in E. coli. Angew. Chem. Int. Ed. Engl. 54, 12791-12794; Orelle, C., et al. (2015) Protein synthesis by ribosomes with tethered subunits. Nature 524, 119-124; Carlson, E. D. (2015) Creating Ribo-T: (Design, Build, Test)n. ACS Synth. Biol. 4, 1173-1175; and Schmied, W. H., et al. (2018) Controlling orthogonal ribosome subunit interactions enables evolution of new function. Nature 564, 444-448; the disclosures of which are incorporated herein by reference in their entirety.) Three-dimensional designs for a tether (106) would require solving the RNA motif pathfinding problem (108) over >100 Å distances and avoiding steric collisions with the ribosome's RNA and protein components (110,
A third problem involves a more complex instance of two RNA motif pathfinding problems (112,
Additional issues exist in protein scaffolding. Scaffold proteins physically link individual molecules to increase the efficiency of their interaction and have been found to be critical to many cellular signaling processes. (See e.g., Good, M. C., et al., Scaffold proteins: hubs for controlling the flow of cellular information. Science, 2011. 332(6030): p. 680-6; the disclosure of which is incorporated by reference herein in its entirety.) Engineers have realized the potential of these scaffold molecules to reshape cellular behavior and have redesigned scaffold proteins for several applications including altering MAP kinase pathway signaling dynamics and enhancing production of specific metabolites. (See e.g., Dueber, J. E., et al., Synthetic protein scaffolds provide modular control over metabolic flux. Nature Biotechnology, 2009. 27(8): p. 753-U107; and Bashor, C. J., et al., Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics. Science, 2008. 319(5869): p. 1539-1543; the disclosures of which is incorporated by reference herein in their entirety.) Synthetic RNA molecules offer increased design flexibility over protein scaffolds and have also been used to spatially arrange proteins to increase metabolic pathway yields and control synthetic transcriptional programs. (See e.g., Delebecque, C. J., et al., Designing and using RNA scaffolds to assemble proteins in vivo. Nature Protocols, 2012. 7(10): p. 1797-1807; Delebecque, C. J., et al., Organization of Intracellular Reactions with Rationally Designed RNA Assemblies. Science, 2011. 333(6041): p. 470-474; Zalatan, J. G., et al., Engineering Complex Synthetic Transcriptional Programs with CRISPR RNA Scaffolds. Cell, 2015. 160(1-2): p. 339-350; and Sachdeva, G., et al., In vivo co-localization of enzymes on RNA scaffolds increases metabolic production in a geometrically dependent manner. Nucleic Acids Research, 2014. 42(14): p. 9493-9503; the disclosures of which is incorporated by reference herein in their entirety.) However, both engineered RNA and protein scaffolds rely on known protein-protein or protein-RNA interactions and thus require protein- or RNA-binding proteins to be fused to the proteins to be scaffolded. This requirement precludes the use of scaffolds for therapeutic applications and makes it much more difficult to control the precise three-dimensional arrangement of the scaffolded proteins.
Turning to
At 204, certain embodiments design a candidate RNA structure, or candidate path, connecting two points of RNA, where the path is comprised of one or more RNA motifs in the one or more motif libraries. In this 204, connection points are defined to be linked. These connection points can be on one or more RNA molecules, such as to link two RNA molecules together or to link two ends of a single RNA molecule.
Various embodiments perform the path designing in a step-by-step in a depth-first manner, where a first motif is joined to a first point to achieve the closest distance to a second point prior to a second motif being added, then a third motif is added to achieve the closest distance to the terminating point. This process is performed, until a candidate path is designed between the first and second points. In various embodiments, the pathfinding will be performed in a bidirectional manner, such that candidate paths will generated starting at the first point and terminating at the second point in addition to candidate paths being generated starting at the second point and terminating at the first point. Additional embodiments will further always begin with a canonical motif, and some embodiments will always end with a canonical motif. Some embodiments will further alternate canonical and noncanonical motifs until a candidate path is identified. Further embodiments will allow for specific settings, such that canonical motifs are selected for larger lengths, while noncanonical motifs are selected for smaller lengths. An illustration of this pathfinding process is illustrated in
Further embodiments will design the path using structures of specific motifs rather than the RNA sequence of the specific motif to be included into the path. For example, some embodiments will allow a user to specify a specific RNA structure (e.g., an RNA aptamer) to be included in the path in lieu of a canonical or noncanonical motif. In embodiments incorporating a specific RNA structure, the method 200 incorporates a de novo scaffold around the existing structure, which will result in a structure that is more stable and active (in the case of functional structures). This pathway runs counter to prevailing methodologies (discussed further below), which attempt to place RNA structures into known scaffolds, thus plugging such structures into preconstructed scaffolds, which require vast amounts of effort without much success in generating functional scaffolds.
In 206, if the candidate path was found based on structure, many embodiments will fill in the candidate path with sequences that best match the target secondary structure. Additional embodiments will fill in the candidate path with sequences that minimize alternative secondary structures.
Once the candidate path sequences have been identified, many embodiments filter the candidate paths at 208. In 208, factors or limitations are utilized to limit the total output of method 200. Such factors include minimum and/or maximum number of motifs (e.g., canonical motifs and noncanonical motifs), minimum and/or maximum number of residues (e.g., the number of bases in the entire RNA strand), and/or minimum and/or maximum stability (e.g., number of Watson-Crick base pairs).
At 210 of certain embodiments, oligonucleotides are synthesized representing the designed RNA nanostructure. Various embodiments synthesize the RNA nanostructure chemically via various known technologies, while additional embodiments synthesize the RNA nanostructure via biochemical. Example methods of synthesis include phosphoramidite, T7 polymerase, and any other known or applicable means of synthesizing an RNA nanostructure. In various embodiments, the oligonucleotides will include just the developed path from a starting point to an ending point, while in some embodiments, the oligonucleotide includes a portion (including the entirety) of the molecule at the starting point and/or a portion (including the entirety) of the molecule at the ending point. Certain embodiments will synthesize the oligonucleotide using RNA base pairs, while some embodiments will synthesize the oligonucleotide using DNA base pairs, and additional embodiments will synthesize the oligonucleotide using a combination of RNA and DNA base pairs. Further, embodiments synthesize the oligonucleotide double stranded, single stranded, or a combination of double and single stranded.
At 212, the RNA nanostructure is put into use. Using an RNA nanostructure can include a number of uses, such as a medicament or to enhance RNA function, such as the means described in depth below.
It should be noted that in numerous embodiments, some components in method 200 will be performed in a different order, performed simultaneously with prior components, and/or omitted. For example, filtering 208 can be completed simultaneously with the pathfinding 204, such that once a path reaches a certain point (e.g., a maximum length and/or a maximum number of motifs) the path is eliminated, and another path is begun. Additionally, if the motif libraries are based on sequence, 206 will be omitted in some embodiments, as there will be no need to fill in the sequence.
Certain embodiments of method 200 are implemented on non-transitory machine readable media, where method 200 is encoded as processor instructions. In many of these embodiments, execution of the processor instructions by a processor causes the processor to perform one or more steps embodied in method 200. Additional embodiments are further directed to systems comprising a processor and memory, where the memory contains instructions that when read by the processor direct the processor to perform one or more steps embodied in method 200.
When implemented on a computer, certain embodiments of method 200 scale linearly with problem size (e.g., distance between starting and ending points). Some embodiments will be performed on a consumer-grade computer (e.g., laptop computer), and
The resulting products of method 200 possess a number of characteristics, including the ability to fold properly, traverse long distances, and/or hold aptamers into a functional conformation.
Various embodiments possess the ability for the RNA nanostructure to properly fold upon synthesis.
aPercent of helical residues that have SHAPE and DMS reactivities < 0.5 reactivity units, suggesting they are in base pairs.
bFor DMS chemical mapping with and without 10 mM Mg2+, a 2-fold reduction in mean DMS reactivity at the four TTR adenines was considered to pass screen.
cDistance traveled in gel of RNA compared to mutant with tetraloop GAAA changed to UUCG. Positive numbers correspond to faster gel mobility (more compact fold) with wild type tetraloop, as expected for correctly folded RNA.
dRNA that was more than half folded with [Mg2+] < 10 mM was considered to pass screen
Various embodiments have the ability to link molecules across long distances.
Additional embodiments generate structures including multi-way junctions. An example of such embodiments is illustrated in
RNA aptamers possess the ability to bind small molecules. Unfortunately, prior methods to improve RNA aptamer function have largely been unsuccessful by producing weakened binding affinity or instability in biological environments. Even after multiple rounds of improvement, many prior attempts resulted in diminishing returns. (See, e.g., Carothers, J. M., et al. (2006) Aptamers selected for higher-affinity binding are not more specific for the target ligand. J. Am. Chem. Soc. 128, 7929-7937; Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646; and Ellington, A. D., and Szostak, J. W. (1990) In vitro selection of RNA molecules that bind specific ligands. Nature 346, 818-822; the disclosures of which are incorporated herein by reference in their entirety.) As such, various embodiments allow for the introduction of RNA aptamers into an RNA nanostructure. Examples of such activity are illustrated below in
aDecrease in reactivity beyond 0.2 exceeds experimental error and considered evidence for ATP binding at ATP aptamer. Values normalized to DMS reactivity of single-stranded adenosines in reference GAGUA hairpins flanking design.
bMean DMS reactivity less than 0.5 taken as evidence for tetraloop/tetraloop-receptor (TTR) formation.
cFold change in DMS reactivity with and without ATP. If both the mean reactivity is under 0.5 and the fold change is under 2 it is considered a success.
dKd lower than reference ATP aptamer demonstrated successful stabilization of ATP aptamer.
eChemical mapping data for ATP-TTR 1 and 2 could not be processed due to strong stops on the capillary electrophoresis readout.
Additionally, the Spinach RNA aptamer binds an analog of the green fluorescent protein chromophore (Z)-4-(3,5-Difluoro-4-hydroxybenzylidene)-1,2-dimethyl-1H-imidazol-5(4H)-one (DFHBI) within a G-quadruplex. Binding to Spinach enhances the fluorescence of DFHBI by ˜1,000-fold relative to unbound ligand, making this RNA useful for biological interrogations. (See Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646 and Kellenberger, C. A., et al. (2015) RNA-Based Fluorescent Biosensors for Live Cell Imaging of Second Messenger Cyclic di-AMP. J. Am. Chem. Soc. 137, 6432-6435; the disclosures of which are incorporated herein by reference in their entirety.) However, the binding affinity, brightness, folding efficiency and biological stability remain poor even after extensive efforts to discover improvements such as the minimized Spinach and Broccoli aptamers. (See Strack, R. L., et al. (2013) A superfolding Spinach2 reveals the dynamic nature of trinucleotide repeat-containing RNA. Nat. Methods 10, 1219-1224; Filonov, G. S., et al. (2014) Broccoli: rapid selection of an RNA mimic of green fluorescent protein by fluorescence-based selection and directed evolution. J. Am. Chem. Soc. 136, 16299-16308; Ketterer, S., et al. (2015) Systematic reconstruction of binding and stability landscapes of the fluorogenic aptamer spinach. Nucleic Acids Res. 43, 9564-9572; and Song, W., et al. (2014) Plug-and-play fluorophores extend the spectral properties of Spinach. J. Am. Chem. Soc. 136, 1198-1201; the disclosures of which are incorporated herein by reference in their entirety.)
Turning to
A number of embodiments are directed to RNA aptamers to scaffold proteins. In some embodiments, the methods are biased toward sequences that form favorable interactions with target proteins and adopt specific three-dimensional structures. Various embodiments design sequence libraries for in vitro selection experiments. Turning to
Many embodiments start with a target protein 902, then computationally identify optimal RNA-binding regions 904 on the surface of the target protein, then design small “anchor” RNA structures 906 that bind to these regions, likely with low affinity, and finally design RNA structures 908 that connect the anchors. In further embodiments, the affinity of the designed structures are improved by randomizing specific regions and performing selection experiments.
Many embodiments identify RNA/protein binding regions by predicting interaction sites between RNA structures and regions on proteins. Certain embodiments utilize a custom scoring function to discriminate between native and non-native structures, where different structures can be calculated as equation 1:
−kT In(P(structure|sequence)) (eq. 1)
The embodiments utilize an expression for the probability of a structure given its primary sequence (e.g., P(structure|sequence)). In particular, the probability of each monomer in an overall complex structure, such as given in equation 2:
P(M1,M2,C|sequence)=P(C|M1,M2,sequence) P(M1,M2,sequence) P(M2|sequence) (eq. 2)
where M1 is the structure of the RNA monomer 1, M2 is the structure of the protein monomer 2, and C is the structure of the complex.
Assuming that P(M1|M2, sequence) is approximately equal to P(M1|sequence), the equation becomes equation 3:
P(M1,M2,C|sequence)=P(C|M1,M2,sequence) P(RNA structure|sequence) P(protein structure|sequence) (eq. 3)
The energy of the RNA/Protein complex is further given by equation 4:
E(M1,M2,C|sequence)=−kT In(P(C|M1,M2, sequence))+ScoreRNA+Scoreprotein (eq. 4)
Medium resolution potentials for both ScoreRNA and Scoreprotein have been previously worked out and implemented within Rosetta. (See Das, R., et al., Atomic accuracy in predicting and designing noncanonical RNA structure. Nat Methods, 2010. 7(4): p. 291-4; Simons, K. T., et al., Assembly of protein tertiary structures from fragments with similar local sequences using simulated annealing and Bayesian scoring functions. Journal of Molecular Biology, 1997. 268(1): p. 209-225; Simons, K. T., et al., Improved recognition of native-like protein structures using a combination of sequence-dependent and sequence-independent features of proteins. Proteins-Structure Function and Genetics, 1999. 34(1): p. 82-95; and Das, R. and D. Baker, Automated de novo prediction of native-like RNA tertiary structures. Proceedings of the National Academy of Sciences of the United States of America, 2007. 104(37): p. 14664-14669; the disclosures of which are incorporated herein by reference in their entireties.) Additionally, the expression for P(C|M1, M2, sequence) can be decomposed similar to protein-protein docking in equation 5: (See Gray, J. J., et al., Protein-protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations. Journal of Molecular Biology, 2003. 331(1): p. 281-299; the disclosure of which is incorporated herein by reference in its entirety.)
where P(sequence|M1, M2) is constant and can be neglected. Additionally, P(sequence|C, M1, M2) can be expanded following framework outlined for knowledge-based protein score function in Rosetta, as in equation 6: (See
The first term is the residue environment term (Senv) and the second term is the residue pair term (Spair). The environments are defined as interface or non-interface and for proteins buried or exposed and for RNA base-paired or not base-paired. Many embodiments use a coarse-grained representation of both the protein and RNA residues in which the sidechains are represented as a single centroid atom. Accordingly, the distances in this potential are computed between these centroid atoms.
P(C|M1, M2) is the sequence-independent part of the interaction and includes terms describing well-formed complexes. To start, this include two terms approximating the attractive and repulsive parts of van der Waals interactions in equation 7:
P(C|M1,M2)˜e−S
Scontact is proportional to the number of residues between the two monomers that are within an optimal distance range to be determined from the training set of structures described below. Sclash is calculated using atom type dependent distance cutoffs, dij0 determined from the training set following the same method as for the protein potential in equation 8:
S
clash=(dij0)2−(dij)2 (eq. 8)
This leads to a final expression for the protein-RNA score function in equation 9:
E(M1,M2,C|sequence)=wenvSenv+wpairSpair+wcontactScontact+wclashSclash+wRNAScoreRNA+wproteinScoreprotein (eq. 9)
where wenv, wpair, wcontact, wclash, wRNA, and wprotein are weights that are fit to optimize prediction of native structures.
The probabilities of protein/RNA interactions, used to derive Senv, Spair, Scontact, and Sclash is approximated from the frequencies of these interactions in the non-redundant set of protein/RNA structures found in the Protein Database (PDB). As of June 2016, there are 1283 crystal structures containing both protein and RNA chains, with resolution better than 3.5 Å and less than 70% sequence identity. Additional embodiments further refine the set of structures to ensure it only contains non-redundant structures where the protein and RNA are in the same biological unit.
The proposed form of P(C|M1, M2) described here may be insufficient for successful discrimination of native complexes. The protein/RNA complexes from the PDB are analyzed in certain embodiments to identify additional structural features of well-formed RNA/protein complexes such as possible orientation preferences of secondary structure elements. Some embodiments include systematically testing the inclusion of these additional terms to find the score function that best predicts correctly formed protein/RNA structures.
At 906 of many embodiments, small “anchor” RNA structures are designed at 906 of many embodiments. RNA binding proteins with high affinity for their RNA targets are often composed of many modules, each of which binds a short RNA sequence with relatively low affinity. (See e.g., Lunde, B. M., et al., RNA-binding proteins: modular design for efficient function. Nature Reviews Molecular Cell Biology, 2007. 8(6): p. 479-490; the disclosure of which is incorporated by reference herein in its entirety.) Various embodiments design high affinity protein binding RNA aptamers. De novo design of these structures can be accomplished through two different paths in accordance with various embodiments. Some embodiments design small “anchor” RNA structures that bind weakly to specific protein surfaces, while additional embodiments design connecting RNA structures. Certain embodiments combine these paths, to incorporate small, anchor RNA structures with connecting RNA structures.
By choosing the sites of anchor structures and the paths of the RNA connections between them, embodiments design libraries of RNA aptamers de novo that are likely to have specific structural features. To do this, some embodiments first implement a method for determining specific patches of the protein surface that are most optimal for interacting with RNA, then certain embodiments design RNA structures at the protein surface. Several methods have been developed for predicting the RNA binding sites of RNA binding proteins using both structure and sequence-based approaches. (See e.g., Chen, Y. C., et al., Identifying RNA-binding residues based on evolutionary conserved structural and energetic features. Nucleic Acids Research, 2014. 42(3); Zhao, H. Y., et al., Structure-based prediction of RNA-binding domains and RNA-binding sites and application to structural genomics targets. Nucleic Acids Research, 2011. 39(8): p. 3017-3025; and Perez-Cano, L. and J. Fernandez-Recio, Optimal Protein-RNA Area, OPRA: A propensity-based method to identify RNA-binding sites on proteins. Proteins-Structure Function and Bioinformatics, 2010. 78(1): p. 25-35; the disclosures of which are incorporated by reference herein in their entireties.) Many embodiments adapt a structure-based method to predict patches of an arbitrary protein surface that are most optimal for interacting with RNA. Certain embodiments adapt Optimal protein-RNA area (OPRA) to predict patches of an arbitrary protein surface that are most optimal for interacting with RNA. (See e.g., Perez-Cano, L. and J. Fernandez-Recio, Optimal Protein-RNA Area, OPRA: A propensity-based method to identify RNA-binding sites on proteins. Proteins-Structure Function and Bioinformatics, cited above.) OPRA uses the probability of each amino acid being at an RNA/protein interface, calculated from a training set of RNA/protein complex structures, to assign an energy value to each amino acid. Then, for each amino acid on the surface of the protein, these energy values are summed over all of the neighboring residues within a certain distance cutoff, to give a set of patch scores. Some embodiments calculate updated probabilities for each amino acid using novel training sets as developed in research. Certain embodiments utilize Rosetta to output optimal patch centers as a list of amino acids. (See e.g., Leaver-Fay, A., et al., ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol, 2011. 487: p. 545-74; the disclosure of which is incorporated by reference herein in its entirety.) A number of embodiments utilize these amino acids to serve to aid in designing connecting RNA structures.
At 908 of many embodiments, connecting structures are designed to connect the anchor RNA structures from 906. In many embodiments, the connecting RNA structures are designed using the structural modularity of RNA motifs to build new RNA structures by combining motifs found in the Protein Database (PDB). Certain methods used in embodiments treat proteins as steric constraints by representing residues of an input structure as beads. However, further embodiments design the optimal connection structures by considering simple interactions with the protein. For example, some embodiments implement a representation for proteins that conserves information about residues and/or include a custom scorer object that rewards favorable interactions between the RNA and the protein for the design of RNA structures around proteins. In various embodiments, favorable interactions are defined as RNA structures that come within approximately 5 Å of positively charged protein residues. Further embodiments use a combination of methods described within this disclosure.
A schematic of method 900 is illustrated in
Turning to
Further embodiments of RNA nanostructures are connected to at least one anchor structure 108, where the anchor structures are selected from aptamers, tetraloops and/or tetraloop receptors (e.g., TTRs, including mini-TTRs), RNA-protein anchors, ribosomes, and other RNA structures.
Certain embodiments of RNA nanostructures comprise an anchor structure located between RNA motifs 102, such as illustrated in
Additional embodiments further comprise a combination of one or more centrally located anchor structures 110 flanked by one or more among RNA motifs 102 with an anchor structure 108 located at least one end of one or more, such as illustrated in
It should also be noted that certain embodiments are circularized in structure, such that one “end” of the RNA nanostructure is connected to the distal end of the RNA nanostructure, such as illustrated in
Although the following embodiments provide details on certain embodiments of the inventions, it should be understood that these are only exemplary in nature, and are not intended to limit the scope of the invention.
Methods: To build a curated motif library of all RNA structural components, a set of non-redundant RNA crystal structures managed by the Leontis and Zirbel groups (version 1.45: rna.bgsu.edu/rna3dhub/nrlist/release/1.45) were obtained. (See Petrov, A. I., et al. (2013) Automated classification of RNA 3D motifs and the RNA 3D Motif Atlas. RNA 19, 1327-1340; the disclosure of which is incorporated herein by reference in its entirety.) This set specifically removes redundant RNA structures that are identical to previously solved structures, such as ribosomes crystallized with different antibiotics. Each RNA structure to extract every motif with Dissecting the Spatial Structure of RNA (DSSR); (see Lu, X.-J., et al. (2015) DSSR: an integrated software tool for dissecting the spatial structure of RNA. Nucleic Acids Res. 43, e142; the disclosure of which is incorporated herein by reference in its entirety;) were processed with the following command:
x3dna-dssr −i file.pdb −o file_dssr.out
Each extracted motif were checked to confirm that it was the correct type, as DSSR sometimes classifies tertiary contacts as higher-order junctions and vice-versa. For each motif collected from DSSR, we ran the X3DNA find_pair and analyze programs to determine the reference frame for the first and last base pair of each motif to allow for alignment between motifs:
The naming convention for each motif involves the motif classification, the originating PDB accession code, and a unique number to distinguish from other motifs of the same type, all separated by periods. For example, TWOWAY.1GID.2, is a two-way junction from the PDB 1GID and is the third two-way junction to be found in this structure. All motifs retain their original residue numbering, chain IDs and relative position compared to their originating structure.
In addition to the motifs derived from the PDB, the make-na web server (structure.usc.edu/make-na/server.html) were utilized to generate idealized helices of between 2 and 22 base pairs in length. (see Montange, R. K., and Batey, R. T. (2008) Riboswitches: emerging themes in RNA structure and function. Annu. Rev. Biophys. 37, 117-133; the disclosure of which is incorporated herein by reference in its entirety.) All motifs in these generated libraries are bundled with some embodiments and are grouped together by type (junctions, hairpins, etc.) in sqlite3 databases in the directory RNAMake/RNAMake/resources/motif_libraries/(github.com/RNAMake/RNAMake/tree/master/RNAMake/resources/motif_libraries_new).
To build new RNA nanostructures, certain embodiments seek a path for RNA helices and noncanonical motifs that can connect two base pairs separated by a target translation and rotation. A depth-first search algorithm to discover such RNA paths were developed. The algorithm is guided by a heuristic cost function f inspired by prior manual design efforts. (See Grabow, W. W., and Jaeger, L. (2014) RNA self-assembly and RNA nanotechnology. Acc. Chem. Res. 47, 1871-18802, 25; and Dibrov, S. M., et al. (2011) Self-assembling RNA square. Proc. Natl. Acad. Sci. USA 108, 6405-6408; the disclosures of which are incorporated herein by reference in their entirety.) The algorithm is composed of two terms:
f(path)=h(path)+g(path) (eq. 1)
The first term, h(path), describes how close the last base pair in the path is to the target base pair; h(path)=0 corresponds to a perfect overlap in translation and rotation. The functional form for h(path) depends on the spatial position of each base pair's centroid d and an orthonormal coordinate frame R defining the rotational orientation of each base pair:
h(path)=|{right arrow over (d1)}−{right arrow over (d2)}|+W(|{right arrow over (d1)}−{right arrow over (d2)}|)Σi3Σj3abs(R1ij−R2ij) (eq. 2)
(See Filonov, G. S., et al. (2014) Broccoli: rapid selection of an RNA mimic of green fluorescent protein by fluorescence-based selection and directed evolution. J. Am. Chem. Soc. 136, 16299-16308; the disclosure of which is incorporated herein by reference in its entirety.)
Where d is measured in Angstroms. The weight W(d) reduces the importance of the current base pair and the target base pair with similar alignment when they are spatially far apart. This term conveys the intuition that aligning the two coordinate frames becomes important only as the path of the motif and helices approaches the target base pair. Embodiments readily allow for the exploration of alternative forms of the cost function terms in (eq. 2) and (eq. 3), including more standard rotationally invariant metrics to define rotation matrix differences; (see Huynh, D. Q. (2009) Metrics for 3D rotations: comparison and analysis. J. Math. Imaging Vis. 35, 155-164; the disclosure of which is incorporated herein by reference in its entirety;) or base-pair-to-base-pair RMSDs based on quaternions; (see Karney, C. F. F. (2007) Quaternions in molecular modeling. J Mol Graph Model 25, 595-604; the disclosure of which is incorporated herein by reference in its entirety;) but these were not tested in the current study.
The second term in the cost function (eq. 1) is g(path), which parameterizes the properties of the non-canonical RNA motifs and helices comprising the path at each stage of the calculation:
where Sss is a secondary structure score for all the motifs and helices in the path. This Sss term favors longer canonical helices as well as motifs with frequently recurring base pairs, as follows. All base pairs found in the RNA motif are scored based on their relative occurrences in all high-resolution crystal structures; all unpaired residues receive a penalty, and Watson-Crick base pairs receive an additional bonus score (Table 3).
Values were derived based on logarithms of the frequencies of these elements in the crystallographic database, i.e. the inverse Boltzmann approximation; (see Finkelstein, A. V., et al. (1995) Why do protein architectures have Boltzmann-like statistics? Proteins 23, 142-150; the disclosure of which is incorporated herein by reference in its entirety;) so that that frequency of the elements in some embodiment designs was similar to what is seen in natural RNA tertiary structures. In addition to the secondary structure score, Nmotifs penalizes the total number of motifs in the path, here taken as the number of non-canonical motifs plus the number of canonical motifs (e.g., helices, independent of helix length).
The search adds motifs and helices to the path in a depth-first manner, while the total cost function f(path) decreases, back-tracking if f(path) increases. Any solutions with h(path) less than 5, i.e., overlap at approximately nucleotide resolution between the path's last base pair and the target base pair, are accepted into a list of final designs. The balance between g(path) and h(path) allows some embodiments to reduce the number of motif combinations considered, finding most solutions in a few seconds. For each solution, EteRNAbot, was used a secondary structure optimization algorithm that has undergone extensive empirical tests to fill in helix sequences. (See Lee, J., et al. (2014) RNA design rules from a massive open laboratory. Proc. Natl. Acad. Sci. USA 111, 2122-2127; the disclosure of which is incorporated herein by reference in its entirety.)
Proteins that are included in the coordinates supplied to Embodiments are represented as steric beads centered at the Cα atom of each amino acid. This representation allows embodiments to avoid steric clashes with proteins, particularly for the ribosome tethering problems.
Results: The above method generated a multitude RNA nanostructure designs, as seen in
Conclusion: Embodiments reveal a novel approach to solving RNA pathfinding problems.
Background: The problem of creating a well-folded RNA nanostructure was first solved two decades ago by repurposing the well-characterized tetraloop/receptor (TTR) tertiary contact to bring together two separate RNA chains, analogous to the P4-P6 domain of the Tetrahymena group I self-splicing intron and other natural functional RNAs. While later RNA nanotechnology studies used the TTR module and other structural motifs to design different nanostructures, the resulting RNAs original and later designs have all been multi-chain assemblies. (See Bindewald, E., et al. (2008) Computational strategies for the automated design of RNA nanoscale structures from building blocks using NanoTiler. J Mol Graph Model 27, 299-308; Dibrov, S. M., et al. (2011) Self-assembling RNA square. Proc. Natl. Acad. Sci. USA 108, 6405-6408; Afonin, K. A., et al. (2014) Multifunctional RNA nanoparticles. Nano Lett. 14, 5662-5671; Khisamutdinov, E. F., et al. (2016) Fabrication of RNA 3D nanoprisms for loading and protection of small RNAs and model drugs. Adv. Mater. Weinheim 28, 10079-10087; and Huang, L., and Lilley, D. M. J. (2016) A quasi-cyclic RNA nano-scale molecular object constructed using kink turns. Nanoscale 8, 15189-15195; the disclosures of which are incorporated herein by reference in their entirety.) Testing embodiments on the TTR problem was chosen due to the prospect of achieving the first de novo single-chain solutions to this fundamental problem, which we hypothesized might also help crystallization.
Methods: To generate TTR linking designs, the coordinates from the X-ray crystal structure of a TTR from the P4-P6 domain of the Tetrahymena ribozyme (residues 146-157, 221-246, and 228-252 from PDB 1GID) were extracted. Second, embodiments were used to build structural segments composed of two-way junctions and helices spanning the last base pair of the hairpin (A146-U157) to base pair U221-A252 of the tetraloop-receptor, thus connecting the TTR into a single continuous strand (
To probe the structures of the TTR linking designs generated by embodiments, quantitative chemical mapping with selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE) and dimethyl sulfate (DMS) were performed. For all 16 designs illustrated in
To evaluate the formation of tertiary structure, the change in DMS reactivity of both tetraloop and tetraloop-receptor adenines as a function of Mg2+ concentration were investigated. Previous studies have demonstrated that TTR formation in the P4-P6 domain is strongly stabilized by Mg2+. As a control for the unfolded state, we measured the DMS reactivities of the tetraloop and tetraloop-receptor adenines of the TTR of the P4-P6 domain without Mg2+ (A248, A151, A152, and A153) were measured.
As an independent test of TTR linking construct folding, each RNA's GAAA tetraloop was replaced with a UUCG tetraloop, which does not form the sequence-specific TTR tertiary contact and is predicted to reduce the RNA's mobility in non-denaturing polyacrylamide gel electrophoresis, as observed for the P4-P6 domain.
After the gel-based and chemical mapping tests above, whether the embodiment designs might allow crystallization and thereby enable high-resolution characterization of the structural accuracy of the designs were tested. Crystals of miniTTR 6 that diffracted at 2.55 Å resolution (I/σ of 1.0) were obtained. Purified miniTTR 6 RNA diluted in buffer A (30 mM HEPES (pH 7.5), 20 mM MgCl2, and 100 mM KCl) was incubated at 65° C. for 2 min, centrifuged at 13,000 rpm for 2 min, and snap-cooled on ice for approximately 5 min before moving to 25° C. to set up crystallization trays. Within 2-4 weeks, miniTTR 6 crystallized at 25° C. as plates or clusters of plates via sitting-drop vapor diffusion by mixing 2 μL of miniTTR 6 at a concentration of 100 μM with 3 μL of crystallization solution containing 40 mM sodium cacodylate (pH 5.5), 20 mM MgCl2, 2 mM cobalt hexammine, and 40% 2-methyl-2,4-pentanediol (MPD). Crystals of miniTTR 6 grew to maximum dimensions of 700×700×20 μm and were stabilized and cryogenically protected by increasing the MPD to a final concentration of 44%. Crystals were flash-frozen by plunging into liquid nitrogen. Diffraction data were collected at 100 K using synchrotron X-ray radiation at beamline 4.2.2 of the Advanced Light Source, Lawrence Berkeley National Laboratory (Berkeley, Calif.). The data were processed and scaled using X-ray Detector Software (XDS). The scaled data were handled using Collaborative Computational Project programs.
The initial structural determination of the miniTTR 6 in the C2 space group was carried out from molecular replacement (MR) in Phaser (CCP4) searching for one copy of a 31-nucleotide model of only the tetraloop and receptor with the identical sequence. The rotational and translational Z-scores were somewhat low, 4.6 and 5.9 respectively, but the maps were of sufficient quality to enable the iterative building of all the residues into the 2Fo-Fc and Fo-Fc maps. Composite omit maps in PHENIX were used to help confirm the model and reduce model bias from the initial MR solution. The models were built using COOT and refined using REFMAC5 and PHENIX. The final model was refined in REFMAC5 and ERRASER, and the overall Rwork and Rfree were refined to 22.9% and 27.4%, respectively. The structure derived from the miniTTR was refined to 2.55 Å against a data set scaled to an overall I/σ of 1.0 at the highest resolution shell with 98.5% completeness.
Results: Of the 1386 nucleotides in the sixteen TTR linking constructs, 1367 (98.7%) were either reactive at target unpaired regions or protected at target helical residues, supporting the predicted secondary structures. All 19 outliers occurred at helix edges (i.e., flanking base pairs of motifs). These data supported the formation of the expected secondary structures for all TTR linking designs (See Table 1).
Several TTR linking constructs required less than 1 mM Mg2+ to fold stably, similarly to or better than reported midpoints for natural TTR-contains RNA nanostructures. Indeed, miniTTR 2 and miniTTR 16 exhibited folding stabilities better than the P4-P6 RNA in side-by-side assays. Furthermore, miniTTR 6 has a much sharper Mg2+ dependence than P4-P6 with an apparent Hill coefficient of over 10. The adenines exhibited reactivities of 1.27, 0.72, 0.70, and 0.90, respectively. The values are normalized to the reactivity of the reference hairpin loops that flank each design. Upon the addition of 10 mM Mg2+, the adenines involved in the TTR became protected from DMS modification in the P4-P6 control. As with this folding control, for 12 of the 16 designs (miniTTRs 1, 2, 5-7, 9-12 and 14-16), we observed a more than two-fold decrease in the reactivity of the TTR adenine residues. These results were consistent with Mg2+-dependent TTR formation. The remaining designs (miniTTRs 3, 4, 8 and 13) did not demonstrate significant changes in DMS reactivity upon addition of 10 mM Mg2+, indicating that the TTR interaction did not form.
Of the 16 TTR linking constructs tested, 12 designs displayed mobility shifts consistent with the formation of the TTR tertiary contact (See Table 1). Constructs 4 and 15 exhibited mobility shifts that were inconsistent with our chemical mapping results. The UUCG mutant of miniTTR design 4 displayed a mobility shift, but it did not demonstrate a full two-fold decrease in TTR DMS reactivity, suggesting partial folding. Compared to its UUCG mutant, miniTTR design 15 in the wild-type form (GAAA tetraloop) exhibited a wide, slow-mobility band. In all other cases, the electrophoretic mobility measurements were concordant with our quantitative SHAPE and DMS chemical mapping data, supporting the formation of the TTR and a compact tertiary fold.
The crystal structure and the embodiment model agreed with an all-heavy-atom RMSD of 4.2 Å, better than the nanometer-scale accuracy typically sought in RNA nanotechnology. The primary discrepancy between the modeled 3D structure and the crystal structure was a single motif, a triple mismatch drawn from the large ribosomal subunit. This motif formed multiple consecutive non-canonical base pairs with high B-factors in our miniTTR 6 crystal instead of the conformation found in the ribosomal structure, which involved flipped out adenosines (residues: O2360-O2363, O2424-O2426, PDB:1S72), as shown in
Conclusion: The stability of the TTR liking designs was particularly notable given that P4-P6 and other natural TTR-containing RNAs are larger than the miniTTR designs and have additional stabilizing tertiary contacts and other attempts to make artificial minimized TTR constructs have given significantly worse stabilities.
Background: The ribosome is a ribonucleoprotein machine dominated by two extensive RNA subunits, the 16S and 23S rRNAs. Previous work constructed a tethered ribosome called Ribo-T, in which the large and small subunit rRNAs were connected by an RNA tether to form a single subunit ribosome. In that work, the major bottleneck involved a year of numerous trial-and-error iterations to identify RNA tethers that were not cleaved by ribonucleases in vivo when wild type ribosomes were replaced in the Squires strain (SQ171fg) of E. coli. SQ171fg cells lack genetic rRNA alleles, surviving off plasmids that can be exchanged using positive and negative selections. Early failure rounds involving ribosomes from prior studies are shown in
Methods: For ribosome tether designs, PDB coordinates 3R8T and 4GD2 were used for the 50S and 30S ribosomal subunit structures respectively. From the 50S coordinates, we removed residues A2854-A2863 and, from the 30S, we removed residues A1445-A1457. These designs contained either four or five noncanonical structural motifs each to tether the H101 helix on a circularly permuted 23S rRNA to the h44 helix on the 16S rRNA (
The designed tethers were cloned into plasmid pRibo-T-A2058G. The backbone was generated for each design using forward (f) and reverse (r) primer pairs in separate PCR reactions using plasmid pRibo-T as a template, Phusion polymerase (NEB), and 3% DMSO. PCR cycling was as follows: 98° C. for 3 min; 25 cycles of 98° C. for 30 sec, 55° C. for 30 sec, 72° C. for 2 min; and 72° C. for 10 min. Circularly permuted 23S ribosomal RNA (rRNA) was generated with forward and reverse primer pairs, the pRibo-T template, and the same PCR conditions as described above. Each PCR reaction was purified by gel extraction from a 0.7% agarose gel with an E.Z.N.A. gel extraction kit (Omega). Each purified backbone (50 ng) was assembled with the respective 23S insert in 3-fold molar excess using Gibson assembly. Assembly reactions were transformed into POP2136 cells, and the cells were grown at 30° C. overnight. Colonies were picked and plasmids were isolated using an E.Z.N.A. miniprep kit (Omega) and confirmed with full plasmid sequencing by ACGT, Inc.
Each purified plasmid (100 ng) was separately transformed into electrocompetent SQ171fg cells containing pCSacB. Cells were recovered in 1 mL of SOC media at 37° C. with shaking for 1 hour. Fresh SOC (1.85 mL) supplemented with 50 μg/mL carbenicillin and 0.25% sucrose was inoculated with 250 μL of recovered cells and incubated overnight at 37° C. with shaking. Cultures (10% and 90%) were plated on LB agar plates supplemented with 50 μg/mL carbenicillin, 5% sucrose and 1 mg/mL erythromycin and incubated at 37° C.
After 48 hours with no visible colonies, the plates were replica plated onto fresh LB agar plates supplemented with 50 μg/mL carbenicillin, 5% sucrose and 1 mg/mL erythromycin and incubated at 37° C. After 72 additional hours, colonies appeared on the plate containing RM-Tether design 4. Eight colonies were streaked onto LB agar supplemented with 50 μg/mL carbenicillin and 1 mg/mL erythromycin and LB agar supplemented with 30 μg/mL kanamycin (to confirm loss of the pCSacB plasmid) and were also used to inoculate 5 mL of LB supplemented with 50 μg/mL carbenicillin and 1 mg/mL erythromycin. Plates were incubated at 37° C., and cultures were incubated at 37° C. with shaking. The OD600 of the cultures was tracked to generate growth curves (Biochrom Libra S4 spectrophotometer). After 5 days at 37° C., total RNA was extracted using an RNA extraction kit from Qiagen. Total RNA was analyzed by gel electrophoresis on a 1% agarose gel with GelRed. Total plasmid was extracted from saturated 5 mL cultures with an E.Z.N.A. miniprep kit (Omega) and sequenced to confirm the correct RM-Tether design 4 sequence.
For in vitro characterization of ribosomes, all constructs (wild type, Ribo-T v1.0, and RM-Tether 4) were cloned to be under control of a T7 promoter. The T7 promoter was introduced into primers, and amplified using the wild type, Ribo-T v1.0, and RM-Tether 4 plasmids as templates for PCR amplification. PCR products were blunt end ligated, transformed into DH5α E. coli cells using electroporation, and plated onto LB-agar/ampicillin plates at 37° C. Plasmid was recovered from resulting clones and sequence confirmed.
In vitro ribosome synthesis, assembly, and translation (iSAT) reactions were set-up as previously described. Briefly, eight 15 μL reactions were prepared and incubated for 2 hours at 37° C., then pooled together.
Sucrose gradients were prepared from buffer C (10 mM Tris-OAc (pH=7.5 at 4° C.), 60 mM NH4Cl, 7.5 mM Mg(OAc)2, 0.5 mM EDTA, 2 mM DTT) with 10 and 40% sucrose in SW41 polycarbonate tubes using a Biocomp Gradient Master. Gradients were placed in SW41 buckets and chilled to 4° C. 120 μL of pooled iSAT reactions were loaded onto the gradients. The gradients were ultra-centrifuged at 22,500 rpm for 17 hours at 4° C., using an Optima L-80 XP ultracentrifuge (Beckman-Coulter) at medium acceleration and braking (setting of 5 for each). Gradients were analyzed with a BR-188 density gradient fractionation system (Brandel) by pushing 60% sucrose into the gradient at 0.75 mL/min (at normal speed). Traces of A254 readings versus elution volumes were obtained for each gradient. Gradient fractions were collected and analyzed for rRNA content by gel electrophoresis in 1% agarose and imaged in a GelDoc Imager (Bio-Rad). Ribosome profile peaks were identified based on the rRNA content as representing 30S or 50S subunits, 70S ribosomes, or polysomes.
Fractions containing 70S ribosomes and polysomes were collected and pooled. These fractions were recovered as previously described, with pelleted iSAT ribosomes resuspended in iSAT buffer, aliquoted, and flash-frozen. These pelleted fractions were re-run on a 1 agarose gel and imaged in a GelDoc Imager to confirm tethering in monosome and polysome peaks.
For SHAPE-seq, in vitro ribosome synthesis, assembly, and translation reactions were set-up as previously described. (See Jewett, M. C., et al. (2013) In vitro integration of ribosomal RNA synthesis, ribosome assembly, and translation. Mol. Syst. Biol. 9, 678; and Fritz, B. R., et al. (2015) Implications of macromolecular crowding and reducing conditions for in vitro ribosome construction. Nucleic Acids Res. 43, 4774-4784; the disclosures of which are incorporated herein by reference in their entirety.) Briefly, 15 μL iSAT reactions each possessing wild type, Ribo-T, or RM-40 were prepared in triplicate, incubated for 2 hours at 37° C., and then placed on ice. To perform SHAPE modification, samples were warmed to 37° C. for 5 minutes, and 7.5 μL of each sample was added to 0.83 μL of 65 mM 1-methyl-7-nitroisatoic anhydride (1M7) or 0.83 μL DMSO (control solvent). Reactions were incubated for 2 minutes, then all samples were Trizol extracted, ethanol precipitated, washed twice with 70% ethanol, and resuspended in 10 μL water. Subsequent library preparation steps were performed as described previously with one exception: 2 custom reverse transcription primers were used to simultaneously probe the regions containing T1 (5′-GGTTAAGCCTCACGG-3′) and T2 (5′-CCCTACGGTTACCTTGTTACGAC-3′). (See Watters, K. E., et al. (2016) Simultaneous characterization of cellular RNA structure and function with in-cell SHAPE-Seq. Nucleic Acids Res. 44, e12; the disclosure of which is incorporated herein by reference in its entirety.) Following 2×75 bp paired-end Illumina sequencing, SHAPE reactivities were calculated as described by Yu et al. mapping both modification-induced stops and mutations. (See Yu et al. (2018) Estimating RNA structure chemical probing reactivities from reverse transcriptase stops and mutations, BioRxiv; the disclosure of which is incorporated herein by reference in its entirety.) Raw reactivities were calculated using Spats v1.9.8, and were then linearly re-scaled to account for estimated differences in SHAPE probe concentration between replicates. Specifically, one replicate was first selected as the reference. Reactivities for the other datasets were divided by the reference at each position, then the median value of this ratio was taken as the scale factor. Reactivities across each dataset were divided by their scale factor. The same experimental replicate was used to scale reactivities, and reactivities are presented as the average value over these re-scaled replicates.
Results: One of these seven constructs, RM-Tether 4 (
Conclusion: Taken together, these data demonstrate that an embodiment-designed ribosomes with structured, chemically stable tethers can replace wild type ribosomes in vivo and more than one such ribosome can be loaded onto a single message in vitro. Embodiments obviate repeated rounds of trial and error that were previously required to achieve these design goals.
Background: Small molecules can be bound and sensed by artificially selected RNA aptamers. Unfortunately, these molecules often exhibit weakened binding affinities or instability in biological environments, and additional rounds of selection to improve aptamers typically give diminishing returns. (See Carothers, J. M., et al. (2006) Aptamers selected for higher-affinity binding are not more specific for the target ligand. J. Am. Chem. Soc. 128, 7929-7937; Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646; and Ellington, A. D., and Szostak, J. W. (1990) In vitro selection of RNA molecules that bind specific ligands. Nature 346, 818-822; the disclosures of which are incorporated herein by reference in their entirety.)
Methods: Starting with PDB 1AM0 we removed residues A6-A18 and A33-A35 to achieve a minimal ATP aptamer flanked by single Watson-Crick base pairs. We moved these residues into a new PDB ‘ATP_min.pdb’.
Results: In all 5210 designs were generated. As with previous construct designs, designs were selected that maximized motif usage and minimized the chain closure score or how close the optimized sequence is to the target base pair. In total, 10 ATP aptamers embedded by an embodiment into scaffolds with tetraloop/receptor contacts, which we called ATP-TTR designs (
Conclusion: These results demonstrate that the TTR peripheral contact efficiently couples to enhance binding of ATP in the aptameric region, as desired. As a further test of this coupling, we confirmed that the Mg2+ requirements for forming the TTR was reduced in the presence compared to the absence of the small molecule ligand in these constructs (
Background: Binding to Spinach enhances the fluorescence of DFHBI by ˜1,000-fold relative to unbound ligand, making this RNA useful for biological interrogations (38, 45), although its binding affinity, brightness, folding efficiency and biological stability remain poor even after extensive efforts to discover improvements such as the minimized Spinach and Broccoli aptamers (46-49). (See Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646; Kellenberger, C. A., et al. (2015) RNA-Based Fluorescent Biosensors for Live Cell Imaging of Second Messenger Cyclic di-AMP. J. Am. Chem. Soc. 137, 6432-6435; Strack, R. L., et al. (2013) A superfolding Spinach2 reveals the dynamic nature of trinucleotide repeat-containing RNA. Nat. Methods 10, 1219-1224; Filonov, G. S., et al. (2014) Broccoli: rapid selection of an RNA mimic of green fluorescent protein by fluorescence-based selection and directed evolution. J. Am. Chem. Soc. 136, 16299-16308; Ketterer, S., et al. (2015) Systematic reconstruction of binding and stability landscapes of the fluorogenic aptamer spinach. Nucleic Acids Res. 43, 9564-9572; and Song, W., et al. (2014) Plug-and-play fluorophores extend the spectral properties of Spinach. J. Am. Chem. Soc. 136, 1198-1201; the disclosures of which are incorporated herein by reference in their entirety.)
Methods: Starting with PDB 6614 we removed residues R19-R31 and R49-R66 to achieve the minimal DFHBI binding aptamer (Spinach_min.pdb).
A stock of DFHBI (Sigma) was prepared in PBSMKT (1×phosphate buffered saline, 5 mM MgCl2, 100 mM KCl, 0.01% Tween-20, pH 7.2) and its absorbance measured using a UV spectrophotometer (NanoDrop, Thermo Scientific). The DFHBI concentration was calculated using an extinction coefficient of 30,100 cm-1/M at 423 nm as previously reported. (See Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646; the disclosure of which is incorporated herein by reference in its entirety.) A DFHBI titration was performed in half area, flat-bottomed black 96-well plates (Corning) at a final RNA concentration of 200 nM with DFHBI concentration ranging from 10 μM to 10 nM prepared in a 1:2 dilution series. After mixing, the plates were covered with an adhesive film to prevent evaporation and temperature-cycled from room temperature to 4° C. twice over the course of 1 hour to allow aptamer-target equilibration while minimizing magnesium-dependent self-cleavage. Measurements were acquired at room temperature and wells were excited at 462±10 nm and emission was measured at 504±15 nm using a Tecan M1000 plate reader. A fluorescence background was obtained at each DFHBI concentration in the absence of RNA and subtracted from the corresponding wells. The corrected signal for each aptamer at every DFHBI concentration was then least-squares fit using a custom MATLAB script using a 1:1 complexation model according to the following equation:
Here, [T] is the concentration of DFHBI, Kd is the dissociation constant of the given aptamer, and Bmax is the maximum brightness obtained for the given concentration of aptamer.
Next, we prepared an RNA titration assay using identical measurement, equilibration, and buffer conditions, except with the amount of DFHBI constant at 400 nM and RNA concentrations ranging from 5 μM down to 5 nM prepared in a 1:2 dilution series. A background fluorescence was obtained at 400 nM DFHBI in the absence of RNA and subtracted from each well. The corrected signal was then least-squares fit using a custom MATLAB script using a 1:1 complexation model according to the following equation:
Where [A] was the concentration of aptamer, f is the folding efficiency, DT is the DFHBI concentration (400 nM), Kd is the dissociation constant calculated for each sequence above, and Fmax is the maximum fluorescence signal at dye-binding saturation. Quantum yields were obtained through direct comparison of Fmax with the literature value for Broccoli (QY=0.72).
Small molecules can be bound and sensed by artificially selected RNA aptamers. Unfortunately, these molecules often exhibit weakened binding affinities or instability in biological environments, and additional rounds of selection to improve aptamers typically give diminishing returns. (See Carothers, J. M., et al. (2006) Aptamers selected for higher-affinity binding are not more specific for the target ligand. J. Am. Chem. Soc. 128, 7929-7937; Paige, J. S., et al. (2011) RNA mimics of green fluorescent protein. Science 333, 642-646; and Ellington, A. D., and Szostak, J. W. (1990) In vitro selection of RNA molecules that bind specific ligands. Nature 346, 818-822; the disclosures of which are incorporated herein by reference in their entirety.)
Each TTR Spinach aptamer was prepared in 60 μL PBSMKT containing 1.66 μM total RNA and 30 μL of this was added to 50 μL of 5 μM DFHBI in PBSMKT in two wells per aptamer. Next, 20 μL of PBSMKT was added to one well per aptamer to give a final concentration of 500 nM RNA and 2.5 μM DFHBI in order to provide a baseline fluorescence. Next, 20 μL of 100% frog egg lysate prepared 4 hours earlier and stored at 4° C., was added to each well and pipet mixed. (Higher lysate concentrations were too optically absorbent to allow fluorescence measurements). Fluorescence measurements were then obtained for every well every 1 minute for 30 minutes, then every 3 minutes for 1 hour, and after every 5 minutes for an additional hour. For evaluation of times to half-fluorescence, the fluorescence of each aptamer in wells containing lysate was normalized to the same aptamer's fluorescence in PBSMKT at every time point in order to account for photobleaching.
Each TTR Spinach aptamer was prepared in PBSMK (1×PBS pH 7.2, 5 mM MgCl2, 100 mM KCl) containing 1 μM RNA and 2.5 μM DFHBI. The RNA/DFHBI mixture was equilibrated on ice for 30 minutes before aliquoting 50 μL into 4 wells per RNA species. As control reactions, 50 μL of PBSMK containing 2.5 μM DFHBI was added to one of these wells per RNA. Immediately prior to use, PBSMLK (1×PBS pH 7.2, 5 mM MgCl2, 40% E. coli lysate, 100 mM KCl) containing 2.5 μM DFHBI was prepared and 50 μL of this mixture was added to each well to give final concentrations of 500 nM RNA, 2.5 μM DFHBI, and 20% E. coli lysate. Immediately upon addition of PBSMLK, fluorescence intensities were obtained for every well and repeated every 30 s for 8 hours using a Tecan M1000 plate reader.
To test the in vivo fluorescence of Spinach-TTR variants, designed sequences were cloned between a T7 promoter and T7 terminator in a plasmid harboring carbenicillin resistance and a ColE1 origin of replication. Plasmids were transformed into chemically competent E. coli strain BL21*(DE3) (F− ompT hsdSB (rB− mB−) gal dcm me131 [DE3]), plated on Difco LB+Agar plates containing 100 μg/mL carbenicillin, and grown overnight at 37° C. A cellular autofluorescence control containing a blank plasmid was also included. Individual colonies were grown overnight in LB containing 100 μg/mL carbenicillin, then diluted 1:50 into fresh LB. After 1 h, Isopropyl-β-D-thiogalactoside (IPTG) was added at a final concentration of 100 μM to induce expression of T7 RNA polymerase. After 4.5 h of additional shaking, cells were diluted 1:200 into lx Phosphate Buffered Saline (PBS) containing 2 mg/mL kanamycin and 200 μM (Z)-4-(3,5-Difluoro-4-hydroxybenzylidene)-1,2-dimethyl-1H-imidazol-5(4H)-one (DFHBI), then incubated at 37° C. for 5 minutes. A BD Accuri C6 Plus flow cytometer fitted with a high-throughput sampler was then used to measure fluorescence of at least 50,000 events for each sample. Measurements were taken for 4 biological replicates.
Flow cytometry data analysis was performed using FlowJo (v10.4.1). Cells were gated by FSC-A and SSC-A, and the same gate was used for all samples. The geometric mean fluorescence was calculated for each sample, then all fluorescence measurements were converted to Molecules of Equivalent Fluorescein (MEFL) using CS&T RUO Beads (BD). The average fluorescence (MEFL) of cells expressing blank plasmid (pJBL002) in the presence of DFHBI was then subtracted from each measured fluorescence value.
Results: In all 697 designs were generated, and a subset were again chosen to maximize number of motifs tested and the chain closure score (how close the designed RNA sequence is to overlay with its target base pair). Out of these designs, 16 ‘Spinach-TTR’ molecules designed by an embodiment to embed the Spinach aptamer into scaffolds with tetraloop/receptor contacts were characterized (
Additionally, six of the seven Spinach-TTR constructs exhibited fluorescence longer than control Spinach and Broccoli sequences. Spinach-TTR 3 exhibited particularly high stability (
Conclusion: These results demonstrate that the TTR peripheral contact efficiently couples to enhance binding of DFHBI in the aptameric region, thus increasing fluorescence. As a further test, these aptameric designs also showed to be more effective than other aptamers at increasing fluorescence as well as more stable, when challenged with cellular lysate, showing that embodiments herein are a vast improvement in the art at stabilizing and improving aptamer function.
Background: Two well-studied RNA binding proteins, MS2 coat protein and PUF3 can be used as model systems for testing the design of RNA connections. MS2 coat protein specifically binds a 19 nucleotide RNA hairpin structure with nanomolar affinity. (See Carey, J., et al, Interaction of R17 coat protein with synthetic variants of its ribonucleic acid binding site. Biochemistry, 1983. 22(20): p. 4723-30; the disclosure of which is incorporated by reference herein in its entirety.) PUF3 binds an 8-nucleotide single stranded RNA sequence with nanomolar affinity. (See Zhu, D. Y., et al., A 5′ cytosine binding pocket in Puf3p specifies regulation of mitochondrial mRNAs. Proceedings of the National Academy of Sciences of the United States of America, 2009. 106(48): p. 20192-20197; the disclosure of which is incorporated by reference herein in its entirety.) Both systems have been extensively characterized and crystal structures of the complexes have been solved. (See e.g., Helgstrand, C., et al., Investigating the structural basis of purine specificity in the structures of MS2 coat protein RNA translational operator hairpins. Nucleic Acids Res, 2002. 30(12): p. 2678-85; the disclosure of which is incorporated by reference herein in its entirety.) Here, designing and testing a library of RNA structures addresses two main questions. First, if removing key binding residues from the RNA targets, e.g. remove the tetraloop from the MS2 hairpin structure, how can the remaining RNA target structure, e.g. the MS2 helix, be built on to create new RNA structures that recover the wildtype binding affinity. Second, can the wildtype RNA structures, e.g., the full MS2 hairpin structure, to create new RNA structures that bind to their target proteins with higher affinity.
Methods: To address these questions, an embodiment designs a library of sequences which systematically varies the RNA anchor structures. Two examples are shown in
Background: Predicting binding affinity increases the predictive capacity for embodiments to design successful RNAs for binding proteins. In particular, some embodiments identify predictive features of successful designs with the goal of increasing the percentage of successful designs in the future. Binding affinity is defined as the free energy difference between the complex and the unbound components.
Methods: An embodiment approximately estimate the free energy of the bound complex as a linear combination of various features such as the number of protein/RNA contacts, the extent to which the RNA wraps around the protein, the predicted free energy of the bound RNA secondary structure, and the number and strength of anchor structures. The unbound free energy of the protein are neglected for simplicity and the unbound free energy of the RNA are estimated as the free energy of all possible secondary structures, i.e. from Vienna. Weights are fit for each of these terms using a simple linear regression to a training subset. The correlation coefficient and the AUC of the resulting model are used to assess its utility.
In silico binding affinity prediction is a very difficult problem: previous work showed that even predicting the relative protein binding affinities of small, closely related RNA sequences is challenging and at best yields results accurate to within 1.5 kcal/mol. Because predicting absolute binding affinity is even more challenging, it is possible that the model described above are not predictive. If that is the case, an embodiment focuses on identifying features that increase the likelihood of a successful design, e.g. designs that detectably bind to the target protein. Again, these features are identified from a training subset of the binding affinity data. As an example, an embodiment may identify that designs that have more protein/RNA contacts are more likely to be successful.
Once the binding affinity model or the predictive features have been established, an embodiment implement a new scoring function to encourage solutions that are predicted to be more successful. The embodiment then designs and test a new library of RNA structures for MS2 and PUF3, in the same manner as described in Example 1.
Background: A need exists to assess designs to both measure binding affinity and to examine the structure of the complex. An embodiment verifies this assumption for a small subset of designs deemed successful in other embodiments.
Methods: The RNA/protein structure are examined by performing one dimensional SHAPE chemical mapping on the bound complexes. A SHAPE profile consistent with the secondary structure of the design is expected, with reduced reactivity in regions predicted to be bound to the protein. Additionally, for a small subset of design failures SHAPE chemical mapping in the presence and absence of the protein is performed. By identifying ways in which the designs are failing, design algorithms may be improved.
Background: Once designed and constructed, aptamer embodiments can be tested for the efficacy in binding particular proteins to which they were designed to bind.
Methods: The aptamers are designed by first identifying several possible RNA anchor structures/sequences methods, such as those described herein. Then for each of these sets of anchor structures, many different connecting RNA structures are designed. Additionally, each of the libraries contains a subset of sequences with specific randomized portions, for a total of approximately 1015 sequences in each library. The benchmark set of proteins contains proteins that range in size and for which previous selection attempts have been both successful and unsuccessful. Table 1 lists an initial set of five possible proteins for the benchmark set. Selections are performed for each of these proteins with the designed libraries. This initial benchmark set helps to identify the optimal way in which to incorporate randomized regions into the designed sequences. The success is assessed by the binding affinities of the selected aptamers.
Background: If or when successful aptamers are identified, the structures of these aptamers can be examined to identify the specific features that contribute to the success.
Methods: First the structures of the RNA are verified by performing one-dimensional SHAPE chemical mapping. By examining the SHAPE profile in the presence and absence of the protein, the regions of the RNA that are likely to be interacting with the protein are identified. In addition to the chemical mapping experiments, verifying that the RNA is binding to the protein where it was predicted on the surface are performed. To do this, successful designs that were predicted to leave functional sites accessible are assessed. For these aptamer embodiments, the binding affinity of ligands known to bind to the functional site after incubating the protein with the RNA aptamer are assessed. If the binding affinity of the ligand remains the same when the protein is bound to the RNA aptamer, this would suggest that the functional site is indeed accessible. For example, there are several ligands known to bind to the different binding pockets on thrombin. Aptamers can be designed that should specifically leave one of these binding sites accessible. Then, thrombin are incubated with one of the successful aptamers, then the binding affinity of one of the known ligands to the thrombin/aptamer complex are measured.
Background: When selection experiments fail, they generally still yield many low-quality aptamers. This often means aptamers that have high nanomolar or low micromolar affinity to the target protein. Currently, there is no simple strategy for optimizing these aptamers to bind with higher affinity.
Methods: First, the structure of the RNA aptamer bound to the target protein will be predicted. Using many (˜100) of the structures that score best, RNA extensions that should wrap around the protein will be designed. A small library of these designs will then be tested experimentally. It is expected that some of these designs will bind to the target protein with higher affinity than the original aptamer.
Background: Certain embodiments will seek to predict a structure of an RNA/protein complex based on RNA sequence and protein structure.
Methods: An embodiment will extend the fragment assembly algorithm for RNA structure prediction within Rosetta. This method builds de novo RNA structures by sampling torsion angles from fragments of RNA structures from the PDB in a Monte Carlo simulation. Protein binding will be incorporated using two different strategies: 1) fold the RNA in the presence of the protein, and 2) fold the RNA without the protein and then dock it onto the protein surface and remodel interface residues. Both of these initial strategies will use a coarse-grained representation of the protein and RNA residues.
The first strategy, folding the RNA in the presence of the protein, will involve both fragment insertion and docking moves. Initially, we will implement a strategy similar to that described previously for the simultaneous folding and docking of symmetric protein complexes, in which every tenth move will be a docking attempt. (See Das, R., et al., Simultaneous prediction of protein folding and docking at high resolution. Proceedings of the National Academy of Sciences of the United States of America, 2009. 106(45): p. 18978-18983; the disclosure of which is incorporated by reference herein in its entirety.) Each move will be scored using the potential described herein.
The novel aspect of the second strategy is essentially the flexible docking algorithm. Initially, the RNA structure will be built with the fragment assembly method. Because the protein will not be present at this stage, structures will be evaluated with the RNA-only potential. The resulting RNA structures will then be docked against the protein and interface residues will be resampled with fragment insertion moves. At this stage, structures will be scored with the RNA/protein potential described herein.
Finally, coarse-grained structures resulting from either of these two strategies will be converted into full-atom representation. The structures will be refined by sampling side chain rotamers in a Monte Carlo simulation and then performing energy minimization using the high-resolution RNA/protein potential described herein.
These methods will be tested on a benchmark set of RNA/protein complexes with known structures. Varying amounts of input information will be provided for each complex, ranging from just the protein structure and the RNA sequence, to the protein structure with one or more “anchor” RNA residues bound, to the protein structure and parts of the RNA structure. The results over this range of input information will help to evaluate the reliability of this method in various practical situations.
Having described several embodiments, it will be recognized by those skilled in the art that various modifications, alternative constructions, and equivalents may be used without departing from the spirit of the invention. Additionally, a number of well-known processes and elements have not been described in order to avoid unnecessarily obscuring the present invention. Accordingly, the above description should not be taken as limiting the scope of the invention.
Those skilled in the art will appreciate that the foregoing examples and descriptions of various preferred embodiments of the present invention are merely illustrative of the invention as a whole, and that variations in the components or steps of the present invention may be made within the spirit and scope of the invention. Accordingly, the present invention is not limited to the specific embodiments described herein, but, rather, is defined by the scope of the appended claims.
This application claims priority to U.S. Provisional Application Ser. No. 62/894,098, entitled “Methods and Systems for Rational Design of RNA Aptamers and Uses Thereof” to Das et al., filed Aug. 30, 2019 and U.S. Provisional Application Ser. No. 62/835,699, entitled “Systems and Methods for Designing RNA Nanostructures and Uses Thereof” to Das et al., filed Apr. 18, 2019; the disclosures of which are herein incorporated by reference in their entireties.
This invention was made with Governmental support under Contract Nos. GM122579 and GM100953 awarded by the National Institutes of Health and under Contract No. DGE-114747 awarded by the National Science Foundation. The government has certain rights in the invention.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/US2020/029018 | 4/20/2020 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 62894098 | Aug 2019 | US | |
| 62835699 | Apr 2019 | US |
