Engineering protein sensors and actuators at multiple scales for neuroscience

Abstract

The mammalian brain supports sophisticated mental functions such as cognition and memory through intricate networks of interconnected neurons. Canonical transmission of information starts with the initiation and propagation of voltage waveforms throughout a neuron’s surface, and a substantial amount of neural communication between neurons relies on chemical transmission over a range of scales. To dissect these sophisticated and varied neural processes, researchers require methods that enable precise, non-invasive activation and monitoring of neural activity with genetic specificity in vivo. Recent advances in optogenetic actuators and genetically encoded indicators, such as voltage indicators (GEVIs) and fluorescent indicators of brain chemistry, have partially met this need. The neuroscience community now aspires toward multi-channel experiments that combine optogenetic stimulation with real-time imaging of neural activity. Such multi-channel experiments, when combined with complementary methodologies in system neuroscience, are crucial for probing interactions between different populations of neurons. However, these experiments often struggle with non-negligible photocurrent and optical crosstalk, which can cause unwanted neuron depolarization or crosstalk fluorescence transients. This has been a barrier to achieving spatiotemporally precise neuron activation and dual-channel experiments essential for detailed neural circuit dissection. To overcome these challenges, ongoing developments are focused on shifting optogenetic actuators towards blue wavelengths and indicators towards red or near-infrared wavelengths. Despite these efforts, the kinetics of blue-shifted optogenetic actuators is still suboptimal, and the range of genetically encoded indicators is still predominantly centered around GFP-based indicators, with their RFP-based counterparts falling behind in brightness and diversity. In this dissertation, we used both low- and high-throughput screening approaches to optimize the kinetics of a blue-shifted optogenetic actuator, and to optimize the brightness of red fluorescent proteins in the mScarlet family, respectively. For the blue-shifted actuator, we applied low-throughput techniques due to the well-established structural and functional relationships of homologous proteins. We focused on targeting critical residues for single or combinatorial amino acid substitutions, enhancing a CoChR variant for improved kinetics by screening a select number of candidates. We adopted a high-throughput screening strategy for the mScarlet family, driven by their synthetic nature and the absence of homology with other fluorescent proteins. We developed a streamlined high-throughput system that integrated pooled functional screening, optical in situ sequencing, and automated image and data processing. This streamlined and high-throughput approach allowed us to successfully engineer RFPs in mScarlet family through multiple rounds of large-scale screenings Furthermore, this screening strategy can be adapted to engineer other genetically encoded tools with specific experimental modifications, showcasing its broad applicability in the field of protein engineering.Reflecting on the current trends in neuroscience where neural activation and in vivo imaging are frequently performed simultaneously, our results align well with the direction of development in both neuroscience and protein engineering. Such alignment underscores the contribution of our work to these fields.