Integrated design of RNA-based regulatory systems for cellular control and information processing.

Engineering Biology

Integrated design of RNA-based regulatory systems for cellular control and information processing.

Current State-of-the-Art

There is a unique opportunity to construct information processing genetic circuitry out of RNA molecules versus proteins due to three factors: 1) a deep history of molecular computation with nucleic acids in vitro¹; 2) the concept of nucleic acid strand displacement as a simple, yet highly modular and programmable mechanism to maintain and propagate changes of molecular state; and 3) the emergence of RNA secondary structure design tools that can implement new designs using these paradigms. Within the realm of in vitro nucleic acid ‘circuits’, there has been great progress in developing in vitro systems that can process information in much the same way as genetic regulatory circuits. In these systems, chemical state is defined as the concentration of specific nucleic acid species, and these states can be changed through designed interactions that can process information, such as logic evaluation², or even perform complex computational tasks³. One specific paradigm for programming these interactions is via nucleic acid strand displacement – a method by which specific nucleic acid (DNA or RNA) hybrids can exchange strands with each other to change the abundance of specific hybrid species. They offer a powerful paradigm for molecular computation in that they leverage the existing DNA/RNA structure design tools and can be abstracted into high-level programming languages. In addition, RNA has additional advantages in being able to enhance functional properties through expanded non-natural nucleic acid chemistries, and by leveraging the wide range of RNA modifications present especially in eukaryotic systems. Overall, we know a great deal more about the fundamental principles of programming reaction cascades with RNAs than we do for proteins, due to the latter being governed by specific protein interactions that are harder to generalize. We therefore highlight RNA circuit design as its own goal while covering protein-based circuit design in Holistic, integrated design of multi-part genetic systems (i.e., circuits and pathways).

While there is great promise in RNA circuit design, and some progress made in porting the most basic elements of strand displacement into designed RNA regulators of gene expression (toehold switches⁴ and small transcription activating RNAs (STARs)⁵), the full repertoire of strand displacement capability has so far not been fully ported to living cellular systems. This represents the major challenge of this goal, with the accordingly great opportunity to expand the programmable molecular control over cellular systems through porting strand displacement into cells.

Breakthrough Capabilities & Milestones

Porting nucleic acid strand displacement technology into cellular systems with RNA instantiations.

RNA implementation of strand displacement cascades in bacteria.

Bottleneck/Challenge: Difficulties in designing RNAs that are stable and robustly fold into desired structures needed for strand displacement within the cell.

Potential Solution: Identification of new RNA-protecting motifs compatible with strand displacement.

Potential Solution: Development and use of RNA-targeting CRISPR nucleases (e.g., dCas13a) for site-specific modification of RNA bases to enhance stability.

Potential Solution: Development of alternative, strand-displacement motifs that are more compatible with the cellular context.

Potential Solution: Further advances of in-cell structure probing techniques to inform design principles for RNA strand displacement regulators.

Potential Solution: Evolutionary optimization using RNA-targeting CRISPR systems to activate mRNAs or regulatory outputs to activate gene expression for selection.

Bottleneck/Challenge: Cellular RNA cascades require large concentration ratios to be effective.

Potential Solution: Design and implement RNA cascades that leverage catalytic interactions.

Potential Solution: Develop design rules to integrate the components of current multi-strand cascades into a single strand.

Bottleneck/Challenge: Cellular RNAs can have high secondary structures that make them challenging to use as inputs for strand displacement cascades.

Potential Solution: Implement RNA signal transducers that modify cellular RNA structures to be compatible.

Bottleneck/Challenge: Filling gaps in RNA parts and their performance. There remains a high level of discrepancy in RNA part performance, particularly across different cell types and in the context of intracellular conditions. For example, dynamic range (On/Off state function) is a critical parameter for gene regulators that determines the stringency of control and their utility for constructing gene networks but varies in different contexts for many existing parts.

Potential Solution: Creation of high-performing RNA parts explicitly characterized for robustness and engineered and tested at magnesium concentrations that are physiologically relevant for intracellular use.

RNA implementation of strand displacement cascades in eukaryotic systems.

Bottleneck/Challenge: Eukaryotic RNAs are subject to more complex regulation and are subject to additional processing (e.g., nuclear transport, capping, polyadenylation, splicing, microRNA pathways).

Potential Solution: Develop the equivalent of compact and simplified bacterial RNA regulators that distill the complexity of these interactions to function robustly in mammalian systems.

Bottleneck/Challenge: RNA concentrations in eukaryotic cells are lower than those in prokaryotic cells and reduce RNA-RNA interactions.

Potential Solution: Implement RNA strand displacement cascades using RNAs that are confined to the nucleus or that are tagged to remain in the nucleus.

Potential Solution: Make use of catalytic RNA systems as described above.

Engineer ‘universal’ computational strand displacement architectures using strand displacement in bacteria.

Bottleneck/Challenge: The most sophisticated strand-displacement architectures require reversible interactions for error correction.

Potential Solution: Leverage existing architectures by implementing strand displacement that allow reversible interactions, or develop a new strand-displacement architecture that is amenable to cellular requirements.

Bottleneck/Challenge: Strand-displacement neural networks require precise control over the relative stoichiometry of strands in the network.

Potential Solution: Develop methods to encode the complete strand-displacement RNA network in a single strand of RNA.

Bottleneck/Challenge: Signal conditioning methods are required in the propagating cascade for proper network function.

Potential Solution: Engineer methods of RNA signal thresholding and amplification required for signal conditioning.

Engineer computational RNA strand displacement networks in mammalian systems.

Bottleneck/Challenge: Methods implemented for encoding RNA strand-displacement networks in bacteria may no longer function in mammalian cells.

Potential Solution: Adopt RNA network encoding that is specific to mammalian cells that accommodates differences in RNA processing, polyadenylation, and potential changes in ribozyme activity.

Bottleneck/Challenge: Lower concentrations of RNAs in mammalian cells prevents proper network function.

Potential Solution: Encode the RNA strand-displacement network within a single strand of RNA.

Computational design of RNA strand displacement neural networks that process the transcriptome.

Bottleneck/Challenge: Strand displacement networks need to take as input transcriptomic RNAs that are improperly folded and processed, partially degraded, and interacting with other cellular machinery.

Potential Solution: Design of strand displacement interaction architectures that are robust to errors in inputs.

Potential Solution: Design of error correction computational layers that can process ‘messy’ input signatures.

Engineer RNA neural networks that dynamically reprogram cell state.

Porting successes in computationally designed bacterial RNA-based genetic regulators into eukaryotic and mammalian systems.

First generation eukaryotic RNA-based gene regulators that utilize RNA:RNA interactions and/or strand-displacement and achieve 10-fold change in gene expression.

Bottleneck/Challenge: Current state-of-the-art RNA-based gene regulators in bacteria utilize RNA structures to control transcription (STARs) and translation (Toeholds, and small RNAs) that are fundamentally different in eukaryotes and do not directly port.

Potential Solution: Understand how initiation, elongation and termination of both transcription and translation can be manipulated through formation of RNA structures and interactions.

Potential Solution: Uncover how natural RNA processing (splicing, polyadenylation, 5’ capping) specific to eukaryotes can be regulated through formation of RNA structures and interactions.

Potential Solution: Investigate orthogonal mechanisms of gene expression and RNA processing in eukaryotic cells, for example, that use machinery from naturally occurring viruses.

Creation of RNA modification machinery that allows programmable site-specific modifications of RNA, focusing on naturally abundant modifications (N6-methyl adenosine, 2'-O-methylation, pseudouridine).

Bottleneck/Challenge: We currently have no ability to program RNA modifications inside living cells and natural modification machines lack sequence flexibility.

Potential Solution: Create CRISPR-based systems able to perform guide RNA directed post-transcriptional modification of RNA.

Potential Solution: Investigate natural RNA modification machinery to identify programmable and/or sequence flexible mechanisms.

Second generation eukaryotic RNA-based gene regulators that are suitable for computational design to create libraries that are highly-orthogonal and high-performing, achieving 100’s-fold change in gene expression.

Use RNA modifications for programming or fine-tuning RNA functions.

Examples include using modifications to dynamically control how RNAs interact, control gene expression, and form nanostructures.

Bottleneck/Challenge: While RNA modifications are known to be abundant and diverse, we have almost no idea about their functional consequences. This fundamentally limits our ability to harness modified nucleotides as building blocks or control points.

Potential Solution: Combine in vitro synthesized and modified RNAs with cell-free protein synthesis systems or purified reactions to systematically investigate effects of modifications.

Potential Solution: Apply programmable RNA modification tools in vivo to determine functional consequences of RNA modifications.

Potential Solution: Use next-generation sequencing technologies and statistical analysis to understand how modifications perturb endogenous RNA processing, structure formation, interactions and translation.

Expand RNA modification apparatus to modify non-natural RNA alphabets to enhance their functional properties.

Engineering enzymes that can perform non-natural RNA modifications to further expand the chemical repertoire of what is possible and extend RNA ligand recognition, catalysis and genetic control.

Footnotes

Qian, L., & Winfree, E. (2011). Scaling up digital circuit computation with DNA strand displacement cascades. Science, 332(6034), 1196–1201. View publication.
Cherry, K. M., & Qian, L. (2018). Scaling up molecular pattern recognition with DNA-based winner-take-all neural networks. Nature, 559(7714), 370–376.View publication.
Seelig, G., Soloveichik, D., Zhang, D. Y., & Winfree, E. (2006). Enzyme-free nucleic acid logic circuits. Science, 314(5805), 1585–1588. View publication.
Qian, L., & Winfree, E. (2011). Scaling up digital circuit computation with DNA strand displacement cascades. Science, 332(6034), 1196–1201.View publication.
Cherry, K. M., & Qian, L. (2018). Scaling up molecular pattern recognition with DNA-based winner-take-all neural networks. Nature, 559(7714), 370–376. View publication.
Green, A. A., Silver, P. A., Collins, J. J., & Yin, P. (2014). Toehold switches: de-novo-designed regulators of gene expression. Cell, 159(4), 925–939. View publication.
Chappell, J., Westbrook, A., Verosloff, M., & Lucks, J. B. (2017). Computational design of small transcription activating RNAs for versatile and dynamic gene regulation. Nature Communications, 8(1), 1051. View publication.

Last updated: June 19, 2019 Back