Neuropeptide oxytocin is central to maternal and reproductive behaviors, anxiolysis, and social interaction, and is thus critical to our social well-being. Our understanding of oxytocin’s function in the brain remains incomplete, however, as oxytocin signaling pathways are uncharacterized. There is a demonstrable need for an imaging probe with the specificity and spatiotemporal resolution requisite for endogenous oxytocin signaling (seconds, microns). Leveraging the optical properties of single-walled carbon nanotubes (SWNT) and molecular recognition capacity of DNA, our lab has previously developed sensors for imaging dopamine in the brain. The traditional method of development for DNA-SWNT sensors requires a low-throughput guess-and-test approach to identify sequences selective towards the target analyte. Consequently, more structurally complex targets, such as oxytocin, have eluded this development strategy. To combat this challenge, we have developed an evolution-based platform termed SELEC. This platform, which is generically applicable to any neurotarget, enables co-evolution of the molecular recognition unit (DNA) together with the signal transducer (SWNT). Herein we present the evolution of synthetic molecular recognition for oxytocin with SELEC. With an initial library of &gt;10<sup>10</sup> random DNA sequences, multiple iterations of competitive DNA-SWNT binding were conducted, followed by desorption, recovery, and enzymatic amplification of adsorbed DNA ligands. After six rounds of SELEC, deep sequencing analysis identified the most enriched DNA ligands with the highest affinity for oxytocin when bound to the SWNT surface. These DNA-SWNT constructs were then screened for a turn-on fluorescence response to oxytocin, which revealed an enrichment of responsive sensors as a function of SELEC round with eight sensor candidates from SELEC rounds 5 and 6 demonstrating a dF/F<sub>o</sub> of &gt;50% in response to 100 µM oxytocin. Top sensor candidates are screened for sensitivity, selectivity, and reversibility, towards their use to image endogenous oxytocin signaling. To further improve the selection of responsive sensor candidates, we are also developing and employing deep learning models to learn the correlation between DNA sequence and fluorescence response to oxytocin. Our results herein suggest SELEC can be generically leveraged to evolve molecular recognition and develop imaging probes for elusive targets such as neuropeptide oxytocin.

High-throughput evolution of near-infrared oxytocin nanosensors

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.