Bioelectricity in Developmental Patterning and Size Control: Evidence and Genetically Encoded Tools in the Zebrafish Model

However, with the advance of technical developments, these shortcomings are being overcome, and GEVI applications are expanding beyond neuroscience into many fields, such as developmental biology.

Thus far, numerous advancements and variations of GEVIs have been developed (Table 3). These GEVIs usually fall into one of three categories (Figure 2A) [135,162,163]: (1). GEVIs based on a voltage-sensitive domain (VSD) within the cell membrane, usually from the tunicate (Ciona intestinalis) voltage-sensitive phosphatase, PTPE [164,165]. The VSD can be linked with either a single fluorescent protein (FP), dual FPs for FRET (Forester resonance energy transfer) signaling, or even bioluminescence. (2). Opsin-based GEVIs, with and without additionally combined FPs to improve brightness. (3). There is a group of hybrid GEVIs that combine these different components with the addition of brighter and more photostable synthetic dyes [135,162,163]. Each GEVI has its unique properties and application niche. Many of these GEVIs have also been examined and utilized in zebrafish research (Table 3).

One commonly used GEVI is ASAP (accelerated sensor of action potentials) based on VSD design. A circular permutated GFP is inserted int the middle of S3–S4 loop of the VSD. Thus, when the VSD protein confirmation is altered by Vm, the intensity of GPF fluorescence will change correspondingly (Figure 2A). When the cell membrane is hyperpolarized, the ASAP1 fluorescence signal is brighter. This ASAP1 was successful, and neuron bioelectricity has been well-documented in many model organisms, including zebrafish. A few updated versions have also been developed to improve its speed, signal-to-noise ratio, and sensitivity [166,167,168,169]. Our lab has generated a ubiquitous transgenic reporter zebrafish line Tg(ubi: ASAP1) [170,171]. With this ASAP1 transgenic fish line, real-time endogenous cellular bioelectric activities can be visualized in fish embryos, larvae, adults, and even tumor tissues [170]. Our results are consistent with an independent report on Tg(UAS: ASAP1), a binary transgenic fish line for tracking larval fish neuronal circuitry within cerebellum, optic tectum and spinal cord [172,173]. In addition, we made new observations in early fish embryos. We found that a transient local membrane depolarization occurs before cleavage furrow formation during zebrafish embryo cleavage stages (1–64 cells). This phenomenon is consistent with calcium signaling measured by GCaMP6s [153]. These Vm changes are not static, but dynamic during the cell division period. Moreover, these Vm dynamic changes are not perfectly synchronized among early cells. These results suggest a biological function of Vm in cell division. Membrane potential changes have been shown to influence the organization of phospholipids. These are known as critical components of the cleavage furrow and cytokinesis [36,174,175].

Once zebrafish embryos enter the blastula stage, the bioelectric signals become whole-cell transient hyperpolarizations, mainly found in the rapidly dividing superficial layers of the blastula (EVL) and yolk syncytial layer (YSL). During gastrulation, Vm transients continued in the EVL and YSL, and started to occur in the deeper cells. Moreover, we noticed differential Vm among different embryonic tissues and somite-specific hyperpolarization events during the zebrafish embryonic segmentation period. These results demonstrated that the ASAP family and potentially other GEVIs could be readily used for measuring embryonic bioelectricity. We expect more bioelectric biology to be revealed by organ- and/or tissue-specific zebrafish transgenic fish lines. For example, different cell types in the zebrafish fins, pigment cells, and the other cell types in the skin can be characterized for their physiological bioelectric properties (for details, see the prospective section).

It is worth noting that GEVIs are not limited to zebrafish. They have also been applied to other organisms. For example, multiple studies have shown the utility of genetically encoded indicators in fruit flies, but have, thus far, only been focused on neuronal-related studies [167,169,176,177,178]. In mice, fewer research studies have utilized GEVIs in vivo, and most of these studies focused on neurological research [178,179,180,181,182]. In addition, Xenopus oocytes were used to characterize Arclight, but did not address any developmental biology [183]. Nevertheless, their results showed that these sensors could be employed to characterize bioelectricity in Xenopus if needed.

Another requirement to elucidate the bioelectric signaling mystery is the direct and specific perturbation of the normal electrical state of cells and tissues. As this was a major task for neuroscience, optogenetic and chemogenetic tools were already developed as experimental approaches (Figure 2B and Table 4). These tools have, thus far, demonstrated the capability to alter cell-specific electrical states of neurons to hyperpolarization and depolarization, allowing a precise level of control in various organisms [137,138,139,221,222,223].

Optogenetics: Optogenetics modulates bioelectricity through microbial (type I) opsins, which are light-sensitive ion pumps and ion channels found in prokaryotic and eukaryotic microbial organisms [224]. These type I opsins can conduct cations or anions under the control of different wavelengths of light [225,226,227]. When exposed to specific wavelengths of light on expressing cells, channels open to allow specific ions, such as H+, Na+, Ca2+, or Cl−, into cells, resulting in increased or decreased Vm. The use of optogenetics in zebrafish has been primarily targeted in neuroscience studies. These have been used to modify swimming behavior [228] and locomotion behavior [229], perturb hair cell sensory receptors [230], axon guidance control [231], and alter olfactory responses [232]. Optogenetic tools have also been used for other zebrafish research fields, such as heart physiology [233,234], and melanophore patterning [235]. In the zebrafish melanocyte study, ChR2 was expressed in the melanophores of zebrafish that were then placed in tanks exposed to blue light to stimulate depolarization [235]. As a result, these transgenic fish began to lose the boundaries of their standard stripe patterns [235]. Interestingly, this was partially reversed after allowing the depolarized cells to return to their average membrane potential, suggesting that endogenous bioelectric signals are essential for maintaining pigment homeostasis. This study also provided direct evidence of bioelectric signals in zebrafish pigment patterning. Recently, a set of optogenetic transgenic zebrafish lines have been created under the control of the UAS promoter [141]. This will accelerate bioelectric research in zebrafish.

Moreover, with the great success of light-sensitive rhodopsin, new optogenetic tools have been invented with which to control gene expression, protein localization, and activity, using a new set of light-sensitive proteins such as phytochromes, blue light using flavin (BLUF) domain photoactive proteins, cryptochromes (e.g., CRY2-CIB1), and light oxygen voltage (LOV) domain proteins. Their applications to developmental biology and zebrafish have already been reviewed [236,237]. These new optogenetic tools are also utilized for investigating zebrafish bioelectricity.

Chemogenetics: Chemogenetics is a genetic approach to perturb cellular electrical activity using synthetic molecules through either mutated G-protein-coupled receptors (GPCRs) or ligand-gated ion channels that no longer function normally, but only in the presence of inert molecules [222,223].

DREADDs (Designer receptors exclusively activated by designer drugs) are one of the mainstream chemogenetic tools that are commonly used in neuroscience, including in the study of behavior, circuits, and diseases [222,238,239,240]. DREADDs are designed based on mutated muscarinic and opioid receptors. Four types (hM3DGq, hM4DGi, hM3DGs, and KORD) alter cellular Vm through downstream signaling changes, such as the arrestin pathway, intracellular Ca2+, and cAMPs that indirectly modulate ion channels. The original agonist is clozapine-N-oxide (CNO), a derived metabolite of clozapine, which is used as an antipsychotic. Thereafter, it was found that CNO can be converted into clozapine and cause psychoactive side effects in murine models [241]. Subsequently, more potent and specific agonists were invented, including JHU37152, JHU37160, and deschloroclozapine (DCZ) [242,243]. In addition to the choice of agonist, the diverse signaling pathways downstream of receptors make electricity manipulation less straightforward. Furthermore, the biological reactions could be variable in different cell types. Recent progress on DREADD structure activation is helpful for us to understand the mechanisms of their activation [244]. Still, careful experimental design was proposed to overcome the two potential issues in animal models [245]. Despite these weaknesses, the DREADD tools have found their way into various animal models, including flies, mice, rats, and primates [246]. In addition, DREADDs have also been extended to other research fields, including diabetes and endocrinology [247,248]. However, they have not been implemented in zebrafish yet. One study tried to utilize DREADD in zebrafish by microinjections, but failed in the endeavor [249].

Instead, they demonstrated that transient receptor potential (TRP) channels worked in the zebrafish embryos, and successfully manipulated Rohon–Beard and trigeminal sensory neurons using islet-1 enhancer-driven TRP channels [249]. TRPV1 was activated by capsaicin, TRPM8 was activated by adding menthol, while TRPA1 activity required temperatures above 28 °C. Activation of TRPs induced dose-dependent locomotion and ablation, and altered wake-sleep behaviors [249]. In addition, the TRPV1-based approach was verified in another zebrafish study for modulating calcium flux in neutrophil [150]. Our laboratory has applied one of the DREADD, hM4DGi, to zebrafish embryos and larvae, and was able to change melanophore pigment cell dispersion [250]. Although the voltage reporter efficiency still needs to be improved, our results suggest that DREADD can be applied to zebrafish bioelectricity research.

PSAMs (pharmacologically selective actuator modules) are another set of chemogenetic tools. PSAMs share a similar key-and-lock concept, utilizing artificial inert molecules as agonists. For the actuator, instead of using mutated GPCRs as a lock, PSAMs use ligand-gated ion channels, such as nicotinic receptors (nAChR), serotonin receptor 3 (5HT3), GABA receptors, and the glycine receptor (GlyR). Each PSAM has a ligand binding domain (LBD) and a channel. These are mutated ligand-gated ion channels that can only be activated by PSEMs (pharmacologically selective effector molecules) [251,252]. Recently, a ligand-binding domain was genetically engineered, and more potent agonists were identified. Thus, this ultrapotent system is helpful for research and suitable for therapeutic applications, as exampled in mice and monkeys [251]. The first generation PSAM-GlyR was expressed in zebrafish horizontal cells (HCs), which connect rod and cone photoreceptors via synapses [253]. Disrupting Vm of HCs resulted in altered light response and lateral inhibition in retinal ganglion cells. This study illustrated that the ultrapotent PSAM-PSEM system could be extended for zebrafish bioelectric research.

In addition to the chemogenetic and optogenetic tools mentioned above, direct genetic modification, i.e., adding or deleting an ion channel regulator, is also achievable with the established tol2-transposon transgenic system and CRISPR technology [254]. This has already been demonstrated, such as in the transgenic Tg(−5.4k-pax3a: kcnj13-IRES-EGFP), where transient ectopic expression of kcnj13 in zebrafish dermomyotome causes a long-finned phenotype in adults [112]. Nevertheless, this approach depends on already-known ion channel regulators and an available tissue-specific promoter. In addition, this genetic modification cannot be turned on and off, as with chemogenetic and optogenetic tools. Nevertheless, these tractable genetic tools are critical for implanting GEVIs, chemogenetics, and optogenetics into zebrafish.

The above-mentioned zebrafish fin and pigment genetic mutations are indirect evidence of bioelectricity in developmental patterning. Direct bioelectricity research in zebrafish is only possible now because of the recent availabilities of new voltage biosensors and manipulators. Although, here, we use zebrafish fin size and pigment cell patterns as examples, many other research directions can be pursued in this field. Below, we propose four major perspectives.

Though electrical signaling in neuronal tissues has been widely accepted and extensively investigated, other embryonic tissues remain completely unexplored. Characterizing non-neuronal tissues/cellular bioelectric signaling during embryogenesis is essentially the first step for deciphering the roles of bioelectricity. The in vivo real-time systematic characterization of vertebrate embryos has only just begun. Our Tg(ubi:ASAP1) fish line provided the first example of endogenous hyperpolarization signals of embryonic tissues [171]. More tissue-specific fish lines can be created with newly developed, more sensitive GEVIs, such as JEDI-2P, voltron2, Ace-mNeon2, and VARNAM2 using Tol2 transgenesis and CRISPR knock-in [178,182,215]. For example, in order to investigate somite cell contribution to the fins in our Dhi2059 mutant, somite compartments (dermatome, sclerotome, and syndetome) can be labeled with ASAP1 and other GEVIs to investigate the bioelectricity of these embryonic tissues in the future. On the other hand, faster and more sensitive imaging technologies, such as light-sheet fluorescence microscopes, have made this task more achievable [171,267,268]. It is possible that not only cellular Vm, but also its fluctuation amplitude, frequency, and rhythms, can serve as cell signals. Such tissue-specific, cell-type-specific, or subcellular bioelectric imaging in zebrafish embryos will reveal unprecedented insight into the functions of bioelectricity during embryonic development. Moreover, this information will be helpful in testing the theoretical concept of “the bioelectric code” [269,270,271]. In addition to GEVIs, biosensors for ions, such as the potassium sensor GINKO2 [272] and the chloride sensor ClopHensor [273], can also add another layer of information on bioelectricity.

The cellular bioelectricity of a given cell type is composed of many ion regulators, as discussed above. However, major contributors of each cell type at a given embryonic stage also remain uncharacterized. Technically, the major challenge is ion regulator redundancy. Forward genetics starts with mutants created by random mutagenesis, and then addresses which gene is responsible for the mutant phenotype. This approach is powerful for identifying critical genes given a phenotype, especially embryonic development, in multiple model organisms, including flies, worms, mice, and zebrafish [94,274,275,276]. However, this approach might miss the underlying genes when genetic or biochemical redundancy exists. It is already known that multiple ion channels, connexins, pumps, and solute carriers can contribute to the overall cellular bioelectronic properties, though some are more prominent than others, given a cell type. There are 817 channel/transporter proteins known in the human genome in the cell membrane [277]. Many transporters underline many cellular functions, but also create redundancy that could make cellular bioelectric homeostasis very robust (Figure 3) [278]. Thus, it is unsurprising that only a limited number of zebrafish mutants were identified with altered fin and pigment patterns. Thus far, all the known fin-size mutations are genetically dominant. Most of them are either ectopically expressed (kcnj13, kcnh2a) or gain-of-function mutations (kcnk5b, aqp3a, and lat4a) (Table 2). This ismost likely because the loss of one ion regulator is generally insufficient to cause overall cellular bioelectric change. However, overexpression or gained function of ion regulators could drive the cell’s bioelectric property out of its physiological range (Figure 3). Thus, this type of mutant displays a developmental patterning phenotype.

Dominant negative mutations of ion regulators may also function by interrupting structurally similar proteins in the same family. For example, the kcc4a in the schleier mutation could be interpreted as a dominant negative form, since the null mutant has no fin phenotype [108]. This redundancy not only leads to no phenotype, but also makes the experimental interpretation difficult. What is the solution to overcome the redundancy? With single-cell sequencing, it is feasible now to profile all the enriched ion regulators given a cell type [279,280]. For example, abundant ion regulators in different cell types within zebrafish somites and fins can be identified using this technique. Alternatively, whole mount in situ hybridization can be performed systematically, to examine their expression during zebrafish embryogenesis, as was performed in the case of the potassium channel gene subfamilies [281,282]. Once their gene expression is known, cell- and/or tissue-specific multiplexing gene knockout or knockdown by CRISPR can be used to examine certain ion regulators’ function in developmental patterning [283,284,285,286,287]. Moreover, chemogenetic and optogenetic tools are also expected to be effective to override endogenous ion regulators in a treated time window (Figure 3).

Another fundamental question about bioelectricity is how it works during embryogenesis. From the current data of the zebrafish pigment mutants, cell–cell interactions are critical for adult pigment patterning. Moreover, the interaction between the pigment cells and other skin cells could also be important, as suggested by the transplant experiment of the mau fish [110]. One key phenomenon of bioelectric patterning we have learned from the Dhi2059 long-finned mutant is that the adult fin patterning can be determined much earlier, even back to the 1–2 dpf fish embryo stages [112]. The zebrafish larva fin buds have not yet developed at this embryonic stage. Very likely, the transient alteration of bioelectric properties of the two waves of fin progenitor cells, which migrate from somites into the fin buds [288,289], is enough to change adult zebrafish fin size. This memory of bioelectric property change may influence the differentiated cell interaction later in the fin anlagens. This bioelectric memory concept has been proposed and validated in the head–tail body axis determination in flatworms by the Levin research group [290]. Depolarizing edema cells during the first three hours after amputation is enough to cause a double-headed phenotype in flatworms, indicating that bioelectric memory is critical for patterning [291]. Thus, it is also reasonable that the bioelectric memory could be affected in other reported long-finned mutants [105,107,108]. Based on this, we have proposed a two-stage model for fin-size regulation [112]: the bioelectric memory may be formed at the somite stage, before the fin progenitor cells migrate to the fin buds. Furthermore, this memory can guide the local cell interactions and eventually determine the fin shape and size. However, the bioelectric memory can also make the developmental patterning mechanism more elusive.

Generally, there will be a time lag for the phenotype development after bioelectric perturbation. Moreover, the bioelectric memory formation might occur within a short time window. Therefore, careful experimental design, including cell lineage tracking, bioelectricity measurement, and perturbation, is essential to decipher this intriguing mechanism.

Discovering unknown downstream signaling pathways of Vm on different functions could help explain all the diverse functions of bioelectricity. There is still a long way before we know how this biophysical cue is integrated into current biochemistry theory. As neural action potentials can deliver a signal to a long-range destination along the axon and relay by neurotransmitters, the non-excitatory embryonic tissues may use a similar mechanism in an atypical way for guiding cell–cell interactions. Instead, the embryonic cells may form weak bioelectrical networks [8]. Different cells and tissues may dominantly utilize specific pathways.

The morphogen protein gradient is one of the most influential developmental patterning principles with solid experimental support from many organisms, including zebrafish [292,293]. Relationships between bioelectricity and morphogen proteins, and their downstream transcription factors, are particularly interesting (Figure 3). This information will help us better understand how bioelectricity works and its interaction or cross-talks with current already-known morphogen proteins. No clear experimental evidence was reported in zebrafish yet, but multiple morphogen-signaling pathways were directly or indirectly linked with bioelectricity in other organisms. For example, cell membrane resting potential can alter a frog’s brain development through Notch signaling [294]. BMP signaling also mediates morphological changes of Kcnj2/Kir2.1 gene mutations in both flies and mice [295,296,297]. However, the ion channel and WNT signaling relationship are less clear in developmental biology, and there is limited evidence from pathological conditions [298]. The most apparent case is bioelectricity and the hedgehog-signaling pathway. Two transient receptor potential (TRP) channels, PKD1L1 and PKD2L1, modulate ciliary calcium concentration, and the loss of these two channels leads to increased Gli1 activity and subsequent hedgehog signaling [299,300]. Additionally, direct evidence was reported that optogenetic depolarization was found to promote smoothened membrane localization and increase hedgehog signaling, which also promotes cellular depolarization in the fly wing disc [301]. Furthermore, two critical components of this pathway, DISP1 and PTCH1, function as cation-powered transporters. DISP1 utilizes the transmembrane sodium ion gradient to release cholesterylated SHH from HEK293 cells.

In comparison, the PTCH1 receptor behaves as a K+-powered cholesterol transporter, which employs a transmembrane potassium ion gradient to antagonize SMO with cholesterol in NIH-3T3 cells [302]. Another report found the PTCH1 inhibits SMO and depends on extracellular sodium ion concentration [303]. Thus, bioelectricity may modulate multiple steps of hedgehog signaling from SHH secretion (establishment of morphogen gradients) to SHH reception (interpretation of morphogen), indicating bioelectricity patterning may be mediated by morphogen-signaling pathways, at least partially.

Bioelectricity has emerged as a new player in developmental patterning and organ size control. Here, we focused on the zebrafish model system and systematically reviewed genetic evidence of bioelectricity from zebrafish mutants. Additionally, we briefly summarized the newly developed genetically encoded voltage indicators and cellular voltage manipulators (optogenetics and chemogenetics) and their potential to be used for zebrafish bioelectricity research. Finally, we discussed future directions and opportunities for bioelectricity research in developmental patterning.

We thank Zhang lab members Sung Jun Park, Ziyu Dong, and Dingxun Wang for proofreading this manuscript.

Conceptualization, G.Z.; writing—original draft preparation, M.R.S. and G.Z.; writing—review and editing, M.R.S. and G.Z. All authors have read and agreed to the published version of the manuscript.

This research was supported by the National Institute of General Medical Sciences of the National Institutes of Health (2R35GM124913) to G.Z. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agents.