Molecular computing with plant cell phenotype serving as quality controlled output

Abstract

The ability of autonomous biomolecular computing devices to interact directly with biological systems and even with living organisms without any interface represents their main advantage over the electronic computers. This study shows that the expression of fluorescent proteins in live plant cells can be utilized as a highly accurate visual output of DNA-based computing. Each of the two possible outputs of a 2-symbol 2-state finite automaton was represented here by either green or cyan fluorescence in eukaryotic cells. The automata were programmed by the choice of several molecules from a library of 8 transition molecules, each containing a recognition site for a type II endonuclease. Two enzymes, endonuclease and a DNA ligase, as well as ATP, represented the hardware. Each input molecule, in the form of a dsDNA, included a string of symbols, 6 bp each, and a 6 bp terminator. The two detection molecules were also dsDNA, each containing a 4-base sticky end, complementary to the appropriately restricted terminator and a gene encoding for a different fluorescent protein. Computation was carried out by mixing all components in a homogeneous solution, leading to autonomous processing of the input molecule via repetitive cycles of digestion, hybridization, and ligation. The output processing procedure involved the creation of a circular dsDNA that contained the gene of either green fluorescent protein or cyan fluorescent protein. Insertion of these plasmids into onion cells by particle bombardment resulted in either green fluorescent or cyan fluorescent live cells as phenotypical output signals. The plasmid formation was an important step because it served as a quality control gate that transformed a rather noisy output into a clean signal. This process of noise elimination allowed for clean and flawless outputs with high fidelity and zero noise.