(Invited) Predicting the Response of Carbon Nanotube Field-Effect Biosensors Using Simulations

Objectives Carbon nanotube field-effect biosensors (bioCNTFETs) are a promising platform to detect biomolecular analytes because they are highly sensitive, specific, low-cost, portable, and integrable into arrays for multiplexed sensing [1]. Their sensitivity originates from carbon nanotubes having electrical properties that are strongly dependent on nearby biomolecules. As a result, the biosensor detects an electrical current through the network of nanotubes that depends on the concentration of analyte. To achieve specificity, carbon nanotubes need to be functionalized with a probe molecule that binds to a specific target analyte. However, it has been demonstrated that the response of bioCNTFETs strongly depends on the attachment site on the probe molecule to the carbon nanotube [2], and this variability is impairing the development of bioCNTFETs. To address this challenge, we are developing an integrated approach combining simulation and experiment: simulations are used to predict which attachment sites on the binding protein are expected to generate the greatest response, then devices are fabricated using phenyl azide photochemistry to precisely attach the binding protein via these sites on the nanotubes. To validate our approach, we characterized a bioCNTFET covalently functionalized with the beta-lactamase inhibitor protein II (BLIP-II) to detect beta-lactamases such as TEM-1 and KPC-2 that are associated with antimicrobial resistance. Results Our simulation protocol is based on molecular dynamics to sample the biomolecule-nanotube interactions and then on Poisson-Boltzmann calculations to estimate the electrostatic potential generated on the nanotube’s surface by the conformations of the biomolecule [3]. Since electrostatic potential changes on the surface of the nanotubes in the BLIP-II biosensor are expected to cause the electrical signal changes detected, we specifically quantify that as TEM-1 or KPC-2 binds to the covalently attached BLIP-II to the nanotube. We looked at two different attachment sites on BLIP-II – A41 and T213 – that are expected to produce a different response. For the A41 site, we show that the electrostatic potential doesn’t change significantly as TEM-1 or KPC-2 binds to BLIP-II because they are too far from the nanotube. However, the T213 site allows significant electrostatic potential changes: TEM-1 shifts the potential to more negative values, while KPC-2 shifts the potential to less negative values. The predictions from our simulations agree with measurements done on devices. For the A41 site, the measured conductance doesn’t change significantly upon binding of TEM-1 or KPC-2, indicating that the electrostatic potential on the nanotube doesn’t change much, as predicted. For the T213 site, the measured conductance increases significantly upon binding of TEM-1, while it decreases significantly for KPC-2. This suggests more p-type carriers in the nanotube (due to a more negative potential on the nanotube) as TEM-1 binds, but less p-type carriers (due to a less negative potential) as KPC-2 binds, as predicted. Conclusions We have developed a simulation protocol that predicts, in agreement with experimental measurements, the impact of the probe’s attachment site on the conductance modulation detected by carbon-nanotube field-effect biosensors. This supports the development of a truly integrated approach combining simulation and experiment to consistently design and fabricate these biosensors to achieve optimized sensitivity. References [1] Lee, C. S. et al. Chembiochem 2022 , e202200282. [2] Xu, X. et al. Angew. Chem. Int. Ed. 2021 , 60 , 20184-20189. [3] Côté, S. et al. Phys. Chem. Chem. Phys. 2022 , 24 , 4174-4186. Figure 1