DeepVID v2: Self-Supervised Denoising with Decoupled Spatiotemporal Enhancement for Low-Photon Voltage Imaging.

Significance: Voltage imaging is a powerful tool for studying the dynamics of neuronal activities in the brain. However, voltage imaging data are fundamentally corrupted by severe Poisson noise in the low-photon regime, which hinders the accurate extraction of neuronal activities. Self-supervised deep learning denoising methods have shown great potential in addressing the challenges in low-photon voltage imaging without the need for ground truth, but usually suffer from the tradeoff between spatial and temporal performance. Aim: We present DeepVID v2, a novel self-supervised denoising framework with decoupled spatial and temporal enhancement capability to significantly augment low-photon voltage imaging. Approach: DeepVID v2 is built on our original DeepVID framework,1,2 which performs frame-based denoising by utilizing a sequence of frames around the central frame targeted for denoising to leverage temporal information and ensure consistency. The network further integrates multiple blind pixels in the central frame to enrich the learning of local spatial information. Additionally, DeepVID v2 introduces a new edge extraction branch to capture fine structural details in order to learn high spatial resolution information. Results: We demonstrate that DeepVID v2 is able to overcome the tradeoff between spatial and temporal performance, and achieve superior denoising capability in resolving both high-resolution spatial structures and rapid temporal neuronal activities. We further show that DeepVID v2 is able to generalize to different imaging conditions, including time-series measurements with various signal-to-noise ratios (SNRs) and in extreme low-photon conditions. Conclusions: Our results underscore DeepVID v2 as a promising tool for enhancing voltage imaging. This framework has the potential to generalize to other low-photon imaging modalities and greatly facilitate the study of neuronal activities in the brain.

High-speed low-light in vivo two-photon voltage imaging of large neuronal populations.

Monitoring spiking activity across large neuronal populations at behaviorally relevant timescales is critical for understanding neural circuit function. Unlike calcium imaging, voltage imaging requires kilohertz sampling rates that reduce fluorescence detection to near shot-noise levels. High-photon flux excitation can overcome photon-limited shot noise, but photobleaching and photodamage restrict the number and duration of simultaneously imaged neurons. We investigated an alternative approach aimed at low two-photon flux, which is voltage imaging below the shot-noise limit. This framework involved developing positive-going voltage indicators with improved spike detection (SpikeyGi and SpikeyGi2); a two-photon microscope ('SMURF') for kilohertz frame rate imaging across a 0.4 mm × 0.4 mm field of view; and a self-supervised denoising algorithm (DeepVID) for inferring fluorescence from shot-noise-limited signals. Through these combined advances, we achieved simultaneous high-speed deep-tissue imaging of more than 100 densely labeled neurons over 1 hour in awake behaving mice. This demonstrates a scalable approach for voltage imaging across increasing neuronal populations.

Voltage imaging in the olfactory bulb using transgenic mouse lines expressing the genetically encoded voltage indicator ArcLight.

Genetically encoded voltage indicators (GEVIs) allow optical recordings of membrane potential changes in defined cell populations. Transgenic reporter animals that facilitate precise and repeatable targeting with high expression levels would further the use of GEVIs in the in vivo mammalian brain. However, the literature on developing and applying transgenic mouse lines as vehicles for GEVI expression is limited. Here we report the first in vivo experiments using a transgenic reporter mouse for the GEVI ArcLight, which utilizes a Cre/tTA dependent expression system (TIGRE 1.0). We developed two mouse lines with ArcLight expression restricted to either olfactory receptor neurons, or a subpopulation of interneurons located in the granule and glomerular layers in the olfactory bulb. The ArcLight expression in these lines was sufficient for in vivo imaging of odorant responses in single trials using epifluorescence and 2-photon imaging. The voltage responses were odor-specific and concentration-dependent, which supported earlier studies about perceptual transformations carried out by the bulb that used calcium sensors of neural activity. This study demonstrates that the ArcLight transgenic line is a flexible genetic tool that can be used to record the neuronal electrical activity of different cell types with a signal-to-noise ratio that is comparable to previous reports using viral transduction.

Genetically encoded fluorescent voltage indicators: are we there yet?

In order to understand how brain activity produces adaptive behavior we need large-scale, high-resolution recordings of neuronal activity. Fluorescent genetically encoded voltage indicators (GEVIs) offer the potential for these recordings to be performed chronically from targeted cells in a minimally invasive manner. As the number of GEVIs successfully tested for in vivo use grows, so has the number of open questions regarding the improvements that would facilitate broad adoption of this technology that surpasses mere 'proof of principle' studies. Our aim in this review is not to provide a status check of the current state of the field, as excellent publications covering this topic already exist. Here, we discuss specific questions regarding GEVI development and application that we think are crucial in achieving this goal.

Directed Evolution of Key Residues in Fluorescent Protein Inverses the Polarity of Voltage Sensitivity in the Genetically Encoded Indicator ArcLight.

Genetically encoded calcium indicators (GECIs) produce unprecedentedly large signals that have enabled routine optical recording of single neuron activity in vivo in rodent brain. Genetically encoded voltage indicators (GEVIs) offer a more direct measure of neuronal electrical status, however the signal-to-noise characteristics and signal polarity of the probes developed to date have precluded routine use in vivo. We applied directed evolution to target modulable areas of the fluorescent protein in GEVI ArcLight to create the first GFP-based GEVI (Marina) that exhibits a ΔF/ΔV with a positive slope relationship. We found that only three rounds of site-directed mutagenesis produced a family of "brightening" GEVIs with voltage sensitivities comparable to that seen in the parent probe ArcLight. This shift in signal polarity is an essential first step to producing voltage indicators with signal-to-noise characteristics comparable to GECIs to support widespread use in vivo.

A Bright and Fast Red Fluorescent Protein Voltage Indicator That Reports Neuronal Activity in Organotypic Brain Slices.

Optical imaging of voltage indicators based on green fluorescent proteins (FPs) or archaerhodopsin has emerged as a powerful approach for detecting the activity of many individual neurons with high spatial and temporal resolution. Relative to green FP-based voltage indicators, a bright red-shifted FP-based voltage indicator has the intrinsic advantages of lower phototoxicity, lower autofluorescent background, and compatibility with blue-light-excitable channelrhodopsins. Here, we report a bright red fluorescent voltage indicator (fluorescent indicator for voltage imaging red; FlicR1) with properties that are comparable to the best available green indicators. To develop FlicR1, we used directed protein evolution and rational engineering to screen libraries of thousands of variants. FlicR1 faithfully reports single action potentials (∼3% ΔF/F) and tracks electrically driven voltage oscillations at 100 Hz in dissociated Sprague Dawley rat hippocampal neurons in single trial recordings. Furthermore, FlicR1 can be easily imaged with wide-field fluorescence microscopy. We demonstrate that FlicR1 can be used in conjunction with a blue-shifted channelrhodopsin for all-optical electrophysiology, although blue light photoactivation of the FlicR1 chromophore presents a challenge for applications that require spatially overlapping yellow and blue excitation.

Adaptive Evolution of Eel Fluorescent Proteins from Fatty Acid Binding Proteins Produces Bright Fluorescence in the Marine Environment.

We report the identification and characterization of two new members of a family of bilirubin-inducible fluorescent proteins (FPs) from marine chlopsid eels and demonstrate a key region of the sequence that serves as an evolutionary switch from non-fluorescent to fluorescent fatty acid-binding proteins (FABPs). Using transcriptomic analysis of two species of brightly fluorescent Kaupichthys eels (Kaupichthys hyoproroides and Kaupichthys n. sp.), two new FPs were identified, cloned and characterized (Chlopsid FP I and Chlopsid FP II). We then performed phylogenetic analysis on 210 FABPs, spanning 16 vertebrate orders, and including 163 vertebrate taxa. We show that the fluorescent FPs diverged as a protein family and are the sister group to brain FABPs. Our results indicate that the evolution of this family involved at least three gene duplication events. We show that fluorescent FABPs possess a unique, conserved tripeptide Gly-Pro-Pro sequence motif, which is not found in non-fluorescent fatty acid binding proteins. This motif arose from a duplication event of the FABP brain isoforms and was under strong purifying selection, leading to the classification of this new FP family. Residues adjacent to the motif are under strong positive selection, suggesting a further refinement of the eel protein's fluorescent properties. We present a phylogenetic reconstruction of this emerging FP family and describe additional fluorescent FABP members from groups of distantly related eels. The elucidation of this class of fish FPs with diverse properties provides new templates for the development of protein-based fluorescent tools. The evolutionary adaptation from fatty acid-binding proteins to fluorescent fatty acid-binding proteins raises intrigue as to the functional role of bright green fluorescence in this cryptic genus of reclusive eels that inhabit a blue, nearly monochromatic, marine environment.

Fluorescent protein voltage probes derived from ArcLight that respond to membrane voltage changes with fast kinetics.

We previously reported the discovery of a fluorescent protein voltage probe, ArcLight, and its derivatives that exhibit large changes in fluorescence intensity in response to changes of plasma membrane voltage. ArcLight allows the reliable detection of single action potentials and sub-threshold activities in individual neurons and dendrites. The response kinetics of ArcLight (τ1-on ~10 ms, τ2-on ~ 50 ms) are comparable with most published genetically-encoded voltage probes. However, probes using voltage-sensing domains other than that from the Ciona intestinalis voltage sensitive phosphatase exhibit faster kinetics. Here we report new versions of ArcLight, in which the Ciona voltage-sensing domain was replaced with those from chicken, zebrafish, frog, mouse or human. We found that the chicken and zebrafish-based ArcLight exhibit faster kinetics, with a time constant (τ) less than 6 ms for a 100 mV depolarization. Although the response amplitude of these two probes (8-9%) is not as large as the Ciona-based ArcLight (~35%), they are better at reporting action potentials from cultured neurons at higher frequency. In contrast, probes based on frog, mouse and human voltage sensing domains were either slower than the Ciona-based ArcLight or had very small signals.

Genetically targeted optical electrophysiology in intact neural circuits.

Nervous systems process information by integrating the electrical activity of neurons in complex networks. This motivates the long-standing interest in using optical methods to simultaneously monitor the membrane potential of multiple genetically targeted neurons via expression of genetically encoded fluorescent voltage indicators (GEVIs) in intact neural circuits. No currently available GEVIs have demonstrated robust signals in intact brain tissue that enable reliable recording of individual electrical events simultaneously in multiple neurons. Here, we show that the recently developed "ArcLight" GEVI robustly reports both subthreshold events and action potentials in genetically targeted neurons in the intact Drosophila fruit fly brain and reveals electrical signals in neurite branches. In the same way that genetically encoded fluorescent sensors have revolutionized the study of intracellular Ca(2+) signals, ArcLight now enables optical measurement in intact neural circuits of membrane potential, the key cellular parameter that underlies neuronal information processing.

Single action potentials and subthreshold electrical events imaged in neurons with a fluorescent protein voltage probe.

Monitoring neuronal electrical activity using fluorescent protein-based voltage sensors has been limited by small response magnitudes and slow kinetics of existing probes. Here we report the development of a fluorescent protein voltage sensor, named ArcLight, and derivative probes that exhibit large changes in fluorescence intensity in response to voltage changes. ArcLight consists of the voltage-sensing domain of Ciona intestinalis voltage-sensitive phosphatase and super ecliptic pHluorin that carries the point mutation A227D. The fluorescence intensity of ArcLight A242 decreases by 35% in response to a 100 mV depolarization when measured in HEK293 cells, which is more than five times larger than the signals from previously reported fluorescent protein voltage sensors. We show that the combination of signal size and response speed of these new probes allows the reliable detection of single action potentials and excitatory potentials in individual neurons and dendrites.

A fluorescent, genetically-encoded voltage probe capable of resolving action potentials.

There is a pressing need in neuroscience for genetically-encoded, fluorescent voltage probes that can be targeted to specific neurons and circuits to allow study of neural activity using fluorescent imaging. We created 90 constructs in which the voltage sensing portion (S1-S4) of Ciona intestinalis voltage sensitive phosphatase (CiVSP) was fused to circularly permuted eGFP. This led to ElectricPk, a probe that is an order of magnitude faster (taus ~1-2 ms) than any currently published fluorescent protein-based voltage probe. ElectricPk can follow the rise and fall of neuronal action potentials with a modest decrease in fluorescence intensity (~0.7% ΔF/F). The probe has a nearly linear fluorescence/membrane potential response to both hyperpolarizing and depolarizing steps. This is the first probe based on CiVSP that captures the rapid movements of the voltage sensor, suggesting that voltage probes designed with circularly permuted fluorescent proteins may have some advantages.

Genetically encoded fluorescent voltage sensors using the voltage-sensing domain of Nematostella and Danio phosphatases exhibit fast kinetics.

A substantial increase in the speed of the optical response of genetically encoded fluorescent protein voltage sensors (FP voltage sensors) was achieved by using the voltage-sensing phosphatase genes of Nematostella vectensis and Danio rerio. A potential N. vectensis voltage-sensing phosphatase was identified in silico. The voltage-sensing domain (S1-S4) of the N. vectensis homolog was used to create an FP voltage sensor called Nema. By replacing the phosphatase with a cerulean/citrine FRET pair, a new FP voltage sensor was synthesized with fast off kinetics (Tau(off)<5ms). However, the signal was small (ΔF/F=0.4%/200mV). FP voltage sensors using the D. rerio voltage-sensing phosphatase homolog, designated Zahra and Zahra 2, exhibited fast on and off kinetics within 2ms of the time constants observed with the organic voltage-sensitive dye, di4-ANEPPS. Mutagenesis of the S4 region of the Danio FP voltage sensor shifted the voltage dependence to more negative potentials but did not noticeably affect the kinetics of the optical signal.

Herbicide phosphinothricin causes direct stimulation hormesis.

Herbicide phosphinothricin (PPT) inhibits glutamine synthetase (GS), a key enzyme in nitrogen assimilation, thus causing ammonia accumulation, glutamine depletion and eventually plant death. However, the growth response of Lotus corniculatus L. plants immersed in solutions with a broad range of PPT concentrations is biphasic, with pronounced stimulating effect on biomass production at concentrations ≤ 50 µM and growth inhibition at higher concentrations. The growth stimulation at low PPT concentrations is a result of activation of chloroplastic isoform GS2, while the growth suppression is caused by inhibition of both cytosolic GS1 and GS2 at higher PPT concentrations. Since the results are obtained in cell-free system (e.g. protein extracts), to which the principles of homeostasis are not applicable, this PPT effect is an unambiguous example of direct stimulation hormesis. A detailed molecular mechanism of concentration-dependent interaction of both PPT and a related GS inhibitor, methionine sulfoximine, with GS holoenzymes is proposed. The mechanism is in concurrence with all experimental and literature data.

Design constraints for mobile, high-speed fluorescence brain imaging in awake animals.

In this paper we present a fully self-contained imaging instrument (30 mm overall length) that is capable of recording high speed and detect relatively small fluorescent signals (0.1% ΔF/F) from brain tissues potentially containing genetically-encoded sensors or dyes. This device potentially enables the study of neuronal activity in awake and mobile animals during natural behaviors without the stress and suppression of anesthesia and restraint. The device is a fully self-contained illumination system, wide field fluorescence microscope (~ 4.8 mm² FOV-25 um lateral resolution-1.8 × magnification-0.39 NA) and CMOS image sensor (32 × 32). The total weight of the system is 10 g and is capable of imaging up to 900 fps. We present voltage dye RH1692 experiments using the system to study the somatosensory cortex of mice during whisker movements using an air puff.

Head-mountable high speed camera for optical neural recording.

We report a head-mountable CMOS camera for recording rapid neuronal activity in freely moving rodents using fluorescent activity reporters. This small, lightweight camera is capable of detecting small changes in light intensity (0.2% ΔI/I) at 500fps. The camera has a resolution of 32×32, sensitivity of 0.62V/lxs, conversion gain of 0.52µV/e(-) and well capacity of 2.1Me(-). The camera, containing intensity offset subtraction circuitry within the imaging chip, is part of a miniaturized epi-fluorescent microscope and represents a first generation, mobile scientific-grade, physiology imaging camera.

Induction of peroxidases and superoxide dismutases in transformed embryogenic calli of alfalfa (Medicago sativa L.).

Peroxidase (POD) and superoxide dismutase (SOD) enzyme activities were analyzed in non-regenerative transformed embryogenic lines of alfalfa (Medicago sativa L.) carrying wound-inducible oryzacystatin I (OC-I), wound-inducible oryzacystatin I antisense (OC-Ias), or hygromycin phosphotransferase (hpt) genes. All of the transformed lines analyzed had elevated levels of all POD isoforms. Three POD isoforms with pI values of approximately 4.5, 4.8, and 8.4, and one additional pair of isoforms with a pI value of approximately 8.8 were separated from tissue extracts of all transgenic lines. Isoelectrofocusing patterns revealed the induction of one isoform of SOD with a pI of about 5.6 in all transgenic lines compared with non-transformed embryogenic tissue. These results indicate that the process of transformation may disrupt redox homeostasis in alfalfa tissues.