Model: RNA Switch Design Framework

A comprehensive approach to designing programmable RNA switches using Functional and Conformational RNA Elements

Our project revolves around a model that captures the structural behaviour of RNA and its role in gene regulation. The model describes how two sequences can fold into functional structures, and how these structures can be dynamically altered by external signals. By translating these structural changes into measurable effects on gene expression, the model provided us with a framework to design regulatory RNA switches, which we were able to build and test successfully in the lab.

In this page you will find first several sections explaining the general model, and then an explanation on how it guided our project with our particular system.

Before Skippit: The World of Transcription Regulation

Over 50 years ago, the discovery of three nuclear RNA polymerases in bacteriophages and bacteria allowed us to characterise and understand the intricacies of transcription.[1] Since then, synthetic biology has relied heavily on the use of transcriptional-level control to regulate gene expression and synthesise proteins in uncountable situations.

Transcriptional-level control of genes is a complex system, as it relies on a large amount of circumstances: a wide variety of DNA-binding factors such as transcription factors (which interact with each other and other proteins), the ability of RNA polymerases to recognise the promoter region, the RNA polymerases' own regulation strategies, the presence of nucleosomes (which inhibit initiation), presence of CpG islands in the promoter sequence, distance of the promoter to enhancer and silencer sequences, amongst others.[4, 10]

It is undeniable that despite its complexity and many playing factors, transcriptional-level control has evolved way beyond its original bounds and has established itself as the king of gene regulation. A clear example is the maximisation of transcription levels achieved in an organism by optimising promoter sequences with computational tools, or by engineering synthetic promoters and enhancers.[5, 6, 7, 8]

Limitations of Transcriptional Control

However, this strategy still has its shortcomings:

  • Cross-talk: Low-specificity transcription factors interfering with cis-regulatory elements[2]
  • Stochastic noise: Phenotypic variability in isogenic cell populations due to molecular binding randomness[3, 9]
  • Transcriptional bursting: Heterogeneity in mRNA and protein levels[12]
  • Expression noise: Alterations in cell fitness due to uncontrolled gene expression[11]

As synthetic biology evolves and positions itself as one of the key technologies for the future, with arising fields such as biomanufacturing and gene therapy, the need for fine-tuned, multi-layered protein expression control tools is increasing.

After Skippit: Beyond the Limits of DNA

Translational-level control provides rapid responses and direct protein regulation without altering transcription.[14, 15] Riboswitches—RNA sequences that change structure upon binding a molecule—are especially attractive: they are compact, programmable, and highly specific. The theophylline-binding aptamer, for example, has been extensively used to regulate downstream elements when linked to open reading frames.[15, 16]

Traditional Approach

Limited to basic systems like simple riboswitches and sRNAs targeting ribosome binding sites

TADPOLE Framework

Generalized framework for designing RNA switches from arbitrary parts, unlocking full potential of translational regulation

This approach of translational regulation is powerful, but has been limited to basic systems like simple riboswitches and sRNAs that target the ribosome binding site. To unlock the full potential of this regulatory layer, we need to move beyond these traditional methods. Our modeling approach, TADPOLE, addresses this limitation by offering a generalized framework for designing RNA switches from arbitrary parts.

Functional RNA Elements (FREs): Molecular Switches Controlled by Shape

A Functional RNA Element (FRE)—previously called Structural RNA Element (SRE)—is a segment of RNA that must fold into a very specific shape to perform its biological function.

Think of RNA as a flexible string that can fold into a precise shape, like a key that fits a lock. Its function depends entirely on maintaining this shape: if the fold is disrupted, the element stops working.

Why Shape Matters

The folded shape of a FRE is what allows it to interact with other molecules or cellular machinery in the correct way. By designing the sequence to favor the correct fold, we can control whether the element is active or inactive.

Examples of FREs

Stop Codon Readthrough Elements (SCRs)

These elements allow the ribosome to ignore a stop signal in the RNA sequence, extending the protein produced.[13, 24]

Example: The SECIS element, which regulates selenoprotein production. Its function depends on forming a very specific hairpin-like fold—if this fold is disrupted, the ribosome no longer reads through the stop codon.[18]

Frameshift Elements

These stimulate a shift in the ribosome reading frame during translation. The shift only happens if the RNA adopts the correct folded shape.

Conformational RNA Elements (CREs): Molecular Sensors

A Conformational RNA Element (CRE) is a segment of RNA that acts as a sensor. Its defining feature is that it changes shape when it detects a specific signal, such as a molecule, a temperature change, or another environmental factor. Crucially, the CRE's shape by itself does not carry out any function; it simply responds to the signal.

How CREs Work

Simple analogy: A CRE is like a Chunk of Chocolate
🍫

  • At room temperature, it is solid
  • When it warms up, it melts
  • This change in state doesn't do anything by itself; the chocolate doesn't trigger any action—it simply responds to the signal (heat)

This illustrates the essence of a Conformational RNA Element (CRE): it detects a signal and changes its shape, but by itself, that change has no direct function. It is purely a sensor.

Types of CREs

Ligand-Sensing CREs

Fold into a shape that can bind a specific molecule (ligand). Binding changes the shape of the RNA segment.

Examples: CREs that bind theophylline, tetracycline, warfarin, or ciprofloxacin.[17]

Environment-Sensing CREs

Respond to physical conditions, such as temperature or pH. Their shape shifts according to the environment.

RNA-RNA Interaction CREs

Interact with another RNA molecule, triggering a conformational change in response to this interaction.

Examples: miniRNAs

Value

The ON/OFF system described above is more than just a molecular switch; it represents a fundamental evolution in synthetic biology, overcoming key limitations of traditional genetic control methods and opening a completely new design space.

A Generalized, Fast, and Modular Design Framework

The real innovation here is generalization. Our approach provides a systematic framework for combining any Functional RNA Element (FRE) with any Conformational RNA Element (CRE), allowing the design of switches that respond to a wide variety of signals while controlling any RNA function dependent on its structure.

Key Advantages of This Approach

Translational-level Control

Operating downstream of transcription (RNA → protein) enables faster responses with lower energy costs compared to transcriptional regulation.

Unprecedented Modularity

Any FRE can be paired with any CRE. This systematic modularity allows design of highly customizable switches.

Versatility and Scalability

Simple detection systems can be expanded into multi-layer RNA circuits that amplify signals or trigger cascades of functions.

In essence, our framework transforms RNA switch design from isolated examples to a systematic, scalable toolkit. Synthetic biologists can now combine and adapt existing elements to create bespoke solutions, accelerating the design process and democratizing access to a powerful new layer of biological control.

Summary

Before vs After Skippit: A Paradigm Shift

Before Skippit

Transcriptional + Limited Translational

Control Level
Mainly DNA → RNA + basic RNA → protein (isolated examples)
Response Speed
Slow (requires transcription)
Energy Cost
High (full transcription machinery)
Modularity
Limited (complex dependencies)
Main Issues
Cross-talk, stochastic noise, transcriptional bursting
Design Approach
Simple riboswitches, basic sRNAs - isolated examples
→
TADPOLE Framework

After Skippit

Advanced Modular Translational Control

Control Level
Sophisticated RNA → Protein (any FRE + any CRE)
Response Speed
Fast (direct protein regulation)
Energy Cost
Low (no transcription needed)
Modularity
Unprecedented (any FRE + any CRE)
Main Advantages
Precise control, programmable switches, multi-layer circuits
Design Approach
Generalized framework: any FRE + any CRE combination

The Result: Democratized Access to Biological Control

This framework transforms RNA switch design from isolated examples to a systematic, scalable toolkit, enabling synthetic biologists to create bespoke solutions that respond to diverse signals while controlling any RNA function dependent on structure.

Combining FREs and CREs: Creating an ON/OFF Switch

Individually, FREs are functional elements that only work if they maintain a precise shape, while CREs are sensors that change shape in response to a signal but have no function by themselves. When we link a CRE to an FRE, we can create a programmable molecular switch that controls the FRE's activity.

How It Works

Our model is based on the principle of structural change from stimuli:


Initial structure  ➜  Application of external signal  ➜  Changed structure


One of these two structures (initial or changed) will maintain the function of the FRE ON (active), and the other will disrupt its structure, therefore turning it OFF (inactive). This ON structure is the functional structure, the structure that the FRE needs to have to perform its function:
Tadpole steps
Se we could have

FRE disrupted structure (OFF)  ➜  External signal  ➜  FRE functional structure (ON)

We call this the ON-ON alternative, as FRE is active when the external signal is present.
Tadpole steps
Or,

Structure disrupted (OFF)  ➜  External signal  ➜  Structure maintained (ON)

We call this the OFF-ON alternative, as the FRE is inactive when the external signal is present.
Tadpole steps
Both models follow the same energetic, complementarity, etc. principles. To facilitate the understanding of the design considerations of the model, we will be using an ON-ON/OFF-OFF type riboswitch as an example to explain the model’s intricacies. All following explanations can be applied to an ON-ON or an ON-OFF switch, the only difference being when the FRE is disrupted according to the external signal.

Considering the ON-ON case

OFF State (no signal)

The CRE is in its resting shape, which stabilizes the FRE in an inactive fold. In this state, the FRE cannot perform its function.

ON State (signal present)

When the CRE detects its signal, it changes shape. This conformational change propagates to the FRE through a connecting linker, allowing the FRE to adopt its active fold.

Simple Analogy: Chocolate Around a Key
🗝️
🍫

    FRE = key
    CRE = chunk of chocolate surrounding the key

    • No signal (chocolate is solid): The chocolate is hard and holds the key in the wrong shape, so it cannot enter the lock → FRE inactive (OFF)
    • Signal present (chocolate melts): The chocolate softens, releasing the key so it can take its proper shape → key can enter the lock → FRE active (ON)

    Here, the CRE itself doesn't do the work; it modulates whether the FRE can fold into its functional shape, creating a signal-dependent ON/OFF switch.

By combining CREs and FREs, we gain precise, modular control over RNA-based regulation. This allows us to design custom RNA switches that respond to a wide variety of signals while controlling any functional element, expanding beyond traditional riboswitches or sRNA-based methods.

Core Assumptions: Model Transparency

Our computational model (meaning the automatised design performed by the software tool) does not reproduce RNA behavior with 100% accuracy. To make it viable and scalable for large-scale design, we have had to make key assumptions, which we detail below.

Structural Prediction

RNA structure is a critical determinant of function, but for many RNA elements, the structure has not been experimentally resolved. In such cases, computational prediction tools provide a practical way to infer possible structures directly from the sequence[37,38].

How Structural Prediction Works

Prediction tools take as input the nucleotide sequence (e.g., AGUCGUAGUAAUUUCCGAUCG) and return a proposed structure in the so-called "dot-bracket notation":

  • . indicates an unpaired nucleotide
  • ( and ) indicate paired nucleotides

For example, RNAfold predicts the sequence AGUCGUAGUAAUUUCCGAUCG to fold into .((((.((....)).))))... This can be visualized as a two-dimensional secondary structure[37].

RNAfold

One of the most widely used tools for RNA secondary structure prediction is RNAfold, part of the ViennaRNA package. It uses a thermodynamics-based algorithm to calculate the minimum free energy (MFE) structure—the most stable conformation expected for the sequence. RNAfold outputs both the predicted structure and its associated free energy value, providing an estimate of stability[37,38].

Limitations of RNAfold and Structural Prediction

While RNAfold and similar tools are powerful, they come with important limitations[40]:

MFE Bias

RNA molecules do not always fold into the single most stable structure. In reality, they often exist as an ensemble of conformations, which RNAfold does not fully capture[40].

Context-Dependence

Predictions are made in isolation, without considering cellular factors such as RNA-binding proteins, molecular crowding, or co-transcriptional folding[40].

Limited Dynamics

RNA structures are dynamic and can shift in response to signals (as CREs do). Static predictions like those of RNAfold cannot easily account for these conformational changes[39].

Accuracy Trade-offs

Thermodynamic parameters are derived from simplified experimental systems and may not always generalize to longer or more complex RNAs[38].

Energy Assumptions for OFF ⇄ ON Switching

Notation and Basic Idea

We work with predicted minimum free energies (MFEs) from a secondary-structure predictor (e.g., RNAfold). Define:

  • GOFF = MFE of the full construct predicted without the external factor (unconstrained fold). This is the free energy of the unbound / OFF configuration.
  • GON = MFE of the full construct predicted with the CRE forced into the conformation that corresponds to having reacted to the external factor. This is the free energy of the would-be ON conformation before accounting for the extra stabilisation supplied by binding.
  • Ebind = stabilisation energy (positive number, in kcal·mol⁻¹) supplied to the system when the CRE actually binds its partner (ligand, miRNA, etc.). We treat Ebind as a positive magnitude (e.g., theophylline ≈ 9.5 kcal·mol⁻¹ in literature[27,28]). If there is no binding (environment-driven CRE), then Ebind = 0.

Define the on–off energy gap (difference) as:

Δ = GON - GOFF

Note: G values are negative (more negative = more stable). With this definition:

  • Δ > 0 means OFF is more stable than ON (because GOFF < GON)
  • Δ < 0 would mean ON is already preferred in absence of trigger (bad for an OFF default)

When the CRE actually binds, the ON state gains the stabilisation Ebind. The effective free energy of ON after binding becomes:

GON,eff = GON - Ebind

Thermodynamic Conditions You Must Check

1. OFF Must Be the Default (No Trigger)

Condition:

Δ = GON - GOFF > 0

Why: In the absence of the external factor we want the system to sit in the OFF state (FRE disrupted). If Δ ≤ 0, the system would be ON by default or ambivalent (unstable OFF).

2. ON Must Become Preferred After Binding (Binding-based CREs)

Condition (after binding):

GON,eff < GOFF ⟺ Δ < Ebind

Combined requirement (binding-based CRE):

0 < Δ < Ebind

Why: This ensures the OFF is preferred before binding, and the ON becomes preferred after the CRE supplies the binding stabilisation. If Δ ≥ Ebind, even binding will not make ON lower energy than OFF (i.e., the switch will be stuck OFF). If Δ ≤ 0, the system is not OFF by default.

Practical Recommendation

Choose Δ roughly half of Ebind (i.e. Δ ≈ Ebind/2). This yields a robust margin in both directions: OFF is meaningfully favoured without ligand, but binding reliably flips the balance.

Example: theophylline Ebind ≈ 9.5 kcal → target Δ ≈ 4.5 kcal[27,28].

3. Environment-Driven CREs (Temperature, pH, etc.)

If the CRE does not supply binding energy but instead changes conformation because of external physical conditions (temperature shift, pH), the logic is different:

  • Set Ebind = 0. The environment itself alters base-pair stabilities and changes the predicted energies.
  • To test viability, predict MFEs at both conditions (e.g., RNAfold supports a temperature parameter):
    • GOFF = MFE predicted under baseline conditions
    • GON = MFE predicted under the changed environmental condition
  • Then require GON < GOFF under the activating environment, and GOFF < GON under baseline.

Why "About Half the Binding Energy"?

The inequality 0 < Δ < Ebind is the correct thermodynamic requirement. Choosing Δ near Ebind/2 gives a comfortable margin so:

  • OFF is clearly favoured in the absence of ligand (so the system is quiet / repressed), and
  • ON becomes clearly favoured after adding the ligand (so the switch flips decisively).

If Δ is too small (close to 0) the system will be noisy and leak (both states may be populated). If Δ is too large (approaching Ebind or exceeding it) the ligand cannot flip the switch.

Practical Numeric Guidance and Examples

Example (Theophylline-like)

  • Literature Ebind ≈ 9.5 kcal·mol⁻¹ [27,28]
  • Target Δ ≈ Ebind/2 ≈ 4.7 kcal·mol⁻¹

Suppose your predicted MFEs are:

  • GOFF = -44.0 kcal·mol⁻¹; GON = -39.3 kcal·mol⁻¹
  • Then Δ = -39.3 - (-44.0) = +4.7 kcal·mol⁻¹ → meets the target
  • After binding: GON,eff = GON - Ebind = -39.3 - 9.5 = -48.8, which is lower than -44.0 → ON will be favoured after ligand binding

Example (miniRNA Trigger, Small Binding)

If Ebind is small (e.g., 1–2 kcal·mol⁻¹), then you need a correspondingly small Δ (≈0.5–1 kcal) — smaller margins reduce robustness and increase sensitivity to modelling error; be cautious and pair with strong structural checks.

If Ebind is Unknown

Use a conservative Minimum ΔMFE between 3–5 kcal·mol⁻¹ as a first pass unless you have stronger estimates. Also perform sensitivity scans across possible Ebind values and check how many candidate linkers survive.

Experimental Validation

We are very pleased to say that of the 4 different systems we tried in the lab, all of them successfully modulated the FRE element in the presence and absence of the external signal (ligand in our case) as an OFF-ON switch. Please, see the How the Model Guided Our Progect Page for a description of our particular designs with the concrete elements used.
We have been able to convert our switches from an in silico design to reality. For further details, you can check the Results or Parts page
About how the results affect the model: we obtained 4 functional switches for each different proposal tested in the lab. We saw how the addition of theophylline provoked some changes in the system that led to a reduction in the level of readthrough.
However, let’s go beyond, analysing the expectations we had on our model:
The systems designed with aptamers are usually built by testing hundreds or even thousands of possible sequences, until one is found that performs the expected behaviour.
However, we could not afford that method so if we wanted to design some switches we needed something to narrow it down. So we built Tadpole, a tool that automatises the design of the desired switches. The aim was to be able to build two structures, that had

  • Similar energies, so none would be much more stable than the other, and the switch would be possible.
  • Enough complementarity (aided by the linker) in order to make the pairing of the CRE to the FRE possible, so it would disrupt it.
The software also predicts the structures, to try to have an approximation to what the final design is going to look like, and if it’s going to be an ON-ON or an OFF-ON. However, as was acknowledged on the Core Assumptions section, RNAFold has limitations on structure prediction, so it’s not 100% reliable. In fact, the structures predicted for our systems indicated an ON-ON system, but the lab results demonstrate it is an OFF-ON.
With this, we can affirm that our model can be used to build RNAswitches, and that we have a software tool able to automate it, with which anyone can make a semi-rational design, so they can create a handful of sequences to test in the lab, instead of what would be the usual when working with aptamers: testing hundreds and hundreds.
We didn’t just make our project possible with this model and our software tool. We created a framework that empowers other teams to design and test these systems more easily and efficiently.

References

  1. Five decades of eukaryotic transcription. Nat Struct Mol Biol 26, 757 (2019). https://doi.org/10.1038/s41594-019-0303-1
  2. Friedlander, T., Prizak, R., Guet, C. et al. (2016). Intrinsic limits to gene regulation by global crosstalk. Nat Commun 7, 12307. https://doi.org/10.1038/ncomms12307
  3. Barroso, G. V., Puzovic, N., & Dutheil, J. Y. (2018). The Evolution of Gene-Specific Transcriptional Noise Is Driven by Selection at the Pathway Level. Genetics, 208(1), 173–189. https://doi.org/10.1534/genetics.117.300467
  4. Cramer, P. (2019). Organization and regulation of gene transcription. Nature 573, 45–54. https://doi.org/10.1038/s41586-019-1517-4
  5. Blazeck, J., & Alper, H. S. (2012). Promoter engineering: Recent advances in controlling transcription at the most fundamental level. Biotechnology Journal. https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/abs/10.1002/biot.201200120
  6. Khan, A., Nasim, N., Pudhuvai, B., Koul, B., Upadhyay, S. K., Sethi, L., & Dey, N. (2023). Plant Synthetic Promoters: Advancement and Prospective. Agriculture, 13(2), 298. https://doi.org/10.3390/agriculture13020298
  7. Xu, N., Wei, L., & Liu, J. (2019). Recent advances in the applications of promoter engineering for the optimization of metabolite biosynthesis. World J Microbiol Biotechnol 35, 33. https://doi.org/10.1007/s11274-019-2606-0
  8. De Winter, S., Konstantakos, V., & Aerts, S. (2025). Modelling and design of transcriptional enhancers. Nat Rev Bioeng 3, 374–389. https://doi.org/10.1038/s44222-025-00280-y
  9. Sánchez, Á., & Kondev, J. (2008). Transcriptional control of noise in gene expression. Proc. Natl. Acad. Sci. U.S.A. 105(13), 5081–5086. https://doi.org/10.1073/pnas.0707904105
  10. Splinter, E., & de Laat, W. (2011). The complex transcription regulatory landscape of our genome: control in three dimensions. The EMBO Journal 30, 4345–4355. https://doi.org/10.1038/emboj.2011.344
  11. Duveau, F., Hodgins-Davis, A., Metzger, B. P. H., et al. (2018). Fitness effects of altering gene expression noise in Saccharomyces cerevisiae. eLife 7:e37272. https://doi.org/10.7554/eLife.37272
  12. Zhang, Q., et al. (2024). Transcriptional bursting dynamics in gene expression. Frontiers in Genetics. https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2024.1451461/full
  13. Davies, K., & Nowak, K. (2006). Molecular mechanisms of muscular dystrophies: old and new players. Nat Rev Mol Cell Biol 7, 762–773. https://doi.org/10.1038/nrm2024
  14. Ogawa, A. (2011). Rational design of artificial riboswitches based on ligand-dependent modulation of internal ribosome entry. RNA, 17(3), 478–488. https://doi.org/10.1261/rna.2433111
  15. Werstuck, G., & Green, M. R. (1998). Controlling gene expression in living cells through small molecule-RNA interactions. Science, 282(5387), 296–298. https://doi.org/10.1126/science.282.5387.296
  16. Shanidze, N., et al. (2020). A Theophylline-Responsive Riboswitch Regulates Expression of Nuclear-Encoded Genes. Plant Physiol, 182(1), 123–135. https://doi.org/10.1104/pp.19.00625
  17. Bayer, T. S., & Smolke, C. D. (2005). Programmable ligand-controlled riboregulators of eukaryotic gene expression. Nat Biotechnol, 23(3), 337–343. https://www.nature.com/articles/nbt1060
  18. Mariotti, M., et al. (2016). Lokiarchaeota marks the transition between the archaeal and eukaryotic selenocysteine encoding systems. Molecular Biology and Evolution, 33(9), 2441–2453. https://academic.oup.com/mbe/article/33/9/2441/2925203
  19. Chen, Y., Ho, J. M. L., Shis, D. L., et al. (2018). Tuning the dynamic range of bacterial promoters regulated by ligand-inducible transcription factors. Nature Communications 9(1), 64. https://www.nature.com/articles/s41467-017-02473-5
  20. Haugen, S. P., Ross, W., & Gourse, R. L. (2008). Advances in bacterial promoter recognition and its control by factors that do not bind DNA. Nat Rev Microbiol 6(7), 507–519. https://doi.org/10.1038/nrmicro1912
  21. Jiang, C., & Pugh, B. F. (2009). Nucleosome positioning and gene regulation: advances through genomics. Nat Rev Genet 10(3), 161–172. https://doi.org/10.1038/nrg2522
  22. Jang, H. S., Shin, W. J., Lee, J. E., & Do, J. T. (2017). CpG and Non-CpG Methylation in Epigenetic Gene Regulation and Brain Function. Genes (Basel) 8(6), 148. https://doi.org/10.3390/genes8060148
  23. Segert, J. A., Gisselbrecht, S. S., & Bulyk, M. L. (2021). Transcriptional Silencers: Driving Gene Expression with the Brakes On. Trends Genet 37(6), 514–527. https://doi.org/10.1016/j.tig.2021.02.002
  24. Rodnina, M. V., et al. (2019). Translational recoding: Canonical translation mechanisms reinterpreted. Nucleic Acids Research 48(3), 1056–1067. https://doi.org/10.1093/nar/gkz783
  25. Ge, H., & Marchisio, M. A. (2021). Aptamers, riboswitches, and ribozymes in S. cerevisiae synthetic biology. Life 11(3), 248. https://doi.org/10.3390/life11030248
  26. Lynch, S. A., Desai, S. K., Sajja, H. K., & Gallivan, J. P. (2007). A high-throughput screen for synthetic riboswitches. Chemistry & Biology 14(2), 173–184. https://doi.org/10.1016/j.chembiol.2006.12.008
  27. Akhter, S., Tang, Z., Wang, J., et al. (2024). Mechanism of ligand binding to theophylline RNA aptamer. J Chem Inf Model 64(3), 1017–1029. https://doi.org/10.1021/acs.jcim.3c01454
  28. Wrist, A., Sun, W., & Summers, R. M. (2020). The theophylline aptamer: 25 years as an important tool in cellular engineering research. ACS Synthetic Biology 9(4), 682–697. https://doi.org/10.1021/acssynbio.9b00475
  29. Suess, B., Fink, B., Berens, C., Stentz, R., & Hillen, W. (2004). A theophylline responsive riboswitch controls gene expression in vivo. Nucleic Acids Research 32(4), 1610–1614. https://doi.org/10.1093/nar/gkh321
  30. Warfield, B. M., & Anderson, P. C. (2017). Molecular simulations and Markov state modeling reveal the structural diversity and dynamics of a theophylline-binding RNA aptamer. PLoS ONE 12(4), e0176229. https://doi.org/10.1371/journal.pone.0176229
  31. Zimmermann, G. R., Jenison, R. D., Wick, C. L., Simorre, J.-P., & Pardi, A. (1997). Interlocking structural motifs mediate molecular discrimination by a theophylline-binding RNA. Nature Structural Biology 4(9), 644–649. https://doi.org/10.1038/nsb0897-644
  32. Hausser, J., Mayo, A., Keren, L., & Alon, U. (2019). Central dogma rates and the tradeoff between precision and economy in gene expression. Nature Communications 10(1), 68. https://doi.org/10.1038/s41467-018-07391-8
  33. Mishler, D. M., & Gallivan, J. P. (2014). A family of synthetic riboswitches adopts a kinetic trapping mechanism. Nucleic Acids Research 42(10), 6753–6761. https://doi.org/10.1093/nar/gku262
  34. Gebauer, F., & Hentze, M. W. (2004). Molecular mechanisms of translational control. Nature Reviews Molecular Cell Biology 5(10), 827–835. https://doi.org/10.1038/nrm1488
  35. Touat-Hamici, Z., Bulteau, A.-L., Bianga, J., Lobanov, A. V., & Gladyshev, V. N. (2018). Selenium-regulated hierarchy of human selenoproteome. BBA - General Subjects 1862(11), 2493–2505. https://doi.org/10.1016/j.bbagen.2018.04.012
  36. Liang, J. C., Chang, A. L., Kennedy, A. B., & Smolke, C. D. (2012). A high-throughput, quantitative cell-based screen for efficient tailoring of RNA device activity. Nucleic Acids Research 40(20), e154. https://doi.org/10.1093/nar/gks636
  37. Lorenz, R., Bernhart, S. H., Höner zu Siederdissen, C., Tafer, H., Flamm, C., Stadler, P. F., & Hofacker, I. L. (2011). ViennaRNA Package 2.0. Algorithms for Molecular Biology, 6, 26. https://doi.org/10.1186/1748-7188-6-26
  38. Turner, D. H., & Mathews, D. H. (2009). NNDB: The nearest-neighbor parameter database for predicting stability of nucleic acid secondary structure. Nucleic Acids Research, 38(Database issue), D280–D282. https://doi.org/10.1093/nar/gkp892
  39. Jenison, R. D., Gill, S. C., Pardi, A., & Polisky, B. (1994). High-resolution molecular discrimination by RNA. Science, 263(5152), 1425–1429. https://doi.org/10.1126/science.7510417
  40. Mathews, D. H., Moss, W. N., & Turner, D. H. (2006). RNA secondary structure prediction. Current Opinion in Structural Biology, 16(3), 270–278. https://doi.org/10.1016/j.sbi.2006.04.008
↑