Beyond the niche - unlocking the full potential of synthetic riboswitches

Abstract

Synthetic riboswitches have undergone great development in the past decade, evolving into valuable regulatory tools. Operating entirely at the RNA level and independently of auxiliary proteins, they offer a promising alternative to protein-based systems such as TetON/OFF or CRISPR-Cas. As compact, modular RNA elements they unite sensing and regulatory functions within a single molecule, giving them the advantages of high modularity, portability and low metabolic burden. Here, we explore the unique features of synthetic riboswitches, highlight key applications, assess current bottlenecks and limitations and put them in context with emerging solutions, to emphasise the potential of synthetic riboswitches.

Synthetic riboswitches are a burgeoning regulatory tool in the field of molecular biology. Here, the authors explore the unique features of synthetic riboswitches, highlight key applications, assess current bottlenecks and limitations and put them in context with emerging solutions.

Introduction

Synthetic riboswitches are RNA-based genetic control elements. They are compact, highly structured regulators located in untranslated regions of a target gene and control its expression in response to a small molecule ligand. Although their natural counterparts predominantly occur in bacteria and appear to have been lost in eukarya, synthetic riboswitches have been engineered to function in all branches of the tree of life¹. This versatility is based on their modular design (Fig. 1a). A binding domain (aptamer) recognizes the ligand in a completely protein-independent manner while an adjacent domain exerts control over gene expression. The binding of the ligand triggers a conformational change that is transmitted from the aptamer to the regulatory domain, thus modifying its behavior and rendering the domain response ligand dependent. The choice of the aptamer dictates to which ligand the riboswitch will respond, whereas the regulatory domain determines the modus operandi. The engineering of novel regulatory domains, some of which are unknown to exist in nature, has led to synthetic riboswitches that utilize a wide variety of mechanisms, such as blocking ribosomal scanning, controlling RNA stability, splicing, polyadenylation, or RNA interference^2–6. Even greater adaptability potential lies in the choice of the aptamer domain, which can either be adapted from nature or created entirely de novo using SELEX^7,8 (Systematic Evolution of Ligands by EXponential enrichment). With SELEX, aptamers against virtually any target molecule can be identified, which allows engineering of fully synthetic riboswitches that respond to customized ligands. Figure 2 provides a short historic overview of key developments in the field of synthetic riboswitch design.

Fig. 1. Schematic architecture of synthetic riboswitches and their applications.

a General architecture of a synthetic riboswitch is shown in the middle section, being constructed from an aptamer and a regulatory domain. The whole riboswitch undergoes conformational changes upon ligand binding, which alters the activity of the regulatory domain. Both domains can be exchanged, either changing the ligand they are responding to, or the regulated mechanism. b Illustration of projects utilizing synthetic riboswitches for different applications. A tetracycline-dependent aptazyme controlling mRNA stability is used in Caenorhabditis elegans to establish a Huntington’s disease model. In the absence of tetracycline the aptazyme is active, cleaving off the Poly-A tail of the mRNA, which leads to rapid degradation of the mRNA. In the presence of tetracycline, the aptazyme is inactive, keeping the mRNA stable and allowing gene expression³¹. c A synthetic riboswitch controlling mRNA splicing is used in resting B lymphocytes to regulate CD20 expression in a tetracycline-dependent manner. The tetracycline aptamer included the 3’ splice site in its basal stem and is placed on the 5’ end of a synthetic exon (sExon) including premature stop codons. In presence of tetracycline, the basal stem of the aptamer is stabilized, which masks the 3’ splice site from the splice machinery and excludes the synthetic exon from the mRNA. This leads to the correct mRNA sequence including no premature stop codons and ultimately allowing CD20 expression. In the absence of tetracycline however, the synthetic exon is included, leading to premature termination of translation and preventing CD20 expression³². d A synthetic lysine riboswitch was re-engineered from an OFF- to an ON-switch at its expression platform, enabling lysine-responsive control of intracellular transport and enhancing lysine yield in C. glutamicum by repressing undesired biosynthetic pathways³³. e Translation control in artificial cells using a histamine aptamer. The histamine aptamer is masking the Shine-Dalgarno sequence (SD) and the start codon (AUG) in the absence of histamine. In the presence of histamine, the binding component structure of the aptamer is stabilized, freeing the SD and the start codon and allowing translation³⁶.

Fig. 2. Timeline of synthetic riboswitch development.

Timeline showing milestones in the development of synthetic riboswitches. Icons in the upper left of the referenced publications indicate the application of the riboswitch in vitro, in bacteria, in yeast, in eukaryotes or a breakthrough in selection of aptamers. Anticipated future developments to improve synthetic riboswitches are indicated in light gray.

The responsiveness to small molecules sets synthetic riboswitches apart from other promising riboregulators, such as STAR activator and toehold switches that make use of strand displacement or devices that exploit RNA-binding proteins. Advances in these areas are covered by a number of excellent reviews and will therefore not be further discussed here^9–11. We are going to focus on synthetic riboswitches to explore their potential, identify current limitations and compare their capabilities with those of protein-based gene regulatory systems. Our aim is to highlight the unique advantages of synthetic riboswitches across a wide range of research fields and disciplines, showcase the progress made over recent decades, and outline key steps needed to unlock their full potential.

Unique advantages of riboswitches over protein-based systems

Synthetic riboswitches offer distinct advantages that make them valuable regulatory tools, particularly in settings where other systems may perform suboptimally or fail altogether¹². There is no question that the most commonly used systems for external control of gene expression are those that make use of regulatory proteins, such as LacI, TetR or CRISPR-Cas^13–15. While TetON/OFF and LacI provide accessible and proven ways of heterologous gene regulation in response to their inducers, CRISPR-Cas systems have revolutionized gene modification and regulation due to their programmability, modularity and broad applicability. In particular, dCas variants fused to activator or repressor domains can precisely regulate gene expression across various organisms, including human cells¹⁶. Despite their excellent performance, these systems share the limitation of requiring permanent expression of a heterologous protein as their central component. This requirement can lead to challenges such as imposing a considerable metabolic burden on the host cell, triggering host immune system responses, overloading the protein degradation machinery or causing retroactive effects^17–19. Such effects do not occur for synthetic riboswitches, due to their completely RNA-based nature. As cis-acting RNA elements, they combine the advantages of direct and fast control over the RNA molecule they are located on with being cofactor-independent, making them perfectly suited for applications where the co-expression of proteins is problematic. They impose substantially lower biosynthetic and degradation demands than protein-based regulators. This reduced metabolic burden, combined with their inherent orthogonality, enables more efficient resource allocation within the host cell, supporting higher biomass yields and more precise control of gene expression^20,21.

Another key characteristic of synthetic riboswitches is their modular architecture, which allows for a high degree of customizability while maintaining an overall compact size. The possibility to select sensory domains de novo for new target molecules by SELEX sets them apart from other regulatory systems that are limited to existing domains. For instance, changing the binding specificity of a protein towards a new ligand remains a mostly unresolved challenge to date. Moreover, the modularity of synthetic riboswitches also allows for designs utilizing one aptamer domain in combination with various different regulatory domains, thus enabling the control of different mechanisms in response to the same ligand²². This can either be used to control complex processes with different pathways in parallel or combined to leverage regulatory capability²³. Furthermore, new designs have been describing dual-input riboswitches responding to multiple ligands²⁴. Ultimately, synthetic riboswitches offer the possibility to be combined with any other RNA-utilizing mechanism, such as RNA interference or CRISPR-Cas^25,26. In sum, these features allow for the construction of highly sophisticated regulatory systems able to perform in very specialized settings, e.g. ligand-controlled gene editing or miRNA processing.

Despite their modularity, synthetic riboswitches are very small, often less than 200 nucleotides, adding little extra burden for the host cell during transcription as part of the RNA they control²¹. This is of particular importance for artificial cells that have to cope with limited resources, or in settings where the whole system is meant to be RNA-based^27,28 (e.g., RNA vaccines or RNA-based viral vectors). The compact design of synthetic riboswitches suggests good portability, a minimal footprint and an overall low risk of side effects. These criteria are of great value for application outside of model organisms²⁹. In particular, aptazymes (aptamer-controlled ribozymes) controlling mRNA stability or riboswitch-controlled synthetic exons are easily integrated into the host genome (into the 3’UTR or intronic sequences). Their overall reduced complexity compared to other regulatory systems minimizes the risk of unexpected side-effects. Specifically, unlike protein-based systems such as CRISPR-Cas, the inherent cis-acting nature of aptazymes, combined with precise genomic integration and tunable cleavage on/off rates, significantly mitigates the potential for off-target effects. Thus, they may be considered convenient tools that can be widely applied for regulatory purposes. Furthermore, the dose-dependent response of synthetic riboswitches enables precise and quantitative regulation of gene expression and cellular behaviors based on specific ligand concentrations³⁰. This characteristic, coupled with their inherent programmability, rapid response kinetics, and low metabolic cost, highlights their unique advantages over protein-based systems.

While synthetic riboswitches may exhibit a more limited dynamic range compared to protein-based systems, their distinct advantages lie in their rapid response kinetics and significantly lower metabolic cost. These properties enable precise and efficient gene regulation, making them particularly attractive for applications where speed and resource efficiency are paramount (see Table 1).

Table 1.

Comparison of riboswitches and protein-based systems

Performance	Riboswitches	Protein-based Systems	References
Dynamic range	10–300- fold, Dependent on direct sequence context	1000-fold, sometimes higher, Dependent on genomic context, Transcription factor levels, promoter characteristics, off-target	23,90,91,92,93 vs 94,95,96,97
Response time	Faster Maximum expression after 6–8 h	Slower Maximum expression within 24–48 h	98 vs 99,100,101
Metabolic cost	Lower Lower synthesis and degradation costs due to RNA nature	Higher Continuous heterologous protein synthesis/degradation for dynamic regulation	20,21 vs 102,103

Synthetic riboswitches as powerful tools

The great versatility of synthetic riboswitches is reflected in their successful application across a range of research questions. Among them is the creation of a Huntington’s disease model in Caenorhabditis elegans by controlling expression of the human Huntington exon 1 (Htt) with an extended polyglutamine tract closely mimicking a mechanistic cause of Huntington’s disease in humans³¹. The straight-forward insertion of a tetracycline-controlled aptazyme into the 3’ UTR of the Htt gene allowed the regulated expression through altered mRNA stability. In the absence of tetracycline, the aptazyme cleaves itself, leading to rapid degradation of the mRNA (Fig. 1b). Importantly, its insertion was achieved without altering the genetic context. This enabled the utilization of the neuron-specific rab-3p promoter and thus the correct, tissue-specific expression that was crucial for the establishment of the disease model.

Another compelling example is the control of CD20 expression in B lymphocytes. Here, CRISPR-Cas was used to insert a synthetic exon into the endogenous CD20 locus, with splicing of this exon controlled by a tetracycline-responsive riboswitch. The synthetic exon contained premature stop codons that prevented correct translation of CD20 if the synthetic exon was not spliced out from the mRNA (Fig. 1c). This enabled the investigation of the behavior of B-lymphocytes at different levels of CD20 expression and revealed for the first time its importance for membrane protein organization. Here, a short riboswitch-controlled synthetic exon was sufficient to establish a functional regulatory system without additional components or further alterations in the host genome³².

Synthetic riboswitches have also been successfully employed to dynamically regulate metabolic pathways in living cells. A lysine riboswitch from Escherichia coli was re-engineered from an OFF- to an ON-switch and applied to control the lysine transport in response to intracellular lysine levels, resulting in an significant increase in lysine yield in a recombinant Corynebacterium glutamicum strain³³ (Fig. 1d). Moreover, aptamer-based RNA sensors, similar in their function to synthetic riboswitches, enable real-time metabolic monitoring. By coupling ligand recognition to fluorescent light-up aptamers, these sensors enable direct and dynamic visualization of specific metabolites in a dose-dependent manner within living cells, thereby providing real-time insight into cellular metabolic states^34,35.

Great potential also lies in the application of riboswitches to wire and control artificial cells within cell-free systems. Their protein-independent and cis-acting mode of action combined with their small size makes them particularly efficient in the context of synthetic cell entities with limited energetic and metabolic resources²⁷. For instance, a histamine riboswitch was engineered to control molecular cargo release and “suicide” of the artificial cell in form of membrane disintegration³⁶ (Fig. 1e). Building on this riboswitch, a droplet-droplet communication system was established, which enabled receiver droplets to respond to histamine release of sender droplets, thus imitating bacterial quorum sensing³⁷. Several recent studies have further shown that synthetic riboswitches in cell-free systems offer a promising approach for achieving precise and scalable control over genetic circuits and biosensing applications^38–40.

Overall, we are confident that these selected highlights clearly demonstrate the potential of synthetic riboswitches as versatile and applicable tools. They have evolved beyond mere proof-of-concept studies to become effective instruments for precise gene regulation in cells, whole organisms, and even cell-free systems.

Current limitations and research challenges

Despite the obvious potential of synthetic riboswitches, their application has so far been mostly restricted to the field of synthetic RNA biology. Their niche existence may be explained with certain technical challenges and performance limitations.

One major bottleneck is the scarcity of well-characterized synthetic regulatory aptamer domains. The optimization required to obtain defined and functional aptamers from SELEX-enriched RNA pools is labor-intensive. In addition to binding affinity and specificity, regulatory aptamers must undergo substantial conformational rearrangements upon ligand binding - rearrangements that must pass through the entire riboswitch structure to modulate gene expression⁴¹. Such structural dynamics, as demonstrated for the tetracycline, guanine, and neomycin aptamers^42–44, are critical for in vivo functionality, yet they require extensive, exploratory work. As a result, the repertoire of reliable aptamers remains currently limited to a few candidates studied in great detail.

Moreover, all RNA-based regulators including riboswitches can exhibit context-dependent behavior that limits their straightforward portability across species and genetic systems. Flanking sequences may alter aptamer folding and ligand-binding dynamics, reducing predictability and performance⁴⁵. Although such effects are not universally observed, they require careful consideration in riboswitch design and application. Also, cellular contexts such as molecular crowding, differences in co-transcriptional folding, RNA-binding proteins, helicase activity or ionic conditions may influence riboswitch performance and limit cross-species portability^46–53. As a consequence, constructs that perform reliably in vitro may reveal unpredictable or diminished activity when expressed in a cellular environment due to interactions with cellular components or changed folding kinetics⁵⁴.

Rationally designed riboswitches sometimes deviate from their design expectations, displaying suboptimal expression levels or limited dynamic control. Current rational design and engineering approaches often involve the use of thermodynamic riboswitch calculators⁵⁵ or kinetic models⁵⁶ that depend heavily on accurate binding free-energy parameters for target aptamers. Although these methods can predict equilibrium secondary structures, they often ignore non-equilibrium folding pathways and intermediate conformations^57–59 that occur during expression-platform rearrangements in vitro and in vivo. Moreover, tertiary interactions influence functional activity^42,60 that remains difficult to model computationally with current tools⁶¹.

Ultimately, all of the above currently limits the rational design of synthetic riboswitches leading to laborious, experimental screening approaches for their development.

Emerging solutions

Recent advances in experimental methodologies simplify and accelerate synthetic riboswitch engineering. The combination of conventional SELEX with subsequent in vivo screening has already enabled the identification of novel riboswitches⁶². This post-SELEX screening step is particularly valuable as it allows for the high-throughput identification of aptamer candidates that undergo ligand-induced conformational changes - an essential feature for effective riboswitch function. A further big leap was achieved with the implementation of Capture-SELEX, which selectively enriches conformationally dynamic aptamers⁶³. Furthermore, microfluidic droplet-based sorting techniques^37,64, cellular selection and screening strategies⁶⁵, and the broad availability of next-generation sequencing (NGS) have significantly improved the throughput and functional outcome of riboswitch selection and optimization. These emerging technologies will expand the repertoire of well-characterized regulatory aptamers and will facilitate the engineering process.

Vast quantities of RNA-centric data are generated thus, paving the way for novel data-driven approaches. Over the last decade, deep learning (DL) has emerged as a powerful tool in biology, driving advances in structural biology⁶⁶, genomics⁶⁷, and transcriptomics⁶⁸. Unlike thermodynamic models, DL algorithms directly learn hierarchical feature representations from large-scale, high-dimensional datasets, achieving superior accuracy in regression and classification tasks for modeling complex biological behaviors⁶⁹. By uncovering latent patterns in these datasets, DL can circumvent the limitations imposed by cellular and genetic context and non-equilibrium folding states, and consequently establish the groundwork for data-driven, intelligent design of riboswitches from sequence to function.

Integration of DL has already been shown to enhance the prediction accuracy for riboswitch properties such as binding-sequence identification^70,71, classification⁷², and even binding affinity^73–75. The sequence-to-function prediction of riboswitches still relies heavily on large, high-quality functional datasets, a challenge given the tradeoff between assay throughput and accuracy. Compared to proteins, however, RNA offers certain advantages. It’s secondary structures are generally more predictable from sequence, although factors such as misfolding and non-equilibrium intermediates can still complicate design. Nevertheless, the relatively higher programmability of RNA sequences renders them more amenable to sequence-based approaches compared to proteins, as demonstrated in several successful riboregulator engineering efforts. Bridging the gap from predicting static sequence properties (e.g., motif discovery) to harnessing dynamic riboswitch behavior will likely require the integration of high-throughput experimental methods such as Flow-seq⁷⁶, which systematically map sequence–function relationships. These approaches are ideally suited to be incorporated into a DL-driven design of riboregulators, as reported for enhanced, programmable toehold switch regulators in E. coli, where they outperformed traditional thermodynamic models^77,78. The outcomes may be even further improved by creating iterative “Design–Build–Test–Learn” (DBTL) cycles^79,80. For example, a successful application of this paradigm was demonstrated on the Eterna platform, where large-scale, crowdsourced RNA design combined with high-throughput testing enabled the engineering of multiple-input RNA ratio sensors and logic gates. More recently, a DL-guided iterative approach has been employed to optimize performance of Boolean logic NAND gate behavior⁸¹.

Essential to progress toward the long-term goal of automated riboswitch design (Fig. 2) will be accurate tertiary structure prediction and dynamic simulations. Unlike proteins, which benefit from well-established structure prediction tools such as AlphaFold3⁸² and RoseTTAFold⁸³, RNA molecules still lack comparably robust models for tertiary structure prediction. This shortfall is largely attributable to the limited availability of experimentally determined RNA 3D structures and the intrinsic conformational flexibility of RNA. Nevertheless, recent DL-based methods such as RhoFold⁸⁴, trRosettaRNA⁸⁵, and DeepFoldRNA⁸⁶ have begun to close this gap by enabling increasingly accurate predictions of RNA 3D structures directly from sequence. Notably, generative design tools that explicitly integrate DL - such as RhoDesign⁸⁷ and gRNAde⁸⁸ - have demonstrated early but promising capabilities in generating RNA sequences conditioned on defined 3D structural geometries, thus marking an exciting step toward structure-based RNA design. Although the accurate design of dynamic, multi-state biomolecules remains an unsolved problem even in the more advanced field of protein engineering, achieving ligand-induced conformational switching in riboswitches represents a challenge of equal or greater complexity that will require continued innovation.

As research into synthetic riboswitches progresses and interest in and engagement with the RNA biology community grows, we anticipate that DL–driven DBTL cycles, coupled with more advanced SELEX approaches, will become indispensable tools for overcoming current challenges and advancing synthetic riboswitches toward modular ‘plug-and-play’ applications⁸⁹.

Our vision

In light of all the existing distinct advantages (modularity, protein-independency, low metabolic burden, portability) and the development of new strategies and solutions to overcome the current limitations of synthetic riboswitches (improved selection strategies and DL approaches), we are confident to see their broad application in fields as diverse as safeguarded RNA therapeutics, smart therapeutic bacteria, and self-defending plants, as well as shaping cellular behavior on the non-coding RNA level (Fig. 3). The inherent and unique properties of synthetic riboswitches allow them to regulate any RNA-based system, rendering them a powerful tool that may prove a game changer in many areas of research. Among these applications, RNA therapeutics are rapidly advancing toward clinical realization. Nevertheless, as this emerging field continues to expand, potential risks—such as off-target effects and immunogenicity—must be carefully addressed. The integration of aptazymes as safeguard mechanisms could provide an additional layer of control, minimizing risks while enhancing therapeutic precision.

Fig. 3. Potential uses of synthetic riboswitches in different application areas.

a Intestinal disease biomarkers are important to monitor gut health. The cartoon shows genetically engineered bacteria colonizing the intestinal lumen, equipped with synthetic riboswitches that detect specific biomarker molecules. Ligand binding to the riboswitch triggers a downstream actuator. The resulting output signal can be quantitatively measured in fecal samples, enabling noninvasive monitoring of gut health or disease states. b In agriculture, biological control of insect pests is important for environmental protection. Insect herbivores deliver effector proteins in their oral secretions that suppress plant immune signaling pathways. A synthetic riboswitch could be engineered to sense downstream signaling molecules, such as ATP or calcium, whose dynamics are altered in response to effector activity. Upon ligand binding, the riboswitch activates downstream defense genes, effectively bypassing effector-mediated inhibition and restoring plant immune responses against herbivore attack. c A synthetic lncRNA could be engineered with an aptazyme off-switch embedded. In the absence of ligands, the ribozyme remains inactive, and the lncRNA retains its poly-A tail, allowing normal performance. The aptazyme undergoes a conformational rearrangement upon ligand binding, activating its self-cleavage activity. The ribozyme then cleaves the poly-A tail, triggering rapid RNA degradation, which terminates the lncRNA’s regulatory roles. d An aptazyme Off-Switch can be engineered to enhance mRNA vaccine safety by enabling post-delivery control of antigen expression, which mitigates the risk of serious adverse events. 1: mRNA coding for desired antigens. 2: A small-molecule ligand is administered orally, enters the circulation, and penetrates the target cells. 3: Upon binding the aptazyme embedded in the mRNA’s 5′-UTR, the ligand induces a conformational change that activates the ribozyme’s self-cleavage activity, resulting in the degradation of the mRNA. 4: By selectively degrading the vaccine mRNA, the Off-Switch suppresses excessive antigen expression and mitigates the risk of anaphylactic reactions.

Taken together, we are convinced that the time has come for synthetic riboswitches to break out of the niche and unleash their full potential.

Author contributions

J.H. and T.W. wrote the manuscript, created figures and wrote the revised version of the manuscript. B.S. acquired funding, created a manuscript concept, wrote the manuscript and worked on the revision.

Peer review

Peer review information

Nature Communications thanks Chaitanya Joshi, Ashok Palaniappan and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Competing interests

The authors declare no competing interests.