VolPy: Automated and scalable analysis pipelines for voltage imaging datasets

Deep Learning Enhanced Label-Free Action Potential Detection Using Plasmonic-Based Electrochemical Impedance Microscopy

Measuring neuronal electrical activity, such as action potential propagation in cells, requires the sensitive detection of the weak electrical signal with high spatial and temporal resolution. None of the existing tools can fulfill this need. Recently, plasmonic-based electrochemical impedance microscopy (P-EIM) was demonstrated for the label-free mapping of the ignition and propagation of action potentials in neuron cells with subcellular resolution. However, limited by the signal-to-noise ratio in the high-speed P-EIM video, action potential mapping was achieved by averaging 90 cycles of signals. Such extensive averaging is not desired and may not always be feasible due to factors such as neuronal desensitization. In this study, we utilized advanced signal processing techniques to detect action potentials in P-EIM extracted signals with fewer averaged cycles. Matched filtering successfully detected action potential signals with as few as averaging five cycles of signals. Long short-term memory (LSTM) recurrent neural network achieved the best performance and was able to detect single-cycle stimulated action potential successfully [satisfactory area under the receiver operating characteristic curve (AUC) equal to 0.855]. Therefore, we show that deep learning-based signal processing can dramatically improve the usability of P-EIM mapping of neuronal electrical signals.

Scanless two-photon voltage imaging

Two-photon voltage imaging has long been heralded as a transformative approach capable of answering many long-standing questions in modern neuroscience. However, exploiting its full potential requires the development of novel imaging approaches well suited to the photophysical properties of genetically encoded voltage indicators. We demonstrate that parallel excitation approaches developed for scanless two-photon photostimulation enable high-SNR two-photon voltage imaging. We use whole-cell patch-clamp electrophysiology to perform a thorough characterization of scanless two-photon voltage imaging using three parallel illumination approaches and lasers with different repetition rates and wavelengths. We demonstrate voltage recordings of high-frequency spike trains and sub-threshold depolarizations from neurons expressing the soma-targeted genetically encoded voltage indicator JEDI-2P-Kv. Using a low repetition-rate laser, we perform multi-cell recordings from up to fifteen targets simultaneously. We co-express JEDI-2P-Kv and the channelrhodopsin ChroME-ST and capitalize on their overlapping two-photon absorption spectra to simultaneously evoke and image action potentials using a single laser source. We also demonstrate in vivo scanless two-photon imaging of multiple cells simultaneously up to 250 µm deep in the barrel cortex of head-fixed, anaesthetised mice.

Large-scale deep tissue voltage imaging with targeted-illumination confocal microscopy

Voltage imaging with cellular specificity has been made possible by advances in genetically encoded voltage indicators. However, the kilohertz rates required for voltage imaging lead to weak signals. Moreover, out-of-focus fluorescence and tissue scattering produce background that both undermines the signal-to-noise ratio and induces crosstalk between cells, making reliable in vivo imaging in densely labeled tissue highly challenging. We describe a microscope that combines the distinct advantages of targeted illumination and confocal gating while also maximizing signal detection efficiency. The resulting benefits in signal-to-noise ratio and crosstalk reduction are quantified experimentally and theoretically. Our microscope provides a versatile solution for enabling high-fidelity in vivo voltage imaging at large scales and penetration depths, which we demonstrate across a wide range of imaging conditions and different genetically encoded voltage indicator classes.

A subclass of evening cells promotes the switch from arousal to sleep at dusk

Animals exhibit rhythmic patterns of behavior that are shaped by an internal circadian clock and the external environment. Although light intensity varies across the day, there are particularly robust differences at twilight (dawn/dusk). These periods are also associated with major changes in behavioral states, such as the transition from arousal to sleep. However, the neural mechanisms by which time and environmental conditions promote these behavioral transitions are poorly defined. Here, we show that the E1 subclass of Drosophila evening clock neurons promotes the transition from arousal to sleep at dusk. We first demonstrate that the cell-autonomous clocks of E2 neurons primarily drive and adjust the phase of evening anticipation, the canonical behavior associated with "evening" clock neurons. We next show that conditionally silencing E1 neurons causes a significant delay in sleep onset after dusk. However, rather than simply promoting sleep, activating E1 neurons produces time- and light-dependent effects on behavior. Activation of E1 neurons has no effect early in the day but then triggers arousal before dusk and induces sleep after dusk. Strikingly, these activation-induced phenotypes depend on the presence of light during the day. Despite their influence on behavior around dusk, in vivo voltage imaging of E1 neurons reveals that their spiking rate and pattern do not significantly change throughout the day. Moreover, E1-specific clock ablation has no effect on arousal or sleep. Thus, we suggest that, rather than specifying "evening" time, E1 neurons act, in concert with other rhythmic neurons, to promote behavioral transitions at dusk.

Imaging the voltage of neurons distributed across entire brains of larval zebrafish

Neurons interact in networks distributed throughout the brain. Although much effort has focused on whole-brain calcium imaging, recent advances in genetically encoded voltage indicators (GEVIs) raise the possibility of imaging voltage of neurons distributed across brains. To achieve this, a microscope must image at high volumetric rate and signal-to-noise ratio. We present a remote scanning light-sheet microscope capable of imaging GEVI-expressing neurons distributed throughout entire brains of larval zebrafish at a volumetric rate of 200.8 Hz. We measured voltage of ∼1/3 of the neurons of the brain, distributed throughout. We observed that neurons firing at different times during a sequence were located at different brain locations, for sequences elicited by a visual stimulus, which mapped onto locations throughout the optic tectum, as well as during stimulus-independent bursts, which mapped onto locations in the cerebellum and medulla. Whole-brain voltage imaging may open up frontiers in the fundamental operation of neural systems.

Widefield imaging of rapid pan-cortical voltage dynamics with an indicator evolved for one-photon microscopy

Widefield imaging with genetically encoded voltage indicators (GEVIs) is a promising approach for understanding the role of large cortical networks in the neural coding of behavior. However, the limited performance of current GEVIs restricts their deployment for single-trial imaging of rapid neuronal voltage dynamics. Here, we developed a high-throughput platform to screen for GEVIs that combine fast kinetics with high brightness, sensitivity, and photostability under widefield one-photon illumination. Rounds of directed evolution produced JEDI-1P, a green-emitting fluorescent indicator with enhanced performance across all metrics. Next, we optimized a neonatal intracerebroventricular delivery method to achieve cost-effective and wide-spread JEDI-1P expression in mice. We also developed an approach to correct optical measurements from hemodynamic and motion artifacts effectively. Finally, we achieved stable brain-wide voltage imaging and successfully tracked gamma-frequency whisker and visual stimulations in awake mice in single trials, opening the door to investigating the role of high-frequency signals in brain computations.

Statistically unbiased prediction enables accurate denoising of voltage imaging data

Here we report SUPPORT (statistically unbiased prediction utilizing spatiotemporal information in imaging data), a self-supervised learning method for removing Poisson-Gaussian noise in voltage imaging data. SUPPORT is based on the insight that a pixel value in voltage imaging data is highly dependent on its spatiotemporal neighboring pixels, even when its temporally adjacent frames alone do not provide useful information for statistical prediction. Such dependency is captured and used by a convolutional neural network with a spatiotemporal blind spot to accurately denoise voltage imaging data in which the existence of the action potential in a time frame cannot be inferred by the information in other frames. Through simulations and experiments, we show that SUPPORT enables precise denoising of voltage imaging data and other types of microscopy image while preserving the underlying dynamics within the scene.

FIOLA: an accelerated pipeline for fluorescence imaging online analysis

Optical microscopy methods such as calcium and voltage imaging enable fast activity readout of large neuronal populations using light. However, the lack of corresponding advances in online algorithms has slowed progress in retrieving information about neural activity during or shortly after an experiment. This gap not only prevents the execution of real-time closed-loop experiments, but also hampers fast experiment-analysis-theory turnover for high-throughput imaging modalities. Reliable extraction of neural activity from fluorescence imaging frames at speeds compatible with indicator dynamics and imaging modalities poses a challenge. We therefore developed FIOLA, a framework for fluorescence imaging online analysis that extracts neuronal activity from calcium and voltage imaging movies at speeds one order of magnitude faster than state-of-the-art methods. FIOLA exploits algorithms optimized for parallel processing on GPUs and CPUs. We demonstrate reliable and scalable performance of FIOLA on both simulated and real calcium and voltage imaging datasets. Finally, we present an online experimental scenario to provide guidance in setting FIOLA parameters and to highlight the trade-offs of our approach.

Sensitivity optimization of a rhodopsin-based fluorescent voltage indicator

The ability to optically image cellular transmembrane voltages at millisecond-timescale resolutions can offer unprecedented insight into the function of living brains in behaving animals. Here, we present a point mutation that increases the sensitivity of Ace2 opsin-based voltage indicators. We use the mutation to develop Voltron2, an improved chemigeneic voltage indicator that has a 65% higher sensitivity to single APs and 3-fold higher sensitivity to subthreshold potentials than Voltron. Voltron2 retained the sub-millisecond kinetics and photostability of its predecessor, although with lower baseline fluorescence. In multiple in vitro and in vivo comparisons with its predecessor across multiple species, we found Voltron2 to be more sensitive to APs and subthreshold fluctuations. Finally, we used Voltron2 to study and evaluate the possible mechanisms of interneuron synchronization in the mouse hippocampus. Overall, we have discovered a generalizable mutation that significantly increases the sensitivity of Ace2 rhodopsin-based sensors, improving their voltage reporting capability.

Voltage imaging reveals that hippocampal interneurons tune memory-encoding pyramidal sequences

Hippocampal spiking sequences encode and link behavioral information across time. How inhibition sculpts these sequences remains unknown. We performed longitudinal voltage imaging of CA1 parvalbumin- and somatostatin-expressing interneurons in mice during an odor-cued working memory task, before and after training. During this task, pyramidal odor-specific sequences encode the cue throughout a delay period. In contrast, most interneurons encoded odor delivery, but not odor identity, nor delay time. Population inhibition was stable across days, with constant field turnover, though some cells retained odor-responses for days. At odor onset, a brief, synchronous burst of parvalbumin cells was followed by widespread membrane hyperpolarization and then rebound theta-paced spiking, synchronized across cells. Two-photon calcium imaging revealed that most pyramidal cells were suppressed throughout the odor. Positive pyramidal odor-responses coincided with interneuronal rebound spiking; otherwise, they had weak odor-selectivity. Therefore, inhibition increases the signal-to-noise ratio of cue representations, which is crucial for entraining downstream targets.

Bringing to light the physiological and pathological firing patterns of human induced pluripotent stem cell-derived neurons using optical recordings

Human induced pluripotent stem cells (hiPSCs) are a promising approach to study neurological and neuropsychiatric diseases. Most methods to record the activity of these cells have major drawbacks as they are invasive or they do not allow single cell resolution. Genetically encoded voltage indicators (GEVIs) open the path to high throughput visualization of undisturbed neuronal activity. However, conventional GEVIs perturb membrane integrity through inserting multiple copies of transmembrane domains into the plasma membrane. To circumvent large add-ons to the plasma membrane, we used a minimally invasive novel hybrid dark quencher GEVI to record the physiological and pathological firing patterns of hiPSCs-derived sensory neurons from patients with inherited erythromelalgia, a chronic pain condition associated with recurrent attacks of redness and swelling in the distal extremities. We observed considerable differences in action potential firing patterns between patient and control neurons that were previously overlooked with other recording methods. Our system also performed well in hiPSC-derived forebrain neurons where it detected spontaneous synchronous bursting behavior, thus opening the path to future applications in other cell types and disease models including Parkinson's disease, Alzheimer's disease, epilepsy, and schizophrenia, conditions associated with disturbances of neuronal activity and synchrony.

Dual-polarity voltage imaging of the concurrent dynamics of multiple neuron types

Genetically encoded fluorescent voltage indicators are ideally suited to reveal the millisecond-scale interactions among and between targeted cell populations. However, current indicators lack the requisite sensitivity for in vivo multipopulation imaging. We describe next-generation green and red voltage sensors, Ace-mNeon2 and VARNAM2, and their reverse response-polarity variants pAce and pAceR. Our indicators enable 0.4- to 1-kilohertz voltage recordings from >50 spiking neurons per field of view in awake mice and ~30-minute continuous imaging in flies. Using dual-polarity multiplexed imaging, we uncovered brain state–dependent antagonism between neocortical somatostatin-expressing (SST + ) and vasoactive intestinal peptide–expressing (VIP + ) interneurons and contributions to hippocampal field potentials from cell ensembles with distinct axonal projections. By combining three mutually compatible indicators, we performed simultaneous triple-population imaging. These approaches will empower investigations of the dynamic interplay between neuronal subclasses at single-spike resolution.

Review of data processing of functional optical microscopy for neuroscience

Functional optical imaging in neuroscience is rapidly growing with the development of optical systems and fluorescence indicators. To realize the potential of these massive spatiotemporal datasets for relating neuronal activity to behavior and stimuli and uncovering local circuits in the brain, accurate automated processing is increasingly essential. We cover recent computational developments in the full data processing pipeline of functional optical microscopy for neuroscience data and discuss ongoing and emerging challenges.

Sustained deep-tissue voltage recording using a fast indicator evolved for two-photon microscopy

Genetically encoded voltage indicators are emerging tools for monitoring voltage dynamics with cell-type specificity. However, current indicators enable a narrow range of applications due to poor performance under two-photon microscopy, a method of choice for deep-tissue recording. To improve indicators, we developed a multiparameter high-throughput platform to optimize voltage indicators for two-photon microscopy. Using this system, we identified JEDI-2P, an indicator that is faster, brighter, and more sensitive and photostable than its predecessors. We demonstrate that JEDI-2P can report light-evoked responses in axonal termini of Drosophila interneurons and the dendrites and somata of amacrine cells of isolated mouse retina. JEDI-2P can also optically record the voltage dynamics of individual cortical neurons in awake behaving mice for more than 30 min using both resonant-scanning and ULoVE random-access microscopy. Finally, ULoVE recording of JEDI-2P can robustly detect spikes at depths exceeding 400 μm and report voltage correlations in pairs of neurons.

Voltage imaging identifies spinal circuits that modulate locomotor adaptation in zebrafish

Motor systems must continuously adapt their output to maintain a desired trajectory. While the spinal circuits underlying rhythmic locomotion are well described, little is known about how the network modulates its output strength. A major challenge has been the difficulty of recording from spinal neurons during behavior. Here, we use voltage imaging to map the membrane potential of large populations of glutamatergic neurons throughout the spinal cord of the larval zebrafish during fictive swimming in a virtual environment. We characterized a previously undescribed subpopulation of tonic-spiking ventral V3 neurons whose spike rate correlated with swimming strength and bout length. Optogenetic activation of V3 neurons led to stronger swimming and longer bouts but did not affect tail beat frequency. Genetic ablation of V3 neurons led to reduced locomotor adaptation. The power of voltage imaging allowed us to identify V3 neurons as a critical driver of locomotor adaptation in zebrafish.

Imaging Voltage with Microbial Rhodopsins

Membrane potential is the critical parameter that reflects the excitability of a neuron, and it is usually measured by electrophysiological recordings with electrodes. However, this is an invasive approach that is constrained by the problems of lacking spatial resolution and genetic specificity. Recently, the development of a variety of fluorescent probes has made it possible to measure the activity of individual cells with high spatiotemporal resolution. The adaptation of this technique to image electrical activity in neurons has become an informative method to study neural circuits. Genetically encoded voltage indicators (GEVIs) can be used with superior performance to accurately target specific genetic populations and reveal neuronal dynamics on a millisecond scale. Microbial rhodopsins are commonly used as optogenetic actuators to manipulate neuronal activities and to explore the circuit mechanisms of brain function, but they also can be used as fluorescent voltage indicators. In this review, we summarize recent advances in the design and the application of rhodopsin-based GEVIs.

Optical Spike Detection and Connectivity Analysis With a Far-Red Voltage-Sensitive Fluorophore Reveals Changes to Network Connectivity in Development and Disease

The ability to optically record dynamics of neuronal membrane potential promises to revolutionize our understanding of neurobiology. In this study, we show that the far-red voltage sensitive fluorophore, Berkeley Red Sensor of Transmembrane potential-1, or BeRST 1, can be used to monitor neuronal membrane potential changes across dozens of neurons at a sampling rate of 500 Hz. Notably, voltage imaging with BeRST 1 can be implemented with affordable, commercially available illumination sources, optics, and detectors. BeRST 1 is well-tolerated in cultures of rat hippocampal neurons and provides exceptional optical recording fidelity, as judged by dual fluorescence imaging and patch-clamp electrophysiology. We developed a semi-automated spike-picking program to reduce user bias when calling action potentials and used this in conjunction with BeRST 1 to develop an optical spike and connectivity analysis (OSCA) for high-throughput dissection of neuronal activity dynamics. The high temporal resolution of BeRST 1 enables dissection of firing rate changes in response to acute, pharmacological interventions with commonly used inhibitors like gabazine and picrotoxin. Over longer periods of time, BeRST 1 also tracks chronic perturbations to neurons exposed to amyloid beta 1-42 (Aβ 1-42), revealing modest changes to spiking frequency but profound changes to overall network connectivity. Finally, we use OSCA to track changes in neuronal connectivity during maturation in culture, providing a functional readout of network assembly. We envision that use of BeRST 1 and OSCA described here will be of use to the broad neuroscience community.

High-fidelity estimates of spikes and subthreshold waveforms from 1-photon voltage imaging in vivo

The ability to probe the membrane potential of multiple genetically defined neurons simultaneously would have a profound impact on neuroscience research. Genetically encoded voltage indicators are a promising tool for this purpose, and recent developments have achieved a high signal-to-noise ratio in vivo with 1-photon fluorescence imaging. However, these recordings exhibit several sources of noise and signal extraction remains a challenge. We present an improved signal extraction pipeline, spike-guided penalized matrix decomposition-nonnegative matrix factorization (SGPMD-NMF), which resolves supra- and subthreshold voltages in vivo. The method incorporates biophysical and optical constraints. We validate the pipeline with simultaneous patch-clamp and optical recordings from mouse layer 1 in vivo and with simulated and composite datasets with realistic noise. We demonstrate applications to mouse hippocampus expressing paQuasAr3-s or SomArchon1, mouse cortex expressing SomArchon1 or Voltron, and zebrafish spines expressing zArchon1.

VolPy: Automated and scalable analysis pipelines for voltage imaging datasets

Voltage imaging enables monitoring neural activity at sub-millisecond and sub-cellular scale, unlocking the study of subthreshold activity, synchrony, and network dynamics with unprecedented spatio-temporal resolution. However, high data rates (>800MB/s) and low signal-to-noise ratios create bottlenecks for analyzing such datasets. Here we present VolPy, an automated and scalable pipeline to pre-process voltage imaging datasets. VolPy features motion correction, memory mapping, automated segmentation, denoising and spike extraction, all built on a highly parallelizable, modular, and extensible framework optimized for memory and speed. To aid automated segmentation, we introduce a corpus of 24 manually annotated datasets from different preparations, brain areas and voltage indicators. We benchmark VolPy against ground truth segmentation, simulations and electrophysiology recordings, and we compare its performance with existing algorithms in detecting spikes. Our results indicate that VolPy's performance in spike extraction and scalability are state-of-the-art.