Engineering Synthetic RNA Devices for Cell Control (Review) PDF

Reviews Engineering synthetic RNA devices for cell control Peter B. Dykstra1, Matias Kaplan 1...

Reviews Engineering synthetic RNA devices for cell control Peter B. Dykstra1, Matias Kaplan 1 and Christina D. Smolke 1,2 ✉ Abstract | The versatility of RNA in sensing and interacting with small molecules, proteins and other nucleic acids while encoding genetic instructions for protein translation makes it a power- ful substrate for engineering biological systems. RNA devices integrate cellular information sensing, processing and actuation of specific signals into defined functions and have yielded programmable biological systems and novel therapeutics of increasing sophistication. However, challenges centred on expanding the range of analytes that can be sensed and adding new mechanisms of action have hindered the full realization of the field’s promise. Here, we describe recent advances that address these limitations and point to a significant maturation of synthetic RNA-based devices. Sensor RNA is more than a messenger. Genetics research has ability to predict or select for sequences that can bind An element that can detect not only cemented the principles of gene expression and report binding. Having diverse options for selecting signals, including nucleic acid control — that RNA molecules transmit genetic infor- an input ligand tailored to a given system is important sequences, proteins, small mation between DNA and protein — but also uncovered to advance applications and RNA engineering appro­ molecules or non-biological stimuli such as temperature the significant role that RNA plays in directly regulating aches have not yet scaled sensor development. However, and light. cell behaviour (Fig. 1a). Because of its structural flexi- recent developments in high-throughput strategies for bility and ability to sense and interact with a range of generating RNA components that sense new analytes Actuator inputs, RNA has proven to be an important tool in engi- may rapidly change the landscape12. An element that can control neering research and sits on the precipice of widespread Another major challenge has been the limited mech- a process or event. application on a par with that of proteins and DNA. anisms of action by which synthetic RNA devices can Gene-regulatory RNA Synthetic biologists take inspiration from diverse meaningfully affect cell function. Early RNA devices that elements fields of engineering to develop frameworks for the focused on the modulation of mRNA levels and transla- RNA elements that control design, construction and characterization of new bio- tional efficiency within cells10 tapped into a small subset expression of a gene. logical systems1–4. A core concept is the engineering of of the wider possibilities for RNA modulation. Recently, RNA devices systems that can receive an input signal (via a sensor), novel design techniques around protein translation Engineered genetically perform an operation and generate an interpreta- have demonstrated strategies that can achieve increased encoded RNA elements that ble output (via an actuator)5. Endeavours to engineer dynamic range and temporal control13–16. Furthermore, combine sensing and actuation RNA benefit from customizable ligand-sensing and approaches such as the use of synthetic components to activities. gene-regulatory RNA elements and tunable response char- harness natural processes and the de novo utilization of Riboswitches acteristics. The advantageous qualities of RNA have led RNA for novel and artificial cellular functions showcase Natural RNA elements that to the development of diverse synthetic RNA devices advanced mechanisms that open new avenues for the conditionally regulate gene that sense, process and act upon genetic and molecular future of RNA engineering14,17. expression in response to binding of a small molecule. information6 (Fig. 1b). In this Review, we discuss current applications of Early RNA gene control elements were developed RNA devices spanning research, biomanufacturing and shortly after the discovery of functional riboswitches clinical uses. We then review advances in expanding ana- in nature7–9. Owing to the central role of RNA in gene lyte sensing and mechanisms of action with engineered expression, many of these initial engineered systems RNA. Recent reviews have comprehensively covered the 1 Department of focused on regulating translation to affect protein topics of aptamer selection via systematic evolution of Bioengineering, Stanford levels7,8,10. Further advances generated RNA devices that ligands by exponential enrichment (SELEX)18 and mam- University, Stanford, CA, USA. integrated sensing and control activities and enabled malian post-transcriptional circuits19. Here, we examine 2 Chan Zuckerberg Biohub, San Francisco, CA, USA. more precise modulation of cellular functions11. But novel strategies to create new ligand sensors and engi- ✉e-mail: csmolke@ whereas the RNA engineering field matured, several lim- neered RNAs that regulate diverse cellular processes. stanford.edu itations persisted. One long-standing challenge in RNA We conclude with a perspective on the maturation of https://doi.org/10.1038/ device engineering is the limited set of analytes that can the RNA device field and areas where we anticipate new s41576-021-00436-7 be sensed by RNA aptamers due to limitations in our technologies will drive further key advances. NAture Reviews | GENEtiCs volume 23 | April 2022 | 215 0123456789();: Reviews a Nucleus Cytosol Genomic DNA Non-coding RNA Spliceosome Transcription (snRNA and Translational proteins) complex Translation Ribosome Pre mRNA (rRNA and lncRNA circRNA Splicing, proteins) production mRNA maturation mRNA mRNA miRNA tRNA b Natural sequences Selection Rational design Components Sensors Actuators Self-cleavage A6 methylation Ribosome Alternative Small Protein Nucleic acid Temperature processivity splicing molecule m6A Δ OR RNA device Small molecule Self-cleavage mRNA degradation Translation Aptamers Components of synthetic RNA devices supported by the generation of well-characterized, bio- Nucleic acid sequences that can bind a particular ligand, A major research focus in synthetic biology is the devel- logical component parts20. RNA devices are engineered such as a small molecule or opment of modular frameworks for constructing geneti- RNA sequences that combine distinct sensor and actu- protein. cally encoded devices and engineered biological systems ator components to encode higher-order function than 216 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews ◀ Fig. 1 | RNA cellular functions. a | Following transcription, mRNAs are spliced by the sensitivity31–34. Over the past decade, roles for RNA beyond spliceosome, a complex of small nuclear RNAs (snRNA) and proteins, and matured by control of translation through the central dogma have been the addition of markers such as a 5′ cap (green circle) and polyA tail (green line)147,148. better understood — including microRNAs (miRNAs), Outside the nucleus, the canonical function of RNA is the combination of tRNAs, riboso- circular RNAs (circRNAs), long non-coding RNAs mal RNAs (rRNAs) (with accompanying ribosomal proteins, making up the ribosome) and (lncRNAs), spliceosome processing and CRISPR biology mRNAs to form translational complexes that produce most cellular proteins. Circular RNAs (circRNAs) are RNA molecules that form naturally via lariat intermediates as a — providing opportunities to expand the mechanisms splicing by-product149. RNA regulation of cellular processes can be accomplished by and systems through which RNA devices can act. circRNAs, as well as long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) that can regulate levels of exogenous RNA, endogenous RNA and protein expression150,151. b | RNA Emerging applications of RNA devices device engineering. Parts for RNA devices can be obtained by mining natural sequences, RNA devices enable advances in basic research. RNA selection experiments or rational design of sequences. Sensing and actuating compo- devices have been applied as tools to address specific nents can be characterized and combined to form an RNA device with both sensing and bottlenecks and propel basic research forward. One actuation capabilities. such bottleneck is the need to determine the cell state by detecting endogenous protein expression. Techniques the components by themselves (Fig. 1b). In general, the to monitor protein levels such as western blotting, sensor and actuator components are physically coupled immuno­staining and liquid chromatography–mass spec- such that binding of a ligand to the sensor component trometry face difficulties in detecting proteins in living alters the activity of the actuator component5. The quan- cells. RNA devices offer a promising solution by com- titative activity of the actuator component is therefore bining a protein-responsive RNA aptamer with a simple a read-out of the bound state of the sensor component fluorescent read-out. One example of this approach is an and, ultimately, the concentration of the target ligand in mRNA device design that can distinguish between undif- the cellular environment. ferentiated human induced pluripotent stem cells and RNA sensors can detect a range of biological inputs differentiated cells by quantifying endogenous LIN28A including nucleic acids, proteins and small molecules protein levels35 (Fig. 2Aa). This categorization method (Fig. 1b). RNA aptamers are a class of RNA sensors that based on detecting key endogenous protein levels could bind molecular ligands via direct binding interactions. pave the way for future strategies in live-cell detection The quantitative activity of an RNA device is highly or purification of specific cell types for regenerative dependent on aptamer affinity and specificity. The ligand medicine. concentration range that a device is capable of report- RNA devices can also be used to study RNA fold- ing, its sensitivity (this is often reported as the EC50, the ing inside living cells. Current techniques for studying half maximal effective concentration), is closely coupled in vivo RNA folding are lacking, but advances in the to the affinity of the incorporated aptamer. Cells con- engineering of fluorescent RNA aptamers could poten- tain many structurally similar molecules and aptamer tially fill this need. Certain RNA aptamers can combine specificity is critical to ensuring that the device reports with specific small molecules to create a fluorescent on the desired molecular target. Starting from the complex. Researchers have improved upon early demon- analyte-sensing RNA sequences found in natural sys- strations in this space36, creating an RNA aptamer-based tems, researchers have worked to expand the number Förster resonance energy transfer (FRET) system and diversity of molecular ligands that can be detected by by placing fluorescent aptamers on RNA scaffolds37 RNA. Efforts have been split between mining genomes (Fig. 2Ab). This apta-FRET system reports on RNA con- for new aptamers21 and generating de novo aptamers formational changes and offers a general tool for stud- through high-throughput selection22,23. Both of these ying the folding and conformational changes of natural methods have faced limitations in generating new RNA and artificial RNA structures in vivo. aptamers, with RNA aptamers against approximately 100 RNA devices also provide new ways to improve small molecules and 400 proteins currently reported24–27. disease models. For instance, Caenorhabditis elegans RNA actuators are typically encoded by gene- can be used to study diseases such as Huntington’s regulatory or enzymatic components that control a process disease, but suffers as a model system from a dearth or event (Fig. 1b). In early examples, RNA devices inspired of convenient methods for achieving conditional by natural riboswitches were used to control translation gene expression induction. Recent work described a of gene expression in bacteria10. These RNA devices were tetracycline-dependent ribozyme switch that can con- located within the 5′ untranslated region of a target mRNA fer inducible control when inserted into the 3′ untrans- and modulated translation initiation of the encoded gene lated region38 (Fig. 2Ac). This system was used to establish sequence by controlling ribosome binding site acces- an inducible C. elegans polyglutamine Huntington’s sibility via alteration of RNA structure in that region10. disease model that exhibited ligand-controlled Other early RNA devices modulated translation levels polyQ-huntingtin expression, inclusion body formation by controlling the degradation rate of the target mRNA and toxicity. Although this inducible gene expression with a self-cleaving ribozyme switch28. Additional designs platform provides notable advantages over attempts to have incorporated actuators that work through RNA combine multiple components into the host genome, the Ribozyme switch interference-mediated silencing29, antisense-mediated dynamic range of the gene induction was limited com- A type of riboswitch that uses a silencing30 or transcription termination30. The majority pared with other methods used in C. elegans. Future ribozyme, an RNA element that acts through cleaving of RNA devices described to date utilize actuators that efforts may leverage reported screening strategies to RNA, to encode the actuation control gene expression, and more generally have faced optimize switch activity that increase the dynamic range component. limitations in gene-regulatory dynamic range and ligand in a desired model system. NAture Reviews | GENEtiCs volume 23 | April 2022 | 217 0123456789();: Reviews A Basic research a Distinguish cell type by endogenous protein detection b Aptamer-based FRET system c Disease modelling in model organisms Reporter High FRET Reporter Reporter polyQ-huntingtin Reporter Translation ON RNA invader + RNA invader Reporter Endogenous Wild-type Inducible protein C. elegans Huntington’s disease model Reporter Low FRET Translation OFF B Biomanufacturing a High-throughput monitoring of intracellular concentrations b Dynamic control of metabolic flux Peptidoglycan synthesis Glucose Metabolite processing GlcNAc Glycolysis C Human health a Regulation of T cell proliferation b Gene editing control and c Inducible control of gene therapy transcriptional regulation miRNA dCas9 processing miRNA Antisense (6R)-FA IL-2 Apoptosis Anti-VEGF Anti-VEGF Inhibited + + FA aptamer IL-2 Inhibited dCas9 miRNA Expressed Repression GFP AAA processing Cell survival Antisense Fig. 2 | RNA devices enable diverse applications. Versatility of RNA small-molecule concentrations can be monitored with ribozyme switches highlighted in the wide application space of RNA devices ranging from engineered to respond to specific metabolites39. Bb | Metabolite-responsive addressing bottlenecks in basic research (part A), to providing ribozyme switches integrated into cellular pathways enable reprogramming advancements in biomanufacturing (part B) to propelling new frontiers in of networks for dynamic control of metabolic flux40. Ca | Insertion of a human health (part C). Aa | Detecting and quantifying endogenous LIN28A 6R-folinic acid (6R-FA)-responsive aptamer into microRNA (miRNA) protein levels enables qualitative distinction between undifferentiated switches enables regulation of T cell proliferation through control of miRNA human induced pluripotent stem cells and differentiated cells35. Ab | RNA processing44. Cb | CRISPR guide RNA (gRNA) engineered with RNA aptamers can be combined to create an aptamer-based Förster resonance aptamers that use a strand-displacement mechanism can lead to energy transfer (FRET) system for studying RNA folding inside living cells37. transcriptional regulation by ‘dead’ CRISPR–Cas9 (dCas9) (ref. 45). Ac | A tetracycline-dependent ribozyme switch controls polyQ-huntingtin Cc | A ribozyme switch can enable inducible control of anti-vascular expression and inclusion body formation in a novel inducible Caenorhabditis endothelial growth factor (anti-VEGF) proteins in a mouse model of wet elegans polyglutamine Huntington’s disease model38. Ba | Cytoplasmic age-related macular degeneration (AMD)56. GlcNAc, N-acetylglucosamine. 218 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews RNA devices detect and enhance metabolite produc- can show powerful efficacy, but CAR-T cell-associated tion. The ability to sense and monitor intracellular toxicities such as abnormally high cytokine production concentrations of metabolites is important for using and massive in vivo T cell expansion have presented a biological systems to produce valuable small mole- major obstacle41. In response, the field has sought dif- cules at scale. Although existing technologies such as ferent ways to tailor individual treatments and develop high-performance liquid chromatography, liquid chro- safeguards against CAR-T cell-associated toxicities, matography–mass spectrometry and others can be used including suicide switches to control cell death42. One to quantify concentrations of metabolites in biological early foray into genetic control of T cell proliferation samples, such analytical assays are of low throughput used ribozyme switches to modulate expression levels of and only provide bulk assay measurements. An RNA multiple cytokines, directing cells to either proliferate or device responsive to a specific metabolite offers a undergo apoptosis43. Theophylline-mediated regulation high-throughput, real-time and, potentially, single-cell of clonal T cell growth was demonstrated over 2 weeks alternative to current methods. In one example, in vitro in vivo along with effective control over primary human selection was used to generate a series of engineered T cell proliferation. More recent work focused on the ribozyme switches that sense and respond to cyclic regulation of endogenous cytokine receptor subunits diguanosyl-5′-monophosphate (c-di-GMP)39 (Fig. 2Ba). and developed synthetic miRNA-based RNA devices The ribozyme switches were constructed by joining a responsive to the biologically inert drug leucovorin self-cleaving hammerhead ribozyme (HHRz) to an ((6R)-folinic acid (6R-FA))44 (Fig. 2Ca). Combinatorial aptamer from a natural c-di-GMP riboswitch. These targeting of multiple receptor subunits with the incorpo- detectors quantitatively estimated cytoplasmic con- ration of multiple miRNA switches responsive to (6R)-FA centrations of c-di-GMP as low as 90 nM by assaying resulted in dynamic regulation of T cell growth, with up Escherichia coli lysate, thus demonstrating their use as to 4-fold higher growth after addition of (6R)-FA and convenient tools to monitor metabolite levels. up to 26-fold diminished growth after removal of (6R)-FA. More recent work also showcases strategies that Taken together, these approaches offer a versatile strategy incorporate metabolite-responsive ribozyme switches to address CAR-T cell-associated toxicities. into cellular systems for dynamic control of metabolic Applications requiring direct editing of the genome flux. In an effort to overproduce N-acetylglucosamine with CRISPR–Cas can also benefit from RNA devices. (GlcNAc), metabolic networks in Bacillus subtilis were Leveraging the dependence of the CRISPR system on reprogrammed using a glucosamine-6-phosphate a guide RNA (gRNA), researchers designed a control (GlcN6P)-responsive glmS ribozyme switch40 (Fig. 2Bb). system that incorporates small molecule-responsive By linking the GlcN6P-responsive ribozyme switches to aptamers into the 3′ region of the gRNA to modulate the regulation of multiple target proteins involved in the its structure45 (Fig. 2Cb). Initially, the spacer (which is peptidoglycan synthesis, glycolysis and GlcNAc synthe- complementary to the target sequence) is occluded. sis pathways, researchers demonstrated an engineered Ligand binding exposes the spacer sequence. The cell system that doubled the total GlcNAc titre. GlcN6P is exposed sequence allows a catalytically inert ‘dead’ an essential metabolite with variable concentrations and CRISPR–Cas9 (dCas9) to bind to the target sequence although the system worked well at fairly high GlcN6P and silence the gene. More recent work demonstrated concentrations, the limited dynamic range of the glmS similar small molecule-responsive control of CRISPR ribozyme switch meant that optimal GlcNAc synthesis interference (CRISPRi) activity by incorporating similar was difficult to achieve at low GlcN6P levels. However, RNA devices into CRISPR–Cpf1, showcasing the versa- because the intermediate GlcN6P is involved in the bio- tility of this strategy46. A separate example incorporated synthesis of numerous products in addition to GlcNAc, small molecule-responsive ribozyme switches into the with further optimization of the glmS ribozyme switch, 5′ region of the gRNA. Here, the spacer is occluded by this dynamic control strategy can enhance engineering a complementary strand. In contrast to the above struc- of additional metabolic pathways for small-molecule tural rearrangement mechanism, the spacer is exposed, production in B. subtilis and other industrial microbes. and the complex activated, when the complementary strand is cleaved by the cis-acting ribozyme in the Human health as the next frontier of RNA engineering. presence of a ligand47. Advances in RNA device engineering have propelled Small molecule-responsive RNA controllers still applications in human health. The incorporation of face challenges related to the specificity and affin- RNA devices offers distinct advantages in human health ity of the incorporated aptamers. A more clinically applications by facilitating therapeutic applications owing advanced method for cell-specific control of translation to their ability to encode multiple activities (including via genome editing with a CRISPR effector has been sense and respond functions) into a small genetic foot- demonstrated through the use of miRNA-controlled print for cellular delivery. The human immune response translation48–50. This system combined a CRISPR effec- is a key area of interest for applying engineering biolog- tor and an anti-CRISPR gene with miRNA response ical tools to medicine. T cell immunotherapies, such as elements (MRE) in the 3′ untranslated region. The chimeric antigen receptor T cell (CAR-T cell) therapies, described system leveraged tissue-specific expression of seek to reprogramme patients’ immune cells to find and miRNAs to direct repression of the anti-CRISPR pro- attack cancer cells. However, a primary issue with using teins through the MREs and, thus, enable tissue-specific CAR-T cells is the lack of ability to control cell behav- inhibition or activation of the CRISPR effector. The sys- iour and function in vivo after cell delivery. CAR-T cells tem was demonstrated in an in vivo mouse model, where NAture Reviews | GENEtiCs volume 23 | April 2022 | 219 0123456789();: Reviews MREs complementary to miR-122 (a liver-specific device could treat hepatocellular carcinoma by enabling miRNA) directed gene editing to occur only in the liver, a post-transcriptionally regulated RNA replacement even with the RNA device delivered to all tissues via an strategy. In an alternative strategy, RNA aptamers were adeno-associated viral (AAV) vector51. These advances used to target and induce apoptosis in human NCI-H460 in incorporating RNA devices into CRISPR systems offer cells, a model of non-small cell lung cancer59. The RNA new strategies for temporal and conditional control of aptamers were capable of inhibiting NCI-H460 cell via- genome editing and transcriptional regulation that could bility in a dose-dependent manner, with half-maximal lead to reductions in off-target effects52. inhibitory concentration (IC50) values in the 100 nM Clinical gene therapy is another health application range. Although there is potential immunogenicity from space where the incorporation of RNA devices may the AAV vector60, this method, once delivered, draws improve clinical outcomes and safety. A primary method on the small size, specific binding and tissue-penetrative of gene therapy uses AAV vectors to deliver therapeu- ability of RNA aptamer-based devices for targeted tic genes to the patient. The small genetic size of RNA cancer therapy. devices offers particular benefits for AAV vectors, which One concern for the use of RNA devices in human suffer from size constraints, only packaging ~4.7 kb53. health applications is the natural immunogenicity of Engineered ligand-responsive and miRNA-responsive RNA61. The ubiquitous presence of RNA viruses has RNA devices provide a compact control system, requir- marked RNA and, especially, double-stranded RNA to be a ing just ~100 bp of plasmid space54. In one example, a particularly strong pathogen-associated molecular pattern guanine-responsive ribozyme switch was used to sup- (PAMP), leading to rapid immune responses in humans press transgene expression during AAV production and through pathways such as RIG-I62,63. This immuno­ increase subsequent AAV vector yield up to 23-fold after genicity has been shown to be reduced by replacement addition of guanine54. Later work demonstrated the port- of uridines with N1-methylpseudouridine, RNA circu- ability of the RNA device strategy by inserting the same larization and RNA designs that lack double-stranded guanine-responsive ribozyme switch to repress trans- regions64–68. Immunogenicity can be further mitigated by gene expression from a replication-incompetent vesicu- masking RNA PAMPs within lipid nanoparticles during lar stomatitis virus (VSV) vector by more than 26-fold in delivery69. These advances in addressing RNA immuno­ mammalian cells55. The use of small molecule-responsive genicity offer various strategies to streamline the RNA devices has also been demonstrated in a mouse translation of RNA devices to the clinic. model of wet age-related macular degeneration (AMD)56 (Fig. 2Cc). Current treatments for wet AMD Cellular information sensing use repeated large-bolus injections of anti-vascular The ability to design cell-control devices requires scalable endothelial growth factor (anti-VEGF) agents into the strategies for developing genetically encoded aptamers eye. Repetitive bolus dosing is associated with acceler- with affinities and specificities that enable recognition ated retinal degeneration, and there is a clear need to of diverse cellular signals in the complex background of develop a long-term, single-use treatment for wet AMD the cell environment. Although advances have also been that allows for tight temporal control of anti-VEGF agent made in the development of RNA molecules for sensing expression. In this application, a tetracycline-responsive nucleic acid sequences, proteins and even non-biological ribozyme switch delivered via a recombinant AAV vec- stimuli, such as temperature and light70–73, we focus tor enabled small molecule-controlled overexpression of here on recent advances in strategies for the scalable an anti-VEGF protein and led to near complete inhibi- generation of RNA aptamers for small-molecule sensing. tion of wet AMD disease lesions in mice56. More recent work showcased a tetracycline-responsive ribozyme SELEX as a model for de novo generation of nucleic acid switch also able to effectively control AAV-mediated aptamers. High-throughput selection techniques such transgene expression that demonstrated 15-fold induc- as SELEX have been successful in scaling the number tion, reversibility and multiorgan functionality — key of aptamers that sense proteins, but selection for aptam- clinical features that highlight the applicability of RNA ers against small-molecule ligands has had limited devices for gene therapy57. success74. This discrepancy is largely due to the enrich- Finally, RNA devices have been applied to targeted ment strategies used in conventional high-throughput cancer therapies using multiple methodologies that may selection approaches. A large nucleic acid library is mixed overcome the clinical barriers of high cost, laborious with the ligand(s) of interest, and the sequences that techniques or safety concerns common to other treat- bind to the ligand are then separated from those that do ments. One recently described approach takes advan- not bind the ligand. Separation techniques generally tage of a trans-splicing ribozyme to selectively inhibit utilize the size or mass difference between an aptamer hepatocellular carcinoma up to 86% through specific and the aptamer–ligand complex to recover the ligand- replacement of telomerase reverse transcriptase (TERT) binding sequences and favour methods that can be per- RNA in human TERT (hTERT)-positive liver cancers58. formed in high throughput (for example, membrane Adenovirus was used to deliver a trans-splicing ribo- binding or capillary electrophoresis). zyme under the control of miR-122, a miRNA that is Although there is a large size difference between highly expressed in the liver, that specifically targeted, the aptamer and the aptamer–ligand complex when the cleaved and then ligated the hTERT mRNA intro- ligand is a protein, this is not the case when the ligand ducing a therapeutic gene RNA modification. This is a small molecule. Methods developed to recover approach demonstrated that a cancer-specific RNA aptamer–small-molecule ligand complexes generally 220 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews covalently link the small molecule, through a linker, to RNAs to be released from the beads. The released RNA purification modalities such as magnetic beads, resins or is then recovered and separated from the remaining cap- a glass surface. Conjugating the small-molecule ligand tured RNAs, reverse transcribed and sequenced, or put to such linkers results in several challenges, including through further rounds of selection. After nine rounds a lower-throughput process and enrichment of aptam- of selection, Capture-SELEX generated RNA aptamers ers that bind to the conjugated molecule rather than the against adenosine triphosphate (ATP) in a mixture of unconjugated molecule. The purification tags are nor- multiple competing small molecules85. However, new mally conjugated onto reactive elements in the target aptamers created with Capture-SELEX are developed molecule, removing functional groups that can interact outside a biosensor context and must be subsequently with the aptamer22,75. incorporated into an RNA device framework for use in The critical properties of an RNA aptamer are its cellular control, which may require further modification specificity (its ability to distinguish between mole- and testing. Improvements in the method would allow cules of similar structures) and affinity (how tightly it direct selection of aptamers in their final RNA device binds the ligand). High-specificity aptamers are often context and scale up the development of new aptamers. required for RNA devices used in cellular applications where many similarly structured molecules to the tar- DRIVER utilizes ribozyme cleavage to automate selec- get ligand may be present. Such instances include dis- tion strategy. A second recently described method, criminating between small molecules in a biosynthetic DRIVER (de novo rapid in vitro evolution of RNA bio- pathway or sensing specific molecules in a diagnostic sensors), relies on sequence changes between the ligand setting. The required affinity of the aptamer incorpo- bound and unbound states of an aptamer-coupled ribo- rated in an RNA device will depend on the intended zyme device to enable an entirely in-solution selection application and the anticipated concentration range strategy that can be executed autonomously on a liquid of the ligand in the cell environment. Ultimately, the handling robot81 (Fig. 3b). DRIVER utilizes a device EC50 value of the RNA device, which is directly related framework that couples an aptamer to a ribozyme, in to the affinity of the incorporated aptamer76, should which binding of the ligand to the aptamer controls the be within the biologically relevant range of the target ribozyme’s self-cleavage activity. A library is created that ligand. Both specificity and affinity can be tailored dur- contains two randomized regions, a shorter sequence of ing the selection process for an aptamer. For example, up to 8 random nucleotides and a longer sequence of up negative or counterselection steps (that is, selecting to 60 random nucleotides, on loops I and II of the HHRz. against ligands of similar structures) can be incorpo- These libraries are transcribed in the presence or absence rated to tune the specificity of an aptamer77. Similarly, of a small-molecule ligand mixture. A given HHRz mole­ affinity can be increased by lowering the ligand con- cule will either remain intact or undergo self-cleavage centration as the selection progresses to place selective depending on its folded structure and ligand-bound pressure for higher-affinity aptamers75,78,79. Finally, once state. A novel splint oligonucleotide is used to specifi- an aptamer is generated, further tuning of specificity cally recover and amplify cleaved sequences from the and affinity is possible through directed mutagenesis reaction mixture, whereas a standard primer is used or rational engineering80–83. Despite advances in RNA to recover uncleaved sequences. The reaction is then aptamer selection and engineering, the total number run over multiple cycles with or without ligand pres- of small molecules that can be sensed remains lim- ent. In the absence of ligand, only cleaved sequences are ited, and the imminent need for increased numbers of recovered and in the presence of ligand, only uncleaved genetically encoded sensors requires an exponential sequences are recovered. The alternating cycles allow for expansion of our current throughput. Overcoming selection of ligand-responsive ribozyme switches. this barrier will require focused effort on improving In the original demonstration of the DRIVER enrichment efficiency of small-molecule aptamer selec- method, small molecule-responsive RNA biosensors tion, post-selection tuning of aptamer properties and with nanomolar and micromolar affinities were gener- computational modelling of RNA–ligand binding. ated against six small molecules, five of which had no previously reported aptamers81. Aptamers created via Capture-SELEX enhances selection for small-molecule DRIVER can be used in the HHRz framework directly ligands. One recently described method, Capture-SELEX, as genetic controllers or can be isolated and used sep- utilizes structural changes between the ligand bound and arately. However, isolation of functional switches via unbound states of an RNA aptamer to enable a more effi- DRIVER can require more than 100 rounds of selec- cient selection strategy for non-modified small-molecule tion owing to low enrichment efficiencies caused by the ligands23,84 (Fig. 3a). This method immobilizes biotiny- persistence of sequences that naively fold into cleaving lated oligonucleotides, called capture oligos, onto a and non-cleaving conformations independent of ligand streptavidin-covered magnetic bead. An RNA library concentration86. DRIVER makes such intensive selection consisting of 10 randomized nucleotides and 40 rando­ viable by automating the process on a liquid handling mized nucleotides on either side of a capture sequence robot system so that an entire selection experiment complementary to the capture oligos is transcribed and against multiple ligands in parallel can be performed in allowed to bind to the capture oligos. Small molecules a high-throughput manner. Although DRIVER exhibits of interest are added to the mixture and a small sub- robust ability for novel sensor generation, some limi- set of the RNAs will bind to the ligands, disturbing the tations require consideration in its broad application. RNA–DNA interaction and causing the ligand-binding Because bulk selection conditions in DRIVER are not NAture Reviews | GENEtiCs volume 23 | April 2022 | 221 0123456789();: Reviews 5′ Constant Docking sequence 3′ Constant a Library N40 N10 Biotinylated oligos Biotin Transcription and Docking reverse transcription sequence complement Elution of target-binding Capture-SELEX cycle Hybridization RNAs Immobilization Streptavidin-coated Target magnetic beads molecules 5′ Constant Core 3′ Constant b Library N30 or 60 N4–8 NGS for Target analysis molecules Transcription DRIVER cycle Uncleaved regenerated library Selective PCR Rounds oscillate between OR to regenerate selection for cleaved or Self-cleavage OR library uncleaved product via selective PCR Uncleaved Cleaved switch switch Reverse transcription Cleaved regenerated and ligation library Reverse Txn Reverse Txn OR Prefix ligation Uncleaved cDNA Prefix Cleaved cDNA 222 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews ◀ Fig. 3 | Novel methods accelerate RNA sensor selection. Overview of two methods for and Twister ribozymes) and U1-snRNP polyadenyla- selection of RNA aptamers or sensors. Both methods begin with a diverse DNA library tion mechanisms in human cells. Taken together, these consisting of 1012–1015 distinct members. a | In the Capture-SELEX (systematic evolution methods also showcase the modularity possible with of ligands by exponential enrichment) method23,84,85, initial sequences consist of a 5′ and RNA parts — aptamers generated in one context can be 3′ constant region flanking two randomized regions of 10 or 40 nucleotides with a incorporated across diverse device platforms. constant docking sequence in the centre. The transcribed library is then hybridized to a biotinylated oligonucleotide via base pairing of the docking sequence. The complexes are immobilized on streptavidin-coated magnetic beads and mixed with the target RNA processing and cell control molecules. RNA sequences that bind the target molecules and undock from the RNA devices are integrated into cellular regulatory net- biotinylated oligonucleotides are subsequently eluted from the beads, creating the works from which they can control and probe biological starting library for the next selection cycle. b | In the DRIVER (de novo rapid in vitro processes. Here, we highlight recent advances in engi- evolution of RNA biosensors) method81, libraries are designed based on a modified neered RNAs acting as actuators. These novel RNA regu­ hammerhead ribozyme (HHRz) from satellite RNA of tobacco ringspot virus (sTRSV) — a latory strategies provide opportunities for researchers to small, naturally occurring self-cleaving ribozyme — with the two loops replaced by either expand the mechanisms through which RNA devices a randomized 30–60mer or a randomized 4–8mer. Sequences are mixed with target can act. molecules. Desired RNA biosensor sequences cleave in the absence of the target molecule but bind to the target molecule when present, preventing cleavage. In iterative cycles of negative and positive selection, the RNA sequences are reverse transcribed RNA base modifications enable tuning of translation. (txn) and the resultant cDNA of cleaved sequences has a 5′ primer ligated that allows for One of the more nuanced forms of RNA regulation is selective PCR amplification of the sequences corresponding to either the cleaved or based on post-transcriptional modification of bases in uncleaved RNA. The amplified DNA library then serves as the starting library for a new RNA92. Sequencing of tRNAs revealed the first discov- round of DRIVER. Periodically, these libraries can be analysed following quantification ered RNA modifications to be pseudouridine (𝜓) and by next-generation sequencing (NGS). Part a reprinted with permission from ref.85, N-methylation of nucleotides that have been shown Elsevier. to increase processivity of the ribosome and lower the translational error frequency93,94. Newer sequenc- optimized to generate switches responsive to specific ing techniques, such as N6-methyladenosine (m6A) ligands, some DRIVER-derived sensors may require fur- sequencing (m6A-seq)95 and methylated RNA immuno- ther optimization post selection. The DRIVER method precipitation sequencing (MeRIP–seq)96, have allowed also only generated ribozyme switches responsive to a the mapping of the role of RNA post-transcriptional small subset of all tested ligands. It is unknown whether modification in biological processes. Using these tech- this is a limitation of the types of ligands that RNA can niques, m6A has been revealed as vital to gene control sense or a limitation of the pooled, multiplexed process and linked to cancer 92,97. m 6A modifications affect in which the DRIVER selections were performed. Future mRNA stability and translation through a set of meth- work should focus on scaling the throughput of DRIVER yltransferases, demethylases and proteins that can read and demonstrating the utility of DRIVER-selected out m6A modifications — so-called writers, erasers and switches in diverse applications. readers98. RNA actuators have been constructed that can add and remove m6A modifications from a target RNA Sequencing methods enable high-throughput optimi- by fusing methyltransferases to a catalytically inactive zation of existing aptamers. High-throughput meth- CRISPR-associated nuclease Cas13 (ref.99) (Fig. 4Aa). ods have recently been described with streamlined Leveraging Cas13, which can selectively target RNA workflows for generating RNA devices from existing complementary to its loaded gRNA, the fused meth- aptamers. The aptamer sequences are generally inte- yltransferase catalyses the conversion of specific aden- grated into an RNA device framework and then regions osines to m6A. This work showed that the translation of the device are randomized. As the aptamer sequence of mRNAs with known m6A sensitivity could be tuned already exists, the library sizes can be drastically reduced via these targeted modifications. Further studies are from those used in methods such as Capture-SELEX required to fully understand how m6A modifications and DRIVER. The smaller starting library sizes allow affect RNA stability and translation in cells, and whether researchers to collect activity data for all library mem- this approach can be expanded to the broader cellular bers, as well as run selection and functional assays in suite of readers, writers and erasers. Additionally, future the cell systems that the RNA devices will ultimately work can explore whether this type of artificial methyl- be used in87–91. For example, fluorescence-activated ation can be used to programme other processes such as cell sorting followed by sequencing (FACS–seq) and alternative splicing100. RNA-sequencing (RNA-seq) approaches were recently Advances in RNA-targeting proteins have led to a used to screen RNA device libraries based on existing renaissance in RNA base editing for scientific and ther- small-molecule aptamers to identify sequences that apeutic uses101,102. Although much of the attention of result in gene expression control in human cells87. As genome engineering is focused on editing DNA, RNA these approaches can suffer from time-intensive and editing may allow for non-permanent gene therapy cost-intensive steps, efforts have also focused on devel- techniques and targeting of conditions that can only be oping a high-throughput method for the identification altered post-transcriptionally. These actuation mecha- of synthetic RNA devices based on existing aptamers by nisms function by linking an RNA-targeting complex, barcode-free amplicon sequencing88. This method was such as CRISPR–Cas13, CRISPR–Cas9 or endogenous applied to screening RNA device libraries based on exist- human RNA-binding proteins to an enzyme capable ing tetracycline and guanine aptamers and leveraging of modifying an RNA base, such as adenosine deam- ribozyme cleavage (that is, HHRz, hepatitis delta virus inase acting on RNA (ADAR) domains103–105 (Fig. 4Ab). NAture Reviews | GENEtiCs volume 23 | April 2022 | 223 0123456789();: Reviews A RNA post-transcriptional modifications B CircRNA C RNA scaffolds Aa Targeted RNA methylation Ba Extended protein production Ca Control of metabolite production dCas13 Coding Enzyme B Guide RNA circRNA Enzyme A mRNA RNA scaffold A Protein m6A methyltransferase Targeted methylation No RNA RNA scaffold scaffold Coding circRNA No metabolite Metabolite mRNA production production m6A P P Ab Targeted RNA base editing Bb miRNA sequestration Cb Control of apoptosis Hepatitis C virus Apoptosis- A U G C regulatory L7Ae miR-122 miR-122 protein RNA ADAR scaffold miRNA H2O protection of virus RNA dCas13 miRNA ADAR sequestration circRNA No RNA RNA NH3 scaffold scaffold Guide RNA I U G U Inhibition of viral Viral replication replication Cell survival Apoptosis Fig. 4 | Emerging mechanisms of RNA processing and cell control for enzymes to modify RNA nucleobases for post-transcriptional gene editing. novel RNA devices. Recent findings have revealed new roles and Current editors utilize deaminases to convert adenosine and cytosine to mechanisms for RNA beyond control of translation through the central inosine and uracil, respectively103,104,106,152. Ba | circRNA expression vectors dogma, including targeted RNA base editing and post-transcriptional express proteins for longer than equivalent mRNA sequences 107. modifications (part A), circular RNA (circRNA) mechanisms (part B) and Bb | Synthetic circRNAs can be designed as microRNA (miRNA) sponges to novel uses of RNA scaffolds (part C). Future RNA devices can incorporate quench specific miRNAs and inhibit miRNA-dependent viral replication120. these new mechanisms. Aa | Post-transcriptional modifications are involved Ca | RNA can be repurposed as a scaffold to co-localize enzymes and in RNA regulation. N6-Methyladenosine (m6A) is a post-transcriptional increase local enzyme concentrations for control of metabolite modification affecting mRNA stability and translation. Methyltransferases production125. Cb | RNA scaffolds can also be constructed to induce can be fused to dCas13 (a catalytically inactive CRISPR-associated proximity oligomerization of Caspase 8 within the caspase pathway for nuclease) directed by a guide RNA to catalyse the conversion of adenosines apoptosis126. ADAR, adenosine deaminase acting on RNA. Part Aa is to m6A, allowing targeted artificial methylation of RNA which could be adapted from ref.99, Springer Nature Limited. Part Cb is adapted from further utilized99. Ab | RNA base editors use naturally occurring and evolved ref.126, CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/). The RNA-targeting domain brings the deaminase RNA device frameworks for conditional control of edit- enzyme in proximity with an adenosine, thus deami- ing functions. nating the adenosine into an inosine. Recent work has expanded the toolbox to allow for the editing of cyto- Circular RNAs as temporally stable translation platforms sines to uridines106. Future work may focus on devel- and molecular sponges. Another area of RNA regulation oping additional editing domains to allow for all RNA where our understanding of the underlying biological bases to be modified to every possible other base, as well processes has recently advanced is circRNA107. circRNAs as the incorporation of this regulatory mechanism into are RNA molecules that take the form of a covalently 224 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews closed loop due to post-transcriptional processing108 role circRNAs play and, thus, what unintentional effects (Fig. 1a) and have been shown to play a pathological overexpression of engineered circRNAs may have role in numerous diseases, including cancer, Alzheimer on cells. Advances in methods to detect and quantify disease and cardiovascular disease109–111. There is also circRNAs in cells are critical to supporting the develop- evidence that circRNAs can encode proteins and regu- ment of this RNA platform into a broader engineering late translation by serving as sponges for miRNAs and tool for the field. RNA-binding proteins112–114. Methods for synthesizing engineered circRNA RNA scaffolds serve as modular substrate for program- devices in cells have been described that use vari- ming spatial organization. Beyond these examples of ous mechanisms, including ribozyme cleavage and incorporating new RNA biology, scientists have uti- tRNA ligation, group I self-splicing introns and back- lized RNA to achieve genetically encodable, biological splicing115–117. Each of these methods was demonstrated elements that do not have known RNA counterparts in to create synthetic circRNAs that can be translated to nature. One way RNA can be repurposed is as a scaf- proteins in cells. It was hypothesized that circRNAs can fold for various cellular processes124. For instance, RNA persist longer in cells than mRNAs as they lack ends scaffolds have been engineered into cells to co-localize accessible to exonuclease activity; this increased stabil- enzymes in a biosynthetic pathway, increasing the local ity was demonstrated by showing that luciferase activ- concentration of the metabolites and enzymes, and thus ities from a circRNA and an mRNA were comparable, the rate at which reactants can be converted into path- whereas the circRNA exhibited a protein production way products. In one example, researchers working with half-life (that is, time for luciferase activity to decrease E. coli designed a set of RNA sequences encoding aptam- by 50%) of 80 h compared with 45 h for the compara- ers in a 2D scaffold and then co-expressed four enzymes ble mRNA116 (Fig. 4Ba). The persistence of circRNAs has involved in succinate production that were fused to the been leveraged as a platform for the expression of RNA corresponding aptamer-binding protein domains125 devices. Ribozyme-assisted circular RNAs (racRNAs) (Fig. 4Ca). The RNA-scaffolded pathway exhibited an are a class of RNA devices that can be expressed in cells 88% increase in succinate production when compared at high concentrations, increasing device concentra- with a control strain. tions from the low nanomolar to micromolar range115. In another example, scientists replaced a protein- The platform places the RNA sequence of interest based scaffold in the caspase pathway for apoptosis with between two ribozymes, where, once cleaved, the ends an RNA-based scaffold126 (Fig. 4Cb). Caspase 8 normally of the RNA are circularized by ligation through the requires a protein scaffold for proximity oligomeri- endogenous RNA ligase RtcB. This platform was used zation to occur, which triggers the cell-death cascade. to develop RNA devices that detect intracellular SAM Researchers fused Caspase 8 to L7Ae, an RNA-binding concentrations and inhibit the cell-signalling protein protein, and designed an RNA scaffold harbouring NF-κB118. L7Ae binding sequences. When the fusion protein and circRNAs can also be engineered to act as non-natural RNA scaffold were present in cells, the caspase path- miRNA sponges119. One recent study reported the engi- way was triggered, leading to programmed apoptosis. neering of a synthetic circRNA with multiple binding The authors built a targeted RNA device that induced sites for miR-122 to act as a miR-122 sponge120 (Fig. 4Bb). expression of the L7Ae–Caspase 8 fusion upon binding miR-122 protects the 5′ end of the hepatitis C virus of specific miRNAs, resulting in cell death. The result- (HCV) genome from degradation and enhances viral ing RNA device enabled selective control of cell death replication and translation. The engineered circRNAs in HeLa cells, achieving apoptosis rates more than dou- were transcribed and circularized in vitro, and subse- ble those of background levels (up to 70%) with the full quently transfected into human hepatoma cell lines L7Ae–Caspase 8 device. Huh-7 and Huh-7.5, where they sequestered miR-122 Strategies using engineered RNA as a platform to and resulted in lower HCV infection120,121. circRNAs assemble and organize molecules inside cells are cur- have also been developed as sponges for proteins122. rently limited in the size and diversity of known scaf- Researchers engineered and integrated a construct that folds. Development of future RNA scaffolds may require would become a circRNA with 100 CA dinucleotides the design of basic RNA building blocks that could be when transcribed in HEK293 cells. These CA repeats combined in a modular fashion to form diverse and pro- are a known binding site for heterogeneous nuclear rib- grammable structures that can control orientation and onuclear protein L (hnRNPL), a regulator of alternative organization inside cells. Diverse RNA elements could splicing. When the circRNA was expressed, a shift in ultimately be integrated throughout the resulting scaf- alternative splicing was observed across the genome. fold to achieve dynamic control over spatial organization These early studies show that circRNAs can be used for via conformational changes. long-lasting protein production in mammalian cells and function as sponges for other RNAs and RNA-binding Conclusions and perspective proteins. RNA science has flourished in the last decade. Two Despite these recent advances, circRNAs remain mRNA vaccines against SARS-CoV-2 were designed, underexplored in the field largely due to the diffi- manufactured and distributed to hundreds of millions culty in detecting and measuring the concentration of of people in the span of 18 months127–129. This achieve- circRNAs distinct from their cognate linear isoforms123. ment has highlighted the promise of engineered RNA As a result, we still do not fully understand the cellular and has led to substantial public and private-sector NAture Reviews | GENEtiCs volume 23 | April 2022 | 225 0123456789();: Reviews investments in both basic and applied research and of new statistical techniques for RNA tertiary structure development 130–133. Fulfilling the potential of this prediction138–142. Partnerships that focus on the further field will require partnerships between basic and applied development, application and scaling of these tools researchers, organizations, commercial entities and should be incentivized. funding bodies. Inspired by coordinated efforts in the protein and RNA engineering has lagged behind protein engi- enzymology fields, RNA foundries should be established neering, despite RNA’s simpler component parts and to focus on scaling high-throughput methods to gener- robust secondary-structure modelling. However, recent ate open-source libraries of standard, well-characterized progress in RNA engineering tool development pro- RNA components that can be used widely by the field vides optimism as similar advances accelerated the field in support of new device generation23,81,88. Scaling such of protein engineering. The goal of RNA device engi- efforts will ensure a critical open resource to advance neering must be to, at the very least, match nature’s level the broader adoption of RNA devices in research, of proficiency with RNA polymers. To achieve this, the commercial and clinical settings. Larger consortiums field should take inspiration from the development of should be established to focus on tool development to the protein design field. Protein engineers have access advance RNA biology and engineering, including work to large, well-annotated, public libraries of protein data. to improve in vitro transcription, RNA quantification, These shared resources have allowed for the develop- in vitro and in vivo RNA modifications, targeting of ment of powerful computational tools for de novo pro- RNA by small-molecule drugs and the delivery and tein design134–137. Building out similar shared resources immunogenicity of RNA therapeutics69,123,143–146. As for RNA will require coordination within the field. The momentum in the field continues to grow, engineered utility of medium-throughput RNA structural deter- RNA devices will fulfil their potential as precision mination has been spurred on by recent advances in engineered systems. cryogenic electron microscopy, nuclear magnetic reso- nance and chemical mapping, as well as the application Published online 4 January 2022 1. Nshogozabahizi, J. C., Aubrey, K. L., Ross, J. A. & 17. Anzalone, A. V., Lin, A. J., Zairis, S., Rabadan, R. & expression control in cyanobacteria. Appl. Microbiol. Thakor, N. Applications and limitations of regulatory Cornish, V. W. Reprogramming eukaryotic translation Biotechnol. 104, 1977–1991 (2020). RNA elements in synthetic biology and biotechnology. with ligand-responsive synthetic RNA switches. 34. Bayer, T. S. & Smolke, C. D. Programmable J. Appl. Microbiol. 127, 968–984 (2019). Nat. Methods 13, 453–458 (2016). ligand-controlled riboregulators of eukaryotic 2. Kim, J. & Franco, E. RNA nanotechnology in synthetic 18. Spöring, M., Finke, M. & Hartig, J. S. Aptamers in gene expression. Nat. Biotechnol. 23, 337–343 biology. Curr. Opin. Biotechnol. 63, 135–141 (2020). RNA-based switches of gene expression. Curr. Opin. (2005). 3. Schmidt, C. M. & Smolke, C. D. RNA switches for Biotechnol. 63, 34–40 (2020). 35. Kawasaki, S., Fujita, Y., Nagaike, T., Tomita, K. & synthetic biology. Cold Spring Harb. Perspect. Biol. 19. Kawasaki, S., Ono, H., Hirosawa, M. & Saito, H. Saito, H. Synthetic mRNA devices that detect 11, 135–141 (2019). RNA and protein-based nanodevices for mammalian endogenous proteins and distinguish mammalian 4. Park, S. V. et al. Catalytic RNA, ribozyme, and its post-transcriptional circuits. Curr. Opin. Biotechnol. cells. Nucleic Acids Res. 45, e117–e117 (2017). applications in synthetic biology. Biotechnol. Adv. 37, 63, 99–110 (2020). 36. Paige, J. S., Nguyen-Duc, T., Song, W. & Jaffrey, S. R. 107452 (2019). 20. Endy, D. Foundations for engineering biology. Nature Fluorescence imaging of cellular metabolites with 5. Win, M. N., Liang, J. C. & Smolke, C. D. Frameworks 438, 449–453 (2005). RNA. Science 335, 1194–1194 (2012). for programming biological function through 21. Barrick, J. E. & Breaker, R. R. The distributions, 37. Jepsen, M. D. E. et al. Development of a genetically RNA parts and devices. Chem. Biol. 16, 298–310 mechanisms, and structures of metabolite-binding encodable FRET system using fluorescent RNA (2009). riboswitches. Genome Biol. 8, R239 (2007). aptamers. Nat. Commun. 9, 18 (2018). 6. Liang, J. C., Bloom, R. J. & Smolke, C. D. Engineering 22. Ellington, A. D. & Szostak, J. W. In vitro selection of 38. Wurmthaler, L. A., Sack, M., Gense, K., Hartig, J. S. & biological systems with synthetic RNA molecules. RNA molecules that bind specific ligands. Nature 346, Gamerdinger, M. A tetracycline-dependent ribozyme Mol. Cell 43, 915–926 (2011). 818–822 (1990). switch allows conditional induction of gene expression 7. Nahvi, A. et al. Genetic control by a metabolite 23. Lauridsen, L. H., Doessing, H. B., Long, K. S. & in Caenorhabditis elegans. Nat. Commun. 10, 491 binding mRNA. Chem. Biol. 9, 1043–1049 (2002). Nielsen, A. T. in Synthetic Metabolic Pathways: (2019). 8. Winkler, W. C., Cohen-Chalamish, S. & Breaker, R. R. Methods and Protocols (eds. Jensen, M. K. & 39. Gu, H., Furukawa, K. & Breaker, R. R. Engineered An mRNA structure that controls gene expression Keasling, J. D.) 291–306 (Springer, 2018). allosteric ribozymes that sense the bacterial second by binding FMN. Proc. Natl Acad. Sci. USA 99, 24. Baird, G. S. Where are all the aptamers? Am. J. Clin. messenger cyclic diguanosyl 5′-monophosphate. 15908–15913 (2002). Pathol. 134, 529–531 (2010). Anal. Chem. 84, 4935–4941 (2012). 9. Sudarsan, N., Barrick, J. E. & Breaker, R. R. 25. Dunn, M. R., Jimenez, R. M. & Chaput, J. C. Analysis 40. Niu, T. et al. Engineering a glucosamine-6-phosphate Metabolite-binding RNA domains are present in of aptamer discovery and technology. Nat. Rev. Chem. responsive glmS ribozyme switch enables dynamic the genes of eukaryotes. RNA 9, 644–647 (2003). 1, 0076 (2017). control of metabolic flux in Bacillus subtilis for 10. Isaacs, F. J. et al. Engineered riboregulators enable 26. McKeague, M. & DeRosa, M. C. Challenges overproduction of N-acetylglucosamine. ACS Synth. post-transcriptional control of gene expression. and opportunities for small molecule aptamer Biol. 7, 2423–2435 (2018). Nat. Biotechnol. 22, 841–847 (2004). development. J. Nucleic Acids 2012, 1–20 (2012). 41. Sterner, R. C. & Sterner, R. M. CAR-T cell therapy: 11. Suess, B. & Weigand, J. E. Engineered riboswitches: 27. McKeague, M. et al. Analysis of in vitro aptamer current limitations and potential strategies. Blood overview, problems and trends. RNA Biol. 5, 24–29 selection parameters. J. Mol. Evol. 81, 150–161 Cancer J. 11, 1–11 (2021). (2008). (2015). 42. Di Stasi, A. et al. Inducible apoptosis as a safety 12. McKeague, M., Wong, R. S. & Smolke, C. D. 28. Valencia-Sanchez, M. A., Liu, J., Hannon, G. J. & switch for adoptive cell therapy. N. Engl. J. Med. 365, Opportunities in the design and application of RNA Parker, R. Control of translation and mRNA 1673–1683 (2011). for gene expression control. Nucleic Acids Res. 44, degradation by miRNAs and siRNAs. Genes Dev. 20, 43. Chen, Y. Y., Jensen, M. C. & Smolke, C. D. Genetic 2987–2999 (2016). 515–524 (2006). control of mammalian T-cell proliferation with 13. Westbrook, A. M. & Lucks, J. B. Achieving large 29. Bloom, R. J., Winkler, S. M. & Smolke, C. D. Synthetic synthetic RNA regulatory systems. Proc. Natl Acad. dynamic range control of gene expression with a feedback control using an RNAi-based gene-regulatory Sci. USA 107, 8531–8536 (2010). compact RNA transcription–translation regulator. device. J. Biol. Eng. 9, 5 (2015). 44. Wong, R. S., Chen, Y. Y. & Smolke, C. D. Regulation Nucleic Acids Res. 45, 5614–5624 (2017). 30. Lucks, J. B., Qi, L., Mutalik, V. K., Wang, D. & of T cell proliferation with drug-responsive 14. Kim, J. et al. De novo-designed translation-repressing Arkin, A. P. Versatile RNA-sensing transcriptional microRNA switches. Nucleic Acids Res. 46, riboregulators for multi-input cellular logic. regulators for engineering genetic networks. 1541–1552 (2018). Nat. Chem. Biol. 15, 1173–1182 (2019). Proc. Natl Acad. Sci. USA 108, 8617–8622 45. Liu, Y. et al. Directing cellular information flow 15. Rauch, S., Jones, K. A. & Dickinson, B. C. Small (2011). via CRISPR signal conductors. Nat. Methods 13, molecule-inducible RNA-targeting systems for 31. Ausländer, S. & Fussenegger, M. Synthetic RNA-based 938–944 (2016). temporal control of RNA regulation. ACS Cent. Sci. 6, switches for mammalian gene expression control. 46. Liu, Y. et al. Engineering cell signaling using tunable 1987–1996 (2020). Curr. Opin. Biotechnol. 48, 54–60 (2017). CRISPR–Cpf1-based transcription factors. Nat. 16. Chappell, J., Westbrook, A., Verosloff, M. & 32. Chappell, J. et al. The centrality of RNA for Commun. 8, 2095 (2017). Lucks, J. B. Computational design of small engineering gene expression. Biotechnol. J. 8, 47. Tang, W., Hu, J. H. & Liu, D. R. Aptazyme-embedded transcription activating RNAs for versatile and 1379–1395 (2013). guide RNAs enable ligand-responsive genome editing dynamic gene regulation. Nat. Commun. 8, 1051 33. Till, P., Toepel, J., Bühler, B., Mach, R. L. & and transcriptional activation. Nat. Commun. 8, (2017). Mach-Aigner, A. R. Regulatory systems for gene 15939 (2017). 226 | April 2022 | volume 23 www.nature.com/nrg 0123456789();:  Reviews 48. Hoffmann, M. D. et al. Cell-specific CRISPR–Cas9 73. Lotz, T. S. et al. A light-responsive RNA aptamer a Cas13-directed methyltransferase. Nat. Biotechnol. activation by microRNA-dependent expression of for an azobenzene derivative. Nucleic Acids Res. 47, 38, 1431–1440 (2020). anti-CRISPR proteins. Nucleic Acids Res. 47, e75 2029–2040 (2019). 100. Xiao, W. et al. Nuclear m6A reader YTHDC1 (2019). 74. Gold, L. et al. Aptamer-based multiplexed proteomic regulates mRNA splicing. Mol. Cell 61, 507–519 49. Hirosawa, M. et al. Cell-type-specific genome editing technology for biomarker discovery. PLoS ONE 5, (2016). with a microRNA-responsive CRISPR–Cas9 switch. e15004 (2010). 101. Rees, H. A. & Liu, D. R. Base editing: precision Nucleic Acids Res. 45, e118 (2017). 75. Darmostuk, M., Rimpelova, S., Gbelcova, H. & chemistry on the genome and transcriptome of living 50. Hirosawa, M., Fujita, Y. & Saito, H. Cell-type-specific Ruml, T. Current approaches in SELEX: an update cells. Nat. Rev. Genet. 19, 770–788 (2018). CRISPR activation with microRNA-responsive AcrllA4 to aptamer selection technology. Biotechnol. Adv. 33, 102. Marina, R. J., Brannan, K. W., Dong, K. D., Yee, B. A. switch. ACS Synth. Biol. 8, 1575–1582 (2019). 1141–1161 (2015).