Probing neuronal activity with genetically encoded calcium and voltage fluorescent indicators

Monitoring neural activity in individual neurons is crucial for understanding neural circuits and brain functions. The emergence of optical imaging technologies has dramatically transformed the field of neuroscience, enabling detailed observation of large-scale neuronal populations with both cellular and subcellular resolution. This transformation will be further accelerated by the integration of these imaging technologies and advanced big data analysis. Genetically encoded fluorescent indicators to detect neural activity with high signal-to-noise ratios are pivotal in this advancement. In recent years, these indicators have undergone significant developments, greatly enhancing the understanding of neural dynamics and networks. This review highlights the recent progress in genetically encoded calcium and voltage indicators and discusses the future direction of imaging techniques with big data analysis that deepens our understanding of the complexities of the brain.


Introduction
The brain is composed of an extensive and intricate network of neurons, ranging in number from hundreds of millions to billions.These neurons connect in complex patterns and form circuits that process and interpret large amounts of information, playing a crucial role for the execution of higher brain functions, such as cognition and learning.Therefore, probing functional neural circuits at a high spatiotemporal resolution is crucial for understanding how neuronal populations work together to generate internal brain states and behaviors.This research is pivotal not only for exploring the fundamental principles of the brain function but also for addressing neurological and psychiatric diseases.To address these questions, it is essential to simultaneously measure neural activity from numerous neurons (Yuste and Bargmann, 2017).
Electrophysiological approaches using electrodes, such as patchclamp recording, have traditionally been the gold standard for measuring membrane potential (Neher and Sakmann, 1976).However, these electrode-based methods often lack spatial resolution and genetic specificity.Optical imaging with genetically encoded indicators can overcome these drawbacks, enabling the monitoring of large neuronal populations simultaneously with cellular or even subcellular resolution.In addition, recently, the rapid growth of big data analysis technologies has further revolutionized this field.The integration of optical imaging with advanced big data analysis marks a new era in neuroscience, unlocking complex and previously inaccessible insights (Landhuis, 2017).The application of machine learning algorithms and other advanced computational methods to imaging datasets has opened new avenues for understanding complex biological systems.These methodologies are crucial for advancing the understanding of various physiological and pathological conditions, including neural development and synaptic plasticity, and the mechanisms underlying neurodegenerative diseases, thereby providing unprecedented insights into brain function (Schneider et al., 2023).To effectively utilize imaging technologies, genetically encoded fluorescent indicators that can detect neural activity with a high signal-to-noise ratio are essential, as well as sophisticated analysis techniques.In recent years, the field of fluorescent probes has demonstrated remarkable growth and advancement, providing more comprehensive understanding of neural dynamics and networks in the brain.
In this review, we will introduce recent advancements in the design and application of genetically encoded calcium and voltage indicators.In addition, we will discuss future perspectives on the integration of imaging techniques with big data analysis that promises to deepen our understanding of the complexities of the brain.

Calcium imaging
Calcium ions (Ca 2+ ) play a crucial role in regulating a variety of cellular functions, including neural activity and muscle contraction.In cortical neurons, the concentration of intracellular free Ca 2+ is maintained at a low level, typically ranging between 30 and 100 nM, when the neuron is in a resting state with a membrane potential of around − 70 mV (Grienberger and Konnerth, 2012).When action potentials occur, Ca 2+ influx into the neuron through voltage-gated calcium channels and other pathways.This influx leads to a temporary spike in calcium concentration within the soma, often increasing by 10-100-fold compared to the resting state, significantly impacting neuronal processes (Berridge et al., 2000).Simultaneous in vivo calcium imaging and electrophysiology in the same neuron have revealed a correlation between calcium signaling in the soma and neuronal firing (Wei et al., 2020).Given its pivotal role, calcium imaging has emerged as a popular method for measuring neuronal activity.It enables visualization and quantification of changes in calcium concentrations, providing valuable insights into how neurons behave and communicate.Calcium imaging has now become an indispensable technique in the field of neuroscience.
Various fluorescent probes have been developed to visualize calcium dynamics in living cells.A pioneering study in this field was conducted by Tsien and colleagues in the early 1980s.They developed the first calcium-sensitive fluorescent dye, Quin-2 (Tsien et al., 1982).Following this seminal development, a variety of highly sensitive dyes, such as Fura-2, Indo-1, and Fluo-4, were introduced (Gee et al., 2000;Grynkiewicz et al., 1985).While a revolutionary technique, the application of these dyes was limited by the need for delivery through glass pipettes or bulk extracellular loading, which constrained cell-type-specific targeting and imaging conditions.Genetically encoded calcium indicators (GECIs) have overcome these limitations, enabling long-term, repetitive, and unbiased functional imaging of specific types of neurons and even subcellular compartments.

FRET-type GECIs
The initial GECI developed by Miyawaki and colleagues was based on Förster resonance energy transfer (FRET) (Miyawaki et al., 1997).This FRET-type GECI can detect changes in Ca 2+ concentrations by measuring the ratio of fluorescence intensity between two fluorescent proteins, cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP), linked by peptides consisting of the calmodulin (CaM) and the calmodulin-binding peptide of myosin light chain kinase (M13) (Fig. 1A).In the absence of Ca 2+ , the emission is primarily from CFP.In the presence of Ca 2+ , intramolecular conformational changes alter the spatial distance between CFP and YFP, resulting in decreased CFP fluorescence and increased YFP fluorescence.Thus, changes in the CFP and YFP emission spectrum correlate with variations in intracellular Ca 2+ concentration, enabling ratiometric imaging through quantification of the YFP/CFP ratio.Subsequently, more sensitive sensors were developed with other fluorescent proteins and calcium-binding proteins (Nagai et al., 2004;Thestrup et al., 2014).One of the advantages of ratiometric imaging is its ability to reduce noise, such as motion artifacts (Michikawa et al., 2021).However, later generations of GECIs have predominantly utilized a single-fluorophore design.

Single-fluorophore GECIs
The most frequently used intensiometric fluorescent GECI was G-CaMP, developed by Nakai and colleagues (Nakai et al., 2001).G-CaMP consists of a circularly permuted green fluorescent protein (cpGFP), CaM, and M13 peptide (Fig. 1B).When Ca 2+ binds to G-CaMP, it causes a conformational change, resulting in increased fluorescence intensity.However, earlier versions of intensiometric fluorescent GECIs lacked sufficient sensitivity, rendering them unsuitable for in vivo imaging.The development of GCaMP6 and its successor jGCaMP7 series from the Janelia Research Campus marked significant improvements, making GECIs widely applicable for measuring neural activity both in vitro and in vivo (Table 1) (Chen et al., 2013;Dana et al., 2019).While previous GECIs were not as effective as Ca 2+ -sensitive organic dyes like Oregon Green BAPTA-1 (OGB-1), the GCaMP6 series demonstrated superior performance in detecting action potentials.Notably, the GCaMP6 series could detect electrical activity in neuropils such as dendritic spines with high signal-to-noise ratio.Additionally, genetic approaches like Cre/-loxP system further enabled cell-type-specific imaging.GECIs can also be stably expressed in cells over extended periods, facilitating long-term imaging which was not achieved by calcium-sensitive dyes.These advancements facilitated monitoring neural activity during the learning process not only in rodents but also in non-human primates (Chu et al., 2016;Ebina et al., 2018).Beyond two-photon microscopy, fiber photometry and microendoscopy have become viable methods for calcium imaging in freely moving animals (Karigo et al., 2021;Kondo et al., 2018;Yukinaga et al., 2022).Despite the widespread adoption of GECIs, they still have limitations in their temporal resolution, affecting their ability to precisely decode the timing and number of spikes in individual neurons.Recent developments of GECIs have aimed to improve the accuracy and reliability of capturing neural activity.
To address this issue, Bito and colleagues redesigned the original GCaMPs by substituting the M13 with a different Ca 2+ /CaM-binding peptide from Ca 2+ /CaM-dependent protein kinase kinase (CaMKK) (Inoue et al., 2015).This modification led to the generation of a new calcium sensor, XCaMP, which exhibits greater sensitivity and linearity in response to Ca 2+ concentration changes compared to existing sensors (Inoue et al., 2019).XCaMP-Gf represented a significant improvement over GCaMP6, featuring a 2.5-fold increase in signal-to-noise ratio for detecting single action potentials, a 10-fold faster rise time, and decay time constants that are twice as fast.These features significantly improve its ability to accurately decode spike numbers and spike timings, especially in high-frequency firing parvalbumin-positive interneurons.Following a similar approach to XCaMP, the Janelia Research Campus introduced jGCaMP8, another fast and sensitive calcium sensor.In jGCaMP8, the M13 is replaced by a peptide derived from endothelial nitric oxide synthase (Table 1) (Zhang et al., 2023).These advanced sensors, with their improved sensitivity and kinetics, provide more precise detection of spike timings and frequencies compared to conventional GECIs.

A color palette of GECIs
Conventionally, GECIs have primarily utilized GFP.Recent advancements in fluorescent proteins have expanded the color palette of GECIs.In particular, red GECIs, potentially similar capability to GCaMP with additional advantages, have become increasingly salient (Dana et al., 2019;Fenno et al., 2020;Inoue et al., 2015Inoue et al., , 2019;;Ohkura et al., 2012;Yokoyama et al., 2024;Zhao et al., 2011).In recent years, highly sensitive red GECIs have been applied for in vivo imaging (Table 1).One of the most significant applications of red GECIs is for multicolor imaging, which enables the simultaneous labeling and monitoring of different neuronal populations in distinct colors.For instance, using red and green calcium sensors into excitatory and inhibitory neurons, respectively, enables distinct visualization of their activities with cellular resolution (Dana et al., 2016;Inoue et al., 2015;Sakamoto et al., 2022).This approach extends beyond merely monitoring neuronal activity; it also facilitates the visualization of neuromodulators such as dopamine, serotonin, orexin, and oxytocin, as well as intracellular signaling molecules (Duffet et al., 2022;Ino et al., 2022;Unger et al., 2020;Yokoyama et al., 2024;Zhuo et al., 2023).Such broad applications provide deeper insights into the mechanisms of brain functions.Furthermore, due to the lower light scattering at longer wavelengths, red GECIs are suitable for deep brain imaging in areas such as hippocampal CA1 region and medial prefrontal cortex of the mouse brain (Inoue et al., 2019;Kondo et al., 2017).Red GECIs can also be effective for measuring neural activity in non-human primates with larger brains, such as common marmosets.In addition to red GECIs, recent developments have introduced a wider range of colors: blue (XCaMP-B) (Inoue et al., 2019), yellow (XCaMP-Y, jYCaMP, NEMO) (Inoue et al., 2019;Li et al., 2023;Mohr et al., 2020), and even near-infrared (NIR--GECO, iBB-GECO) (Hashizume et al., 2022;Qian et al., 2019).These indicators significantly enhanced the versatility of imaging tools in the field of neuroscience.
To achieve functional deep brain imaging, further development is necessary, particularly in enhancing chromophore properties such as brightness.Recent progress in this field is the development of a novel 'chemigenetic' fluorescent calcium indicator platform.This platform utilizes the self-labeling HaloTag protein, which is conjugated to synthetic fluorophores such as Janelia Fluor dye (Fig. 1C) (Deo et al., 2021).This approach yields brighter red and far-red GECIs suitable for deep brain imaging.

Voltage imaging
Although calcium imaging is a robust method for tracking neural  and RCaMP3) and purified protein (XCaMP-R).
activity, the calcium dynamics revealed by GECIs are not a direct proxy of membrane potential changes.Thus, calcium imaging is limited in its ability to provide a complete description of neuronal activity.First, somatic calcium imaging captures predominantly action potentials (Smetters et al., 1999).Subthreshold excitatory or inhibitory synaptic inputs are practically invisible.Second, due to biophysical limitations, calcium dynamics are significantly slower than the timescale of membrane potential changes.This discrepancy makes it challenging to precisely determine the number of spikes and spike timing with calcium imaging when neurons fire a burst of spikes.Third, calcium dynamics are shaped by complicated interactions between ionic diffusion and extrusion, and they can be significantly altered by intrinsic and extrinsic calcium buffers and the expression of calcium indicators themselves (Neher, 1998).Given these constraints, calcium imaging with GECIs is not an ideal technique for capturing the entire spectrum of neural activity.
Voltage imaging can directly monitor the electrical activity of neurons, including subthreshold events, providing a more precise measurement of neuronal dynamics (Peterka et al., 2011;Storace et al., 2016;Zhang et al., 2021).In parallel with GECIs, substantial progress has been made in developing genetically encoded voltage indicators (GEVIs) since the initial GEVI, named Flash, was reported by Isacoff and colleagues (Siegel and Isacoff, 1997).These GEVIs can specifically target and measure neural activity in distinct types or subcellular compartments (Kwon et al., 2017).In addition, these indicators are capable of detecting subthreshold activity that is not detectable by calcium imaging, thereby enabling more accurate decoding of brain functions (Bando et al., 2019(Bando et al., , 2021;;Cornejo et al., 2022).Therefore, voltage imaging with GEVIs is becoming a potent alternative to calcium imaging.Recent advancements in GEVIs can primarily fall into two categories: 1) those utilizing voltage-sensitive domain (VSD) of voltage-sensitive phosphatase and 2) those utilizing microbial rhodopsins.

VSD-based GEVIs
Initially developed GEVIs were structured by fusing fluorescent proteins to voltage-sensitive ion channels.However, these early GEVIs were not practically usable as voltage sensors due to their poor membrane localization in mammalian cells and low signal-to-noise ratio.Since the discovery of a voltage-dependent phosphatase from Ciona intestinalis (Ci-VSP) in 2005, this has been utilized as the fundamental structure of GEVIs (Murata et al., 2005).The VSP comprises voltage-sensitive domains composed of four transmembrane helices (S1-S4).The S4 helix contains several positively charged amino acids, including arginine and lysine, which move in response to membrane potential changes (Akemann et al., 2010;Jin et al., 2012;Tsutsui et al., 2008).The VSD-based GEVIs incorporate a fluorescent protein adjacent to the S4 helix and detect membrane potential changes by monitoring fluorescence changes, which correlate with conformational changes.In recent years, highly sensitive VSD-based GEVIs employing Gallus gallus (Gg)-VSD have been developed.These are suitable for both one-or two-photon imaging, facilitating the measurement of membrane potential changes in vivo with single-cell resolution (Fig. 2A, Table 2) (Evans et al., 2023;Liu et al., 2022;Lu et al., 2023;Platisa et al., 2023;Villette et al., 2019).

Rhodopsin-based GEVIs
Microbial rhodopsins, consisting of rhodopsin apoprotein and lightabsorbing chromophore retinal, serve as light-sensitive ion pumps, ion channels, and sensors (Zhang et al., 2021).These rhodopsins were initially utilized for optogenetic actuators (Boyden et al., 2005;Chow et al., 2010).However, due to the low quantum yield of the retinal chromophore, its fluorescence was overlooked in earlier studies.Cohen and colleagues found that Archaerhodopsin-3 (Arch), derived from Halorubrum sodomense, exhibits voltage-dependent fluorescent changes from the retinal chromophore in neurons, allowing to accurately monitor membrane potentials with high temporal resolution (Kralj et al., 2011).For voltage imaging, Arch and its variants are typically excited with red light (640 nm) and emit in the infrared spectrum (peak at around 715 nm) (Fig. 2B) (Kralj et al., 2011).The voltage sensitivity of these proteins is attributed to the protonation of the Schiff base in the photointermediate state (Maclaurin et al., 2013).However, their practical applications were initially limited due to weak fluorescence and an insufficient signal-to-noise ratio (Kojima et al., 2020).To address these problems, significant efforts, including mutagenesis related to the photocycle or near the Schiff base, have been made (Piatkevich et al., 2018).These efforts led to improved brightness and dynamic range.As a result, several GEVIs suitable for in vivo one-photon imaging have been developed, further advancing the field (Table 3) (Piatkevich et al., 2019;Tian et al., 2023).

eFRET-based GEVIs
Despite extensive efforts to Arch variants, their brightness remains lower than that of commonly used fluorescent proteins.To overcome this issue, an electrochromic Förster resonance energy transfer (eFRET) strategy was developed.Microbial rhodopsins have an absorption spectrum that overlaps with the emission spectrum of popular fluorescent proteins.Therefore, these fluorescent proteins and other chemical fluorophores can serve as FRET donors, while rhodopsin molecules can serve as FRET acceptors (Bayraktar et al., 2012).eFRET sensors detect the absorption change of rhodopsin through the quenching of an attached fluorescent protein's intensity.When neurons depolarize, the fluorescent protein intensity is decreased by FRET from the fluorescent protein to the rhodopsin (Fig. 2C) (Gong et al., 2014;Zou et al., 2014).Thus, FRET-opsin-based GEVIs detect voltage depolarization by the decrease in emission intensity from the fluorescence donor.The rhodopsins used in these GEVIs are not limited to Arch alone (Zou et al., 2014).Others like Mac (bacteriorhodopsin from Leptosphaeria maculans) and Ace2 (bacteriorhodopsin from Acetabularia acetabulum) were also successfully employed to generate new indicators with fast kinetics and high dynamic range (Gong et al., 2015(Gong et al., , 2014;;Kannan et al., 2022).Given the broad absorption spectrum of microbial rhodopsin, a variety of fluorescent proteins with different emission wavelengths can serve as donors, broadening the versatility of these GEVIs, enabling in vivo one-photon voltage imaging with single-cell resolution (Table 3) (Abdelfattah et al., 2019(Abdelfattah et al., , 2020(Abdelfattah et al., , 2023;;Han et al., 2023;Kannan et al., 2022Kannan et al., , 2018;;Kojima et al., 2020).In addition, by utilizing optical fibers, these sensors can accurately detect oscillatory waves in the brain (Kannan et al., 2018;Marshall et al., 2016).Synthetic fluorescent dyes are also available as eFRET donors (Fig. 2D).Voltron includes a HaloTag domain that enables the use of Janelia Fluor dyes as the eFRET donor.These synthetic dyes are more photostable and brighter than the fluorescent proteins, facilitating in vivo one-photon voltage imaging.It is important to note, however, that both rhodopsin-and eFRET-based GEVIs tend to reduce voltage sensitivity under two-photon excitation (Bando et al., 2019;Maclaurin et al., 2013).

Discussion
Here, we introduced the recent progress of genetically encoded biosensors for monitoring neural activity by optical imaging.The development of GECIs and GEVIs helped significantly advance when decoding neural activity with high spatiotemporal resolution precisely.This development has opened new avenues in neuroscience research, particularly in visualizing activity in a subcellular domain, which was previously challenging with conventional electrophysiology (Chen et al., 2013;Cornejo et al., 2022;Kwon et al., 2017).The development of even more sensitive probes in the future is expected to elucidate circuit mechanisms of higher brain functions and biological phenomena that have not been elucidated.Apart from fluorescent biosensors, imaging apparatuses are also crucial.Notably, rapid progress in microscopy for mesoscale imaging has facilitated the measurement of tens of thousands of neurons with single-cell resolution across a wide field of view (Ota et al., 2021;Sofroniew et al., 2016;Stirman et al., 2016).This mesoscale imaging is invaluable for exploring neural networks across an extensive range of brain regions.Mesoscale imaging is also applicable to fluorescent probes for intercellular signaling molecules and neuromodulators, in addition to neural activity, enabling the unraveling of complex intracellular processes that were previously outside of capabilities.Moreover, these techniques will provide profound insights into how multimodal sensory information is processed and integrated into the brain.
Despite these technological advancements, considerable challenges remain, particularly in managing and processing large-scale imaging data.For example, the mesoscopic two-photon microscopy (FASHIO2-PM) developed by Murayama and colleagues captures a 3 mm×3 mm field of view with 2048 × 2048 pixels at 7.5 Hz frame rate, yielding approximately 4 GB per minute, which is a substantial amount for functional imaging.(Ota et al., 2021).Similarly, voltage imaging, which requires higher frame rates to monitor millisecond-order membrane potential fluctuation, can produce approximately 8 GB per minute (512 × 128 pixels at 1 kHz) (Zhang et al., 2021).Furthermore, voltage imaging signals contain multiple waveforms, which require data processing for accurate definition of action potentials, subthreshold activities, and background noise.Such immense volume and complexity of the data demand a robust computational infrastructure and sophisticated, user-friendly analytical methods accessible to researchers in various fields.Furthermore, integrating various data types, including temporal and spatial data from both calcium imaging and voltage imaging, as well as genomics or proteomics, needs tools for advanced data integration and visualization (Csillag et al., 2023).
Another critical challenge is the standardization and sharing of big data in imaging.The lack of uniform formats for data storage and sharing hampers collaboration and constrains data reusability.Tackling this problem requires a collaborative approach from the scientific community to establish universal data standards and create platforms for open-access data sharing.These efforts would not only promote transparency and reproducibility in research but also encourage collaborative studies that can leverage the advantage of big data.
In conclusion, the integration of imaging technologies with big data analysis represents a significant advancement in the field of neuroscience.The ongoing development of sophisticated fluorescent biosensors and the improvement of data analysis algorithms are essential to deepen the understanding of mechanisms of higher brain functions.These concerted efforts are important to unravel complex physiological processes and contribute significantly to the development of novel therapeutic strategies for neurological and psychiatric disorders.

Declaration of Competing Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.a : Decay kinetics were calculated using a single-exponential function.

Fig. 1 .
Fig. 1.Genetically encoded calcium indicators.(A) FRET-based GECI design.In the absence of Ca 2+ , the emission is primarily from CFP.After binding Ca 2+ , intramolecular conformational changes lead to a reduction of the spatial distance between CFP (donor) and YFP (acceptor).This enables Förster resonance energy transfer (FRET), resulting in a decrease in CFP fluorescence and an increase in YFP fluorescence.(B) Single-fluorophore GECI design.The Ca 2+ binding induces conformational intramolecular changes, leading to an increase in the emitted fluorescence.(C) Chemigenetic GECIs design.HaloTag domain is used to bind fluorescent synthetic dyes.After binding Ca 2+ , conformational changes result in an increase in the emitted fluorescence.

Fig. 2 .
Fig. 2. Genetically encoded voltage indicators.(A) Voltage-sensitive domain (VSD)-based GEVIs.Voltage-dependent movement of a transmembrane S4 helix perturbs the protonation state of a fluorescent protein (cpGFP), resulting in changes in fluorescence emission.(B) Voltage sensing mechanism of microbial rhodopsinbased GEVIs.This type of sensor reports voltage changes through the fluorescence intensity changes of the retinal chromophore caused by protonation of the Schiff base in the photointermediate state.(C) Voltage sensing mechanism of microbial eFRET-based GEVIs (Ace2N-mNeon).At a depolarized stage, the Schiff base of microbial rhodopsin is protonated, and the absorbance of rhodopsin changes.This absorption quenches the fluorescence of the appended fluorescent proteins.(D) Chemigenetic GEVIs.The HaloTag domain is used to bind fluorescent synthetic dyes.Upon the depolarization, the Schiff base of microbial rhodopsin is protonated, leading to a change in the rhodopsin's absorbance.This absorption quenches the fluorescence of the appended bright fluorophores.

Table 1
Comparative performance of GECIs.
a : ΔF/F and kinetics in response to single action potentials were measured in cultured neurons.The dissociation constant (K d ) and Hill coefficient were determined from purified proteins.b: ΔF/F and kinetics in response to single action potentials were measured in living mice.K d and Hill coefficient were determined from HEK293T cell lysate (jRFECO1a

Table 2
Comparative performance of VSD-based GEVIs.

Table 3
Comparative performance of opsin-based GEVIs.