Claire N. Bedbrook

Archaerhodopsin variants with enhanced voltage-sensitive fluorescence in mammalian and Caenorhabditis elegans neurons

Probing the neural circuit dynamics underlying behavior would benefit greatly from improved genetically encoded voltage indicators. The proton pump Archaerhodopsin-3 (Arch), an optogenetic tool commonly used for neuronal inhibition, has been shown to emit voltage sensitive fluorescence. Here we report two Arch variants that in response to 655 nm light have 3–5 times increased fluorescence and 55–99 times reduced photocurrents compared to Arch WT. The most fluorescent variant, Archer1, has 25–40% fluorescence change in response to action potentials while using 9 times lower light intensity compared to other Arch-based voltage sensors. Archer1 is capable of wavelength specific functionality as a voltage sensor under red-light and as an inhibitory actuator under green-light. As a proof-of-concept for the application of Arch-based sensors in vivo, we show fluorescence voltage sensing in behaving C. elegans. Archer1’s characteristics contribute to the goal of all-optical detection and modulation of activity in neuronal networks in vivo.

Recent advances in engineering microbial rhodopsins for optogenetics

Protein engineering of microbial rhodopsins has been successful in generating variants with improved properties for applications in optogenetics. Members of this membrane protein family can act as both actuators and sensors of neuronal activity. Chimeragenesis, structure-guided mutagenesis, and directed evolution have proven effective strategies for tuning absorption wavelength, altering ion specificity and increasing fluorescence. These approaches facilitate the development of useful optogenetic tools and, in some cases, have yielded insights into rhodopsin structure-function relationships.

Structure-guided SCHEMA recombination generates diverse chimeric channelrhodopsins

Significance Critical for regulating cell function, integral membrane proteins (MPs) are key engineering targets. MP engineering is limited because these proteins are difficult to express with proper plasma membrane localization in heterologous systems. We investigate the expression, localization, and light-induced behavior of the light-gated MP channel, channelrhodopsin (ChR), because of its utility in studying neuronal circuitry. We used structure-guided SCHEMA recombination to generate libraries of chimeric ChRs that are diverse in sequence yet still capable of efficient expression, localization, and useful light-induced functionality. The conservative nature of recombination generates unique protein sequences that tend to fold and function. Recombination is also innovative: chimeric ChRs can outperform their parents or even exhibit properties not known in natural ChRs. Integral membrane proteins (MPs) are key engineering targets due to their critical roles in regulating cell function. In engineering MPs, it can be extremely challenging to retain membrane localization capability while changing other desired properties. We have used structure-guided SCHEMA recombination to create a large set of functionally diverse chimeras from three sequence-diverse channelrhodopsins (ChRs). We chose 218 ChR chimeras from two SCHEMA libraries and assayed them for expression and plasma membrane localization in human embryonic kidney cells. The majority of the chimeras express, with 89% of the tested chimeras outperforming the lowest-expressing parent; 12% of the tested chimeras express at even higher levels than any of the parents. A significant fraction (23%) also localize to the membrane better than the lowest-performing parent ChR. Most (93%) of these well-localizing chimeras are also functional light-gated channels. Many chimeras have stronger light-activated inward currents than the three parents, and some have unique off-kinetics and spectral properties relative to the parents. An effective method for generating protein sequence and functional diversity, SCHEMA recombination can be used to gain insights into sequence–function relationships in MPs.

Machine learning to design integral membrane channelrhodopsins for efficient eukaryotic expression and plasma membrane localization

There is growing interest in studying and engineering integral membrane proteins (MPs) that play key roles in sensing and regulating cellular response to diverse external signals. A MP must be expressed, correctly inserted and folded in a lipid bilayer, and trafficked to the proper cellular location in order to function. The sequence and structural determinants of these processes are complex and highly constrained. Here we describe a predictive, machine-learning approach that captures this complexity to facilitate successful MP engineering and design. Machine learning on carefully-chosen training sequences made by structure-guided SCHEMA recombination has enabled us to accurately predict the rare sequences in a diverse library of channelrhodopsins (ChRs) that express and localize to the plasma membrane of mammalian cells. These light-gated channel proteins of microbial origin are of interest for neuroscience applications, where expression and localization to the plasma membrane is a prerequisite for function. We trained Gaussian process (GP) classification and regression models with expression and localization data from 218 ChR chimeras chosen from a 118,098-variant library designed by SCHEMA recombination of three parent ChRs. We use these GP models to identify ChRs that express and localize well and show that our models can elucidate sequence and structure elements important for these processes. We also used the predictive models to convert a naturally occurring ChR incapable of mammalian localization into one that localizes well.

Viral Strategies for Targeting the Central and Peripheral Nervous Systems

Recombinant viruses allow for targeted transgene expression in specific cell populations throughout the nervous system. The adeno-associated virus (AAV) is among the most commonly used viruses for neuroscience research. Recombinant AAVs (rAAVs) are highly versatile and can package most cargo composed of desired genes within the capsids ∼5-kb carrying capacity. Numerous regulatory elements and intersectional strategies have been validated in rAAVs to enable cell type-specific expression. rAAVs can be delivered to specific neuronal populations or globally throughout the animal. The AAV capsids have natural cell type or tissue tropism and trafficking that can be modified for increased specificity. Here, we describe recently engineered AAV capsids and associated cargo that have extended the utility of AAVs in targeting molecularly defined neurons throughout the nervous system, which will further facilitate neuronal circuit interrogation and discovery.

Learned Protein Embeddings for Machine Learning

Motivation: Machine‐learning models trained on protein sequences and their measured functions can infer biological properties of unseen sequences without requiring an understanding of the underlying physical or biological mechanisms. Such models enable the prediction and discovery of sequences with optimal properties. Machine‐learning models generally require that their inputs be vectors, and the conversion from a protein sequence to a vector representation affects the models ability to learn. We propose to learn embedded representations of protein sequences that take advantage of the vast quantity of unmeasured protein sequence data available. These embeddings are low‐dimensional and can greatly simplify downstream modeling. Results: The predictive power of Gaussian process models trained using embeddings is comparable to those trained on existing representations, which suggests that embeddings enable accurate predictions despite having orders of magnitude fewer dimensions. Moreover, embeddings are simpler to obtain because they do not require alignments, structural data, or selection of informative amino‐acid properties. Visualizing the embedding vectors shows meaningful relationships between the embedded proteins are captured. Availability and implementation: The embedding vectors and code to reproduce the results are available at https://github.com/fhalab/embeddings_reproduction/. Supplementary information: Supplementary data are available at Bioinformatics online.

Neuronal activity sensing and modulation with Archers

The interplay of electrical signals within our brains gives us the ability for thought, memory, long-term planning, and empathy. A persisting goal for neuroscience is to understand how these percolating signals encode various thought processes and behaviors, and how their erroneous performance leads to neurological diseases with high societal cost. Traditional approaches for exploring neural circuitry involve low throughput but precise electrophysiological recordings, or large-area functional imaging with lower spatial and temporal resolution. Bridging the two requires a technique that enables single-cell and singleaction potential resolution, while detecting brain-wide electrical changes responsible for behavior. In recent years, genetically encoded fluorescent sensors—with targeted expression to specific cell types—have emerged as tools to monitor the electrical activity of a large neuronal population to correlate this with a given stimulus or behavior. These sensors detect either intracellular calcium concentration or the cellular membrane potential (a more direct measure of the cell’s electrical state). Fluorescent calcium sensors have been successful aids in the study of living systems. However, the use of genetically encoded voltage sensors in vivo (to measure membrane potential) remains limited. This is because no single sensor provides both high sensitivity and speed in intact mammalian tissue, which are required for optical imaging of activity in large volumes in vivo. To address these limitations, we modified Archaerhodopsin3 (Arch), a microbial rhodopsin (light-driven) proton pump (a protein that can move protons across a biological membrane), which is known for its high sensitivity to voltage change and its fast response of fluorescence intensity. Specifically, we addressed Arch’s main limitation: its weak baseline fluorescence. Imaging at this low level of emission requires high laser power, which has thus far prohibited the use of Arch in intact living tissue. Figure 1. Archer1 is a bi-functional tool for optogenetic inhibition and sensing of neuronal activity. It is able to fluoresce when excited with red light, or pump protons out of the cell when excited with green light. H: Hydrogen (proton). (Image reproduced with permission from Nicholas Flytzanis, Claire Bedbrook, and Viviana Gradinaru/California Institute of Technology.)