Surface-immobilized DNA nanomachines

We are working towards the construction of hybrid bioelectronic systems in which DNA nanomachines are integrated with semiconductor circuitry, where DNA performs the function of highly parallel, low-power processing of biomolecular signals, and silicon is used for high-speed data processing and control. This technology has applications in computation and smart biosensing, and surface immobilization of DNA is key to the endeavour, because it enables biological and electronic components to be interfaced [1]. Procedures for immobilizing DNA are very reliable, and we can routinely prepare DNA monolayers on gold with densities of ~10 molecules/cm. Many solution-phase DNA machines are driven by toehold-mediated strand displacement, but the effect of surface-immobilization on this process has not yet been fully established. To address this, we studied strand displacement in a representative surface-immobilized DNA nanomachine, using a quartz crystal microbalance with dissipation monitoring (QCM-D) [2]. This technique involves the use of acoustic waves to probe a surface-immobilized molecular layer, providing realtime information on its mass and structure. We found that surface-immobilization has a significant effect on strand displacement. Our approach allowed us to probe toehold binding and branch migration separately, which is difficult with other techniques, and we established that the rates of these processes could be tuned independently. Consequently, we showed that the overall displacement rate on the surface could be controlled over nearly two orders of magnitude through changes to the concentration or length of the invading strand, or insertion of mismatched bases. We have also constructed and tested a surface-immobilized DNA OR gate based on strand displacement [3]. The ensemble of gates took ~6 minutes to switch, and when the input strand needed to bind very close to the surface the switching efficiency was significantly reduced on the second and subsequent cycles, a finding which may have significant implications for the design of future systems. We estimated the power dissipated during switching to be 0.1nW cm, which compares very favourably with equivalent values for an array of silicon transistors. This provides quantitative evidence for the claim that biomolecular computation is a low-power technology. We have begun to design architectures for hybrid systems and in the future we will start to implement these in the laboratory. One of our objectives is to incorporate electronically-switchable DNA machines [4].