Bioelectrical controls of morphogenesis: from ancient mechanisms of cell coordination to biomedical opportunities

Abstract

Cell-to-cell communication is a cornerstone of multicellular existence. The ancient mechanism of sharing information between cells using the conductance of ions across cell membranes and the propagation of electrical signals through tissue space is a powerful means of efficiently controlling cell decisions and behaviors. Our understanding of how cells use changes in “bioelectrical” signals to elicit systems-level responses has dramatically improved in recent years. We are now in a position to not just describe these changes, but to also predictively alter them to learn more about their importance for developmental biology and regenerative medicine. Recent work is helping researchers construct a more integrative view of how these simple controls can orchestrate downstream changes in protein signaling pathways and gene regulatory networks. In this review, we highlight experiments and analyses that have led to new insights in bioelectrical controls, specifically as key modulators of complex pattern formation and tissue regeneration. We also discuss opportunities for the development of new therapeutic approaches in regenerative medicine applications by exploiting this fundamental biological phenomenon.

Introduction

Each and every cell—and all types, including archaeal, prokaryotic, and eukaryotic cells— can be described in terms of the differences in concentrations of particular ions in the space inside the cell. The overall or “net” balance of the electrochemical gradients together gives an individual cell its resting membrane potential– an important parameter that regulates a variety of cellular mechanisms (Fig. 1). Differences in ionic concentrations can also exist within individual cells or across organelle membranes as well. While this kind of descriptor is most widely applied to neurons and the propagation of bioelectric activity through neural networks (i.e., action potentials), all cells form such networks. Electrical synapses mediated by protein complexes known as gap junctions enable the sharing of a cell’s bioelectric state with neighboring cells. Gap junctions are gated by various physiological processes, allowing organs and tissues to establish rich patterns of regionalization and to dynamically change the topology of these networks in vivo (Fig. 2). “Bioelectricity” is a mode of information transfer arising from changes in membrane potentials across individual cells or fields of cells. This mode of signaling is slower than neural spiking. It integrates with chemical gradients and physical forces to regulate developmental boundaries, gene expression, and organ size and identity. Recent advances have revealed how bioelectrical communication helps to orchestrate cell activity toward self-assembly and repair of specific complex anatomies.

Figure 1. Schematic of bioelectric signaling at a cellular level.

A cell’s V_mem is a function of the various ionic concentrations inside its membrane and in the extracellular space, the open/closed state of the various ion channels (transmitting sodium, potassium, chloride, and calcium, for example), and the bioelectrical state of adjacent cells coupled by gap junctions. Ion channels (such as Nav1.5, Kir2.1, and GlyR, depicted in orange, pink, and green) and gap junctions (such as Connexin43, depicted in blue) are often themselves voltage-sensitive, allowing sophisticated physiological feedback loops that regulate bioelectrical state and implement context-sensitive responses and hysteresis. Ions are depicted as circles. Changes in V_mem are transduced by a number of known mechanisms into the activities of downstream effectors such as calcium, neurotransmitters, or butyrate. These in turn influence events inside the nucleus such as gene transcription as well as cytoskeletal remodeling and physical properties of cells. Genes, such as those that control cellular proliferation, cell migration, and cell adhesion, amongst others, can either be up- or down-regulated in response to these cues. As channels and gap junctions are themselves proteins encoded by genes, their expression is also subject to bioelectric control in a feedback loop. Differences in V_mem between cells in a tissue enable facilitated transport of charged molecules, tying these bioelectric circuits to the dynamics of spatial gradients of morphogens.

Figure 2. Types of tissue and cell behaviors used in regenerative systems that may be guided by bioelectric approaches.

A generic morphology is shown. (1) Wound closure. Wound closure involves complex multi-cellular behaviors, especially the migration of epidermal cells across the cut surface, which may be considered a type of collective cell migration and therefore subject to bioelectric controls already shown operational in development. (2) Activation of progenitor cells and their migration to the wound site. Bioelectric cues can direct cells to re-enter the cell cycle, an often-critical step in converting quiescent stem and progenitor cells into actively proliferating cells (pointy green cells). Some progenitor cells are cued to migrate (curvy arrows) to the site of injury, often nestling beneath the wound epidermis (blue) in the example shown. (3) Progenitor cell may coalescence into a defined structure. In appendage regeneration, this is called a “blastema.” Coalesced progenitor cells may be kept in relatively undifferentiated states during this step. (4) Proliferation of coalesced progenitor cells. This step expands the pool of substrate cells in preparation for differentiation, patterning, and growth. Bioelectric cues may promote the continued proliferation of cells and may also prepare the tissue field for morphogenesis by interacting with signaling pathways and genetic regulatory networks. (5) Patterning gradually reveals the correct anatomical structure, and differentiation of various tissues leads to physiological replacement. Scaling the regenerated structure to the body may result in the control of cell migration and apoptosis by bioelectrical signaling, as discussed in the text for fish fin regeneration.

New tools are enabling an unprecedented view into the bioelectrical lives of cells

For any new phenomenon, a set of tools with which to study it is required to transcend the terrain of phenomenology and enter explanatory mechanism. The toolkit that now exists to study bioelectrical phenomena in living cells, and in many model organisms, has been dramatically improved in recent years. Understanding a process requires that it be describable as precisely as possible, both spatially and temporally. Reporter systems for illuminating the bioelectrical status of cells are integral to this essential descriptive work (Supplemental Table 1). These tools have been applied to a variety of cell types and model organisms. For example, CaMPARI, a sensor of calcium – a common downstream effector of bioelectrical state change - has been engineered into species as diverse as flies, zebrafish, and mice [1]. When cells expressing this CaMPARI are exposed to a specific wavelength of light, the sensor is permanently converted from one color to another. This conversion event that only occurs when intracellular calcium levels quickly change at the same time [1]. Likewise, the development of new dyes and genetically-expressed voltage reporter fluorescent proteins [2] are revealing the conversations occurring among cells during embryogenesis, regeneration, and cancer suppression.

The application of optogenetic technologies is now allowing researchers to change membrane potentials in cells with unprecedented spatio-temporal precision and analyze the outcomes. While this is most commonly used to probe CNS activity, recent work has used light-activated ion channels and pumps to trigger regeneration of tails under non-regenerative conditions [3], normalize KRAS-induced tumors [4], modulate pigmentation [5], and create synthetic excitable tissues out of non-neural cells [6].

Dependency of cellular behaviors—in isolation and collectively—upon bioelectric state

Many cells have now been demonstrated to elaborate specific processes or to perform specific tasks in response to asymmetric electrical cues. An outstanding example comes from budding yeast: application of an exogenous electrical field to cells causes individual yeast to form mating projections at the site of highest hyperpolarization, while the site of most depolarization is correlated with bud formation [7]. Changes in the cellular membrane potential (V_mem) imparted by various expression levels of Kv1.2, a voltage-gated potassium channel, can drive key aspects of cell motility via direct interaction with cortactin in fibroblast-like COS-7 cells [8]. Directionality of human keratinocyte migration in vitro can be driven by exogenous electric fields, and this response occurs independently from the canonical actin regulatory modules such as Cdc42/Rac and Arp2/3 [9]. Instead, the behavior is mediated by extracellular pH and G-protein coupled receptors [9]. These two studies are examples that demonstrate that bioelectrical effector pathways can be similar to, or distinct from, those most commonly implicated in well-studied cellular behaviors such as protrusion and migration. Therefore, while it is important to link bioelectrical signaling to known effector mechanisms, defining possibly novel mechanisms is also essential.

Skin patterning relies on bioelectric signaling. Elongation of chick feathers requires collective migration of precursor mesenchymal cells, and this migration is driven by calcium oscillations created and propagated by gap junctions and calcium channels [10]. Upstream of ion channel expression are canonical morphogens such as Sonic hedgehog and WNTs [10]. Patterning of feathers across the epidermis, including the spacing between feather buds and the direction of bud outgrowth, is also connected to these same bioelectric modulators [10]. Collective migratory behaviors underlie key aspects of tissue regeneration. One instance is the early wound healing responses in which epithelial cells crawl across the injury site to seal it. In fly embryos, this event has now been shown to be dependent upon V-ATPase activity [11] – an ancient ion pump known to also be involved in vertebrate regeneration and left-right patterning [12,13]. Human skin cells pre-treated with electrical stimulation and then transplanted into nude mice facilitated improved wound healing, offering potential clues for future therapies [14]. Intriguingly, wounds in corneas from diabetic patients show lower endogenous electrical activity, correlated with their poorer healing outcomes [15], supporting the idea that bioelectricity is needed for efficient wound healing in the clinic.

Beyond orchestrating shape changes and cell movements, bioelectricity also controls key aspects of cellular proliferation and differentiation behaviors. These are both essential processes in regeneration. Modulation of V_mem via the KCNK10 potassium channel promotes the mitotic clonal expansion of progenitor cells, which is required for normal adipocyte differentiation [16]. In a neuron-glia co-culture context, depolarizing astrocytes can promote the differentiation of human mesenchymal stem cells into neurons [17]. Two recent studies examining cellular differentiation in the mammalian skeleton have demonstrated the importance of ion channels. Both osteoblasts and chondrocytes require the activity of the potassium channel Kir2.1 for proper differentiation [18]. This activity specifically promotes bone mineralization [19]. Osteoblast differentiation and bone mineralization defects are hallmarks of the human disorder Andersen-Tawil’s syndrome (ATS). Exogenous expression of wild-type Kir2.1 is sufficient to rescue these phenotypes in cultured patient cells [19]. Morphology is also dependent on bioelectricity in ATS patients (see below). Known channelopathies, showing the dependence of developmental morphogenesis on the function of a range of ion channels and gap junctions in numerous species are shown in Supplemental tables 2,3. Additionally, another study showed that electrical stimuli can directly promote the osteogenic differentiation of cultured human mesenchymal stem cells [20]. These examples—which are from diverse tissues and cell types—indicate that bioelectricity controls fundamental aspects of cell differentiation states. They also provide potentially tunable targets for controlling cellular decisions for therapeutic applications.

Bioelectric mechanisms for determining nerve repair and connectivity

After nerve injury, complex cellular events (both intrinsic and extrinsic to neurons) dictate repair outcomes. In order for axons to traverse a lesion, they must first elaborate new growth cones. This behavior has recently been shown to require transient Kv3.4 potassium channel expression in both chick and rat [21]. In these contexts, Kv3.4 functions to keep calcium ion influx, which locally lowers excitability and promotes outgrowth [21]. Optogenetic tools, such as channel rhodopsin expression and stimulation, have recently been shown to be an effective means of directing neurite outgrowth in dorsal root ganglia cultures, and these effects are mediated by known growth factors [22]. As proteins responsible for mediating bioelectric changes are often ancient and well-conserved, conserved roles for their functions highlight core regulatory pathways for driving operations such as axon regrowth across divergent species. Following spinal cord resection in the highly-regenerative axolotl, membrane depolarization in key glial cells is observed almost immediately. This event has been linked to activation of the immediate early response gene cFos [23]. Intriguingly, this depolarization must be transient because prolonging it with ivermectin inhibits spinal cord regeneration, possibly by inhibiting proliferation of glial cells [23]. The microRNA miR125b is required for establishing a permissive environment in which spinal axons are coaxed to regenerate across a lesion in axolotl [24]. Similarly, inducing miR125b expression in a rat model improved axon regeneration [24]. Intriguingly, here one of the downstream targets of miR125b was Sema4B [24]. Investigating whether sodium channels might also be a critical effector targets will be important because work in non-neuronal cells has pinpointed the mechanism of miR125b’s regulation to be through the sodium channel epithelial 1 alpha subunit transcript encoding SCNN1A [25].

Connectivity patterns and neural physiology are also subject to bioelectric control. Ivermectin-induced depolarization caused an increase in the complexity of neural connections in vitro [26]. Interestingly, in vivo, the connectivity pattern of the posterior peripheral nervous system during embryogenesis receives input from the nascent brain [27]. Brainless Xenopus develop with highly abnormal neural and muscle patterning. These defects can be partially rescued by misexpression of the HCN2 channel, revealing the importance of ion channel-mediated signaling for propagating patterning information across distances.

To properly interface with the existing body, a replacement part will need to connect its nervous system. This reconnection process might be most achievable by re-activation of axon outgrowth and guidance programs involved in initial development. One example of how a novel sensory organ can become innervated by the body is highlighted in the outcome of transplanting embryonic eye tissue to tails of more mature Xenopus tadpoles. The transplanted eye primordia develop into normal eyes. Applying experimental membrane depolarization (using pharmacological agents or ion channel misexpression) in nearby host cells instigated profound innervation from the graft into the host tissue [28]. The mechanism operates through serotonergic signaling that is guided by the voltage patterns. Voltage-guided serotonergic signaling has been shown to be a tractable control point for improving functional innervation of ectopic eyes to enable vision [29]. As regenerative medicine matures, the ultimate goal in the nervous system will become not just neuronal replacement, but functional integration. Hence, these types of insights into the role that bioelectricity can play in this process stand to inform therapeutic efforts aimed at restoring the precise connectivity of the pre-injury state.

Bioelectricity as an innate patterning mechanism for achieving complex developmental morphologies in development and regeneration

The last five years have seen an explosion in the attribution of developmental pattern defects to mutations in specific proteins that modulate bioelectric state, such as ion channels, gap junctions, and pumps. These findings have spanned the phylogenetic spectrum. In zebrafish, several researchers have identified a requirement for ion channel function in patterning fin outgrowth, both in development and in regeneration. For example, mutations in the potassium channel KCNK5B lead to fin overgrowth disproportionate with body size in both developing and regenerating contexts [30], and diminishing expression of V-ATPase causes inhibition of fin regeneration in adult zebrafish [31]. Inhibiting the calcium-dependent phosphatase calcineurin also alters the size of regenerated zebrafish fins, possibly by resetting their proximo-distal coordinates within the progenitor cells of the blastema [32]. Loss-of-function studies in mice have also identified roles for ion channels in mammalian embryogenesis, such as the activity of KCNJ13 potassium channels in establishment of cell alignment and polarity that is required for normal tracheal tubulogenesis [33]. As at least 20 different ion channel-encoding genes are differentially expressed along the proximo-distal axis in zebrafish tail fin [34]. This work provides an important opportunity to perhaps uncover many more functions for these V_mem regulators in establishing proper morphology.

Ion channels and gap junctions exhibit high degrees of compensation and redundancy, because the signaling is not a function of individual channel genes but the overall bioelectric state. For this reason, single gene knockout screens under-represent the roles of the electrome in development. Nevertheless, in addition to the zebrafish model, the classic genetics model Drosophila has recently been used to reveal a role for numerous ion channels in regulating wing morphogenesis [35,36].

In humans, several diseases and syndromes with characteristic phenotypic morphologies have now been mapped to proteins that modulate bioelectric state. The same study that implicated mutations in KCNH1 with epilepsy also showed affected individuals have finger shortening and characteristic facial features [37]. Similarly, mutations in KCNJ6 potassium channel subunit are causative for another syndrome, Keppen-Lewbinsky, with a separate set of facial characteristics [38]. These recent human studies highlight that patterning of the face, hands, and brain is exquisitely sensitive to bioelectrical controls during development. Recent work in the frog embryo model revealed why Kir2.1 mutations result in the distinctive craniofacial malformations that characterize ATS. In addition to bioelectricity influencing differentiation of osteoblasts, a prepattern of bioelectric regionalizations in the nascent face ectoderm is required for the normal demarcation of gene expression domains and subsequent morphogenesis [39]. ATS is just one of many channelopathies with structural defects [40–43]. Together these findings reveal that bioelectric states in the human embryo are an important part of normal development. Therefore, the data suggest that caution should be used when considering the use of drugs that target ion channels, or downstream neurotransmitter signaling, during pregnancy [44].

Roles for bioelectrical signaling in modulating patterning through regulation of canonical and non-canonical signaling systems in well-studied developmental contexts are emerging. For example, the developing vertebrate spinal cord has for several decades been a very rich model for understanding how patterning—particularly cell fate specification—is dependent upon morphogen gradients. A key outstanding question had been how Sonic hedgehog signaling is attenuated in receiving cells after their motor axon fates have been specified but the morphogen signal remains. Nascent calcium-dependent electrical activity in the developing spinal cord was temporally and spatially linked to downregulation of Shh intracellular pathway effectors [45]. Recently, the mechanism for this coupling was discovered: Shh-instigated calcium spiking in new spinal cord neurons causes an increase in protein kinase A (PKA) activity. Increased PKA activity leads to three separate changes, all of which have the effect of attenuating Shh signaling via decreased transcription of gli1: increase in Gli3^R (repressor) accessing the nucleus, increase in CREB phosphorylation, and translocation of Gli2 (activator) out of the nucleus [46]. The opposite kind of regulation has also been shown: FGF13 and FGF14 work to localize and maintain voltage-gated sodium channels within different compartments of neurons [47]. Whether these types of regulation exist in non-neuronal cells is currently unknown. Figure 3 illustrates one class of mechanisms by which bioelectric gradients arise and are converted into changes in gene expression that impact organogenesis. These themes – the use of bioelectric gradients to drive distribution of morphogens, the linkage of bioelectric cues with cytoskeletal organization, and the amplification of subcellular symmetry breaking into large-scale pattern by bioelectric circuits are echoed by numerous neural and non-neural systems, from plants to vertebrate embryos [48].

Figure 3. Bioelectric mechanisms function in left-right patterning: Xenopus embryonic laterality illustrates how bioelectric signaling links to genetic pathways.

The invariant left-right (LR) asymmetry of heart, viscera, and brains of bilaterian animals requires a robust mechanism for amplifying chiral subcellular structures into distinct patterns of gene expression on the left and right sides. (A) In Xenopus, differences in resting potential between the L and R blastomeres during cleavage result in an electrophoretic force that creates a right-ward gradient of small molecules, including serotonin. (B) The direction of the voltage difference is established by a chiral maternal cytoskeletal element that is fixed with respect to the dorso-ventral and animal-vegetal axes, allowing it to invariantly nucleate transport of four specific ion channels and pumps to the right blastomere during the first 2 cleavages. The resulting directed transport of small signaling molecules through long-range gap junction-coupled paths enables the coordination between the two sides of the body and aligns the morphogen of the LR axis with respect to the other two axes. Transduction into gene expression occurs when the serotonin in right-side cells binds to an intracellular receptor (MAD3 and HDAC1) and thus ensures that the left-sided gene Nodal (XNR-1) is only expressed on the left side. A subsequent cascade of differential gene expression ensures asymmetric development of the heart, gut, and other internal organs. Aspects of this scheme are conserved to the patterning of the left-right axis in numerous other taxa (reviewed in [54,55]). Beyond asymmetry, the same themes of using bioelectrics toamplify and propagate cytoskeletal and other subcellular information to large cell fields are found throughout development and regeneration.

Unmasking latent morphological potential in regenerating systems and correcting developmental errors using bioelectric stimuli

Experimental modulation of bioelectrical signaling pathways has led to an unexpected result that challenge views of how anatomical information is stored across lifespan and generations. Treatment of Dugesia japonica planarians with the gap junction inhibitor octanol for only the immediate 48 hours after head and tail amputation resulted in production of two heads in 25% of the regenerated worms [49]. These animals continued to regenerate as two-headed in subsequent rounds of amputation in plain water, in perpetuity, despite complete washout of octanol. Surprisingly, it turned out that the remaining worms that regenerated a normal one-headed anatomy were not in fact wild-type escapees from the treatment, but represent a novel “cryptic” phenotype: their subsequent amputation again resulted in 25% double-head, the remaining ones maintaining this destabilized phenotype [49]. This experiment demonstrates that transient physiological signaling, not requiring genomic editing, can permanently change the encoded pattern to which animals regenerate upon damage. It also reveals these treatments can produce a unique bi-stable stochastic target morphology. Interestingly, each piece cut from a single cryptic worm makes an independent decision about the possible head number outcomes. This model provides novel opportunities to understand how invariant pattern outcomes are normally achieved by genetic and physiological networks and about the role of noise, stochasticity, and pattern memory in this “epigenetic” process. It should also be noted that these permanent “lines” of 2-headed and destabilized planaria are the only available “mutants” in this model species which until now has offered no natural or induced strains of abnormal morphology as are available in every other model system.

Modulating bioelectrical signaling has recently been exploited as a corrective measure in frog tadpoles whose brain morphologies and learning capacities have been negatively impacted by nicotine exposure. Expression of HCN2 ion channel in these animals [50] can largely rescue the gene expression, anatomical defects, and behavior by forcing the correct bioelectrical prepattern responsible for pattern and size control in the brain [51]. Interestingly, the same strategy can rescue brain defects induced by a dominant negative mutation in Notch, a critical neurogenesis gene. These results show that intervention at the physiological control level could be an effective strategy to address even some genetic defects. Importantly, the repair strategy was predicted by a computational model of bioelectric brain patterning [50]. The success of predictive modeling here illustrates the field is progressing to the point that quantitative mechanistic models can be used to infer specific interventions that, as predicted, induce complex organ repair in vivo.

Outlook: regenerative medicine applications

The ultimate objective of regeneration is the perfect replacement of lost cells and tissues. Our increasing understanding of the rules that govern natural regeneration—as well as its component processes—and how they are bound by bioelectrical states offers critical insights that may be harnessed for future therapies. The understanding of bioelectric circuits has now reached the point that bio-realistic computational models can be interrogated for specific interventions that induce large-scale organ repair [50]. Work in model systems exploits the precision of ion channel misexpression. Yet, the plethora of existing ion channel drugs, many of them already approved for human use, offers the possibility of exploiting small molecule ion channel-modifying compounds as “electroceuticals” [52]. Proof-of-concept for very brief application of pharmaceuticals to an injury site leading to profound improvement in regeneration has been obtained in adult frogs. Amputated Xenopus hindlimbs typically only regenerate a cartilaginous spike covered in skin. However, treating them with progesterone (a drug that is known to induce specific bioelectric tissue states) loaded into a small device worn at the amputation site for just 24 hours can result in markedly better regeneration. Treated limbs were larger limbs and composed of substantially more internal tissues such as nerves and muscle. Remarkably, these limbs often achieve patterns approaching unamputated limbs and dramatically more complex than controls [53]. Future experimentation in this vein with known electroceuticals could prove to be a simple and effective pro-regenerative strategy. Another especially important area for future research is the development of machine learning strategies to help crack the bioelectric code – the mapping of V_mem prepatterns to tissue-and organ-level outcomes.

These future therapeutic approaches may initially be aimed at modulating bioelectric signals to instruct cells to activate in response to injuries. This activation could be either cell cycle re-entry for internal progenitor cells or re-epithelization of the skin across a wound site. Bioelectric signals may also be used to specify polarity and to localize responses after initial activation; for example, progenitor cells in an amputated limb stump need to migrate toward the wound site, and, if they can be commanded to do so, perhaps other key events will follow. Once activated substrate cells have coalesced, providing bioelectric cues to the field may be a means of enhancing cellular proliferation and of instructing key patterning events, such as establishing anatomical axes. Eventual differentiation of the various cell types could be enhanced or specified by bioelectrical therapies. Finally, the clues gained from in vivo studies of appendage regeneration predict that therapeutically-stimulated patient regeneration could result in precise replacement size and shape if electric fields across these regenerating tissues are correctly programmed.