Modular composition frameworks support the rapid assembly of biological components into devices that can be implemented toward the most pressing research challenges facing society. Implicit in these frameworks is the ability to modify individual component behavior to predictably tune and program device performance. However, design principles are needed to guide component modification and accurately predict their effect on device performance. We were interested in delineating design principles for the rapid tuning and programming of RNA-based devices. RNA is a versatile building block for device construction based on a variety of inherent properties: a manageable size of sequence space, predictable folding and sequence recognition through base pairing interactions, diverse regulatory mechanisms, catalytic activity, and specific and tight binding of various biomolecules. By utilizing these properties, researchers have successfully constructed RNA-based information processing devices called riboswitches that program cellular behavior through detection of user-defined signals in the intracellular environment. As the identity of the inputs and outputs are only loosely constrained, synthetic riboswitches have tremendous potential as in vivo biosensors and autonomous control systems. Composition frameworks for RNA device assembly from modular functional components are currently being developed, where recent studies have offered a few design principles. A critical factor that remains to be evaluated is the role of kinetics and thermodynamics in riboswitch performance. Inherent processes such as ligand binding or transcription rate have a profound impact as shown experimentally, although kinetic factors are not currently considered as part of device construction. To address this issue, we investigated riboswitch behavior in silico to develop a quantitative link between the dynamics of riboswitch behavior and overall performance toward gaining further insights into the design and programming of riboswitch function.

To draw general conclusions regarding riboswitch function, we considered three representative regulatory mechanisms: transcriptional termination, translational repression, and mRNA destabilization. Modeling results suggest that riboswitch performance is highly dependent on the competition between irreversible rates such as mRNA degradation or transcriptional termination and reversible rates such as conformational switching or ligand binding. When reversible rates dominate (thermodynamically-driven), riboswitch performance is independent of the selected regulatory mechanism. When the rates are balanced (kinetically-driven), riboswitch performance generally decreases, transcriptional folding becomes important, and performance depends on the selected regulatory mechanism. When irreversible rates dominate (non-functional), riboswitch performance is insensitive to the input signal. In addition, imposing an upper limit on the input signal reduces the maximal dynamic range and establishes a bounded parameter space for optimal riboswitch performance. Model predictions are supported by published experimental data and physical modification of a synthetic riboswitch.

From our modeling efforts, we arrived at multiple design principles to guide the construction and modification of synthetic riboswitches. These principles address selection and modification of device components as well as formulation of composition frameworks. Current work is focused on applying these principles to improving the performance of a synthetic riboswitch actively used in our group for biomedical and biotechnological applications.