Proof of Concept

The first step was to design the riboswitch sequences themselves. To design a riboswitch, we needed to find high-affinity apatamers that were proven to bind to specific target proteins or small molecules in literature. To show modularity of the system in the sense that various biomarkers could be tested for with the same reporter system, we chose six different biomarkers to test: thyroxine (Thyr), bovine thrombin (BThrom), basic fibroblast growth factor (bFGF), vascular endothelial growth factor (VEGF), human monomeric C-reactive protein (mCRP), and interleukin-32γ (IL32y).
Using the De Novo DNA Riboswitch Calulator, we produced hundreds of predicted riboswitch sequences. We selected five candidates for each biomarker that had high predicted translation rates but varying ON and OFF states so that each riboswitch was unique, improving our chances of developing a successful one.

The next step was to test if the riboswitches did what they were supposed to. Before attempting to produce the trehalase gene, we decided to test if the riboswitch could regulate fluorescent proteins first, as shown in previous literature.^[1][2] We characterize the riboswitches, we would use cell-free systems. Specifically, we used Daicel Arbor Bioscience’s myTXTL cell-free expression kit.
We were unsure how sensitive the riboswitches would be, so we wanted to test both the J23100 and T7 promoter. If the promoter was not strong enough, protein expression may not be noticeable. If the promoter was too strong, protein expression levels may be difficult to differentiate, and “leakiness” (protein expression without biomarker present) may become more of a problem.
In the end, all mRFP1 tests with both J23100 and T7 failed to show any increase in fluorescence in the presence of the biomarker. In fact, a the presence of the biomarker showed a represssing effect in some cases.

While we never made it to this stage of testing, the goal was to exchange the mRFP1 gene in each riboswich with the Tre37A gene, which codes for the trehalase enzyme. The Pardee Lab proved that toehold switches could be regulate the translation of the Tre37A gene from C. japonicus in cell-free systems^[3]. We would repeat this with our riboswitches using a “spacer,” which would theoretically allow us to switch out the downstream gene without changing the riboswitch sequence itself. The Riboswitch Calculator tailors the riboswitch sequences to the immediate downstream amino acids in the sequence, but if we add a short “spacer” sequence, we may be able to preserve the riboswitch’s functionality, even in the presence of a new gene.
We would quantify this by taking instantaneous glucose readings of our cell-free system over time using a standard glucometer. In the presence of the biomarker, the trehalase enzyme is produced proportionally. With an excess of trehalose added to the system, the trehalase will convert the trehalose into α-D-glucose and β-D-glucose. Measuring the increase of glucose over time would reveal if the riboswitch was successful in regulating translation of Tre37A.

In conclusion, we have failed to validate that the Riboswitch Calculator can predict translation-regulating riboswitch sequences. To continue, we must troubleshoot our testing protocols further. We must also focus on testing for fewer biomarkers at a time but test more predicted sequences of the remaining biomarkers, which would increase our chances of finding a successful sequence and allow us to better direct our efforts. We will better characterize our riboswitches with J23100 and mRFP1 with more testing and protocol modifications.