A novel method for inducing nerve growth via modulation of host resting potential: gap junction-mediated and serotonergic signaling mechanisms

The transduction of membrane voltage changes to downstream signaling cascades can occur through a variety of mechanisms [68]. In a suppression screen of known candidate transduction mechanisms (data not shown), we identified 5-HT signaling as a possible candidate for signal transduction in ectopic eye innervation. To test the role that host 5-HT may play in the hyperinnervation process, tadpoles were cultured in media supplemented with 50 μM 5-HT. Under these conditions, 17 % of the ectopic eyes hyperinnervated the hosts, a significantly higher rate than that observed in untreated transplants (n =62; P =0.03) (Fig. 4a, ai). Thus, elevated serotonergic signaling could mimic the effect of ivermectin in a gain-of-function experiment. Further, to assess quantitatively the level to which embryos sequestered extracellular 5-HT, stage 41 embryos were analyzed for 5-HT content. WT embryos contained 309 ±93 pg per animal (n =4) and no significant changes in overall content were observed in response to ivermectin exposure, with an observed value of 276 ±82 pg per animal (n =4). Adding 5-HT to the media revealed a bioaccumulation of 5-HT in embryos, with values exceeding 344 ±146 μg per animal (n =4), confirming the ability of Xenopus cells to take up 5-HT from their surroundings.Fig. 4Membrane voltage control of ectopic eye innervation functions through serotonin (5-HT) signaling. (a) Supplementing the Xenopus media with 5-HT can induce hyperinnervation, even in the absence of membrane voltage alteration. (b) Disrupting 5-HT production in Xenopus embryos through para-chlorophenylalanine (TPH) exposure inhibits ectopic eye innervation of host animals. (c) Inhibition of the 5-HT transporter (SERT) with fluoxetine blocks hyperinnervation in response to membrane depolarization.

(d) Selectively blocking 5-HT receptor 3 (5-HT3) with exposure to tropisetron has no effect on ectopic eye innervation of the host. (e) Inhibition of the 5-HT receptors 1, 2, and 7 (5-HT1, 2, 7)_with the broad-spectrum antagonist metergoline abolishes ectopic eye innervation of the host. (f) Disruption of 5-HT receptor 1A and 1B (5-HT1) with the antagonist cyanopindolol does not inhibit host innervation by ectopic eyes. (g) Inhibiting 5-HT receptor 2A (5-HT2) with the compound altanserin has no effect on ectopic eye innervation of the host. (h) Selectively blocking 5-HT receptor 7 (5-HT7) with exposure to SB258719 does not inhibit ectopic eye innervation of the host. (i) Using a combination of altanserin and cyanopindolol, innervation of hosts by ectopic eyes could be suppressed, indicating a downstream role of 5-HT receptors 1 and 2 (5-HT1/2). IVM = ivermectin

In a loss-of-function experiment, embryos were raised from the 1-cell stage in para-chlorophenylalanine, which binds the enzyme tryptophan hydroxylase and inhibits 5-HT synthesis [69]. Para-chlorophenylalanine exposure resulted in an incomplete but significant 20 % reduction of ectopic eye innervation in the presence of ivermectin (6 of 28; P =0.04), further supporting a role for 5-HT in the observed phenotypes (Fig. 4b,bi). Finally, tadpoles receiving transplants were raised in a combination of ivermectin and the selective 5-HT reuptake inhibitor fluoxetine [70], which blocks the movement of 5-HT through the 5-HT transporter (SERT) and has been extensively tested in the Xenopus system [56, 71–75]. In the presence of fluoxetine, none of the resultant tadpoles were hyperinnervated, completely blocking the effect of ivermectin exposure (n =29; P <0.01) (Fig. 4c,ci). We conclude that the dynamics of serotonergic signaling are involved in mediating the effect of V mem upon ectopic nerve growth, and to flesh out more fully the mechanistic pathway leading from V mem change to hyperinnervation, we next asked how the 5-HT levels might be ascertained by cells.

The presence of extracellular 5-HT can be read by cells through a family of 7 membrane-bound 5-HT receptors [76], and we sought to determine whether specific 5-HT receptor family members were implicated in the hyperinnervation effect [77, 78]. We began by inhibiting the 5-HT3 receptor with the compound tropisetron, and 5-HT1/5-HT2/5-HT7 receptors with the compound metergoline, in eye-transplanted embryos exposed to ivermectin. Tropisetron exposure had no effect on hyperinnervation (Fig. 4d), with rates identical to those of ivermectin exposure alone (12 of 34; P =0.55), but metergoline had a striking antagonistic effect, with none of the treated animals demonstrating hyperinnervation (n =31; P <0.01) (Fig. 4e). The 5-HT1, 5-HT2, and 5-HT7 receptors were then targeted individually with cyanopridolol, altanserin, and SB 258719, respectively. However, none of these agents alone resulted in reduced levels of hyperinnervation (n ≥29; P ≥0.44 in each case) (Fig. 4f–h), leading us to test combinations of these compounds. Combined 5-HT1/5-HT2 inhibition with a combination of altanserin and cyanopridolol resulted in a reduction of hyperinnervation from 37 % to 10 % (n =30; P <0.01), suggesting 5-HT-1,2 receptors as likely downstream sensors of voltage changes within host animals (Fig. 4i).

5-HT can be secreted extracellularly as a signaling molecule during development, but it can also move between cells by passing through gap junctions [79, 80], along voltage gradients (towards more negatively charged cells), as occurs, for example, during left–right patterning [81–85]. Gap junctions also play a role in other examples of bioelectrically controlled morphogenetic decisions [86–91], and are expressed in a wide variety of tissue types in Xenopus, including epidermis, tail, lateral plate mesoderm, and neural tissue [92–94]. To determine if intercellular communication mediated by gap junctions might play a role in ivermectin-induced hyperinnervation of ectopic eyes, we inhibited gap junction channel communication with the blocker lindane [95–97]. This compound was a potent inhibitor of ectopic eye innervation in response to ivermectin (n =32; P <0.01), with <5 % of treated animals hyperinnervating the host (Fig. 5a,ai).Fig. 5Gap junction (GJ) communication is essential for ectopic eye innervation of the host. (a) In the presence of the ivermectin (IVM), hyperinnervation by ectopic eyes could be suppressed by exposure to the GJ antagonist lindane. (b) Similar to chemical exposure, hyperinnervation by ectopic eyes in response to IVM could also be suppressed by early mRNA injections of the dominant negative (DN) GJ H7. (c) Animals treated with both the gap junction inhibitor lindane and supplemented serotonin (5-HT) resulted in hyperinnervation

To validate this result with a molecular reagent, we employed the dominant negative gap junction H7 construct. Injection of this mRNA into embryos results in the formation of nonfunctional connexin chimeras, reducing or abolishing native gap junction communication [60, 98, 99]. Embryos were injected with H7 mRNA at the 1-cell stage, resulting in ubiquitous expression, and cultured in ivermectin following transplantation of eye primordium. When donor tissue was visualized, hyperinnervation was strongly inhibited and levels were indistinguishable from those of untreated controls (3 of 38; P =0.41) (Fig. 5b, bi). These results confirm those of chemical gap junction inhibition, and show that gap junction communication is essential for ectopic eyes to hyperinnervate hosts in response to membrane depolarization.

Finally, we sought to determine whether 5-HT signaling was acting upstream or downstream of gap junction communication. Embryos receiving grafts were cultured in a combination of the gap junction inhibitor lindane and 50 μM 5-HT. Animals raised in these conditions and exposed to ivermectin after eye transplant demonstrated hyperinnervation in 23 % of the animals examined, a rate that was statistically indistinguishable from animals treated with 5-HT alone (n =31; P =0.38 (Fig. 5c, ci). The finding that 5-HT can reverse the inhibitory effect of the lindane suggests that 5-HT signaling occurs downstream of gap junction communication.

The present study demonstrates the control of innervation of host tissue by grafted sensory organs in Xenopus tadpoles through the modulation of membrane potential. Ectopic eyes were induced in the tails of developing embryos through the transplantation of eye primordia taken from fluorescently labeled donors, allowing the visualization of donor neurites in the host. Ectopic eyes normally display low levels of innervation within the host fin; however, when host cells are depolarized, transplanted eyes show a drastically increased amount of branched innervation, which spreads throughout the fin and trunk. This effect was biphasic (animals demonstrated either no innervation or drastic hyperinnervation), with 40 % of treated animals examining the phenotype. This likely underestimates the true penetrance because the efficiency is limited by: the dilution of our reagents to low concentrations that will not interfere with normal development, physiological variability among the hosts, and tissue/location variation inherent in manual microtransplantation. Interestingly, the host’s native nerves, and tissues transplanted to their normal location in the head, did not hyperinnervate in response to depolarization. This may be owing to retinal ganglion pathfinding cues in the brain over-riding the bioelectric signaling induced by ivermectin (but being absent outside of the head, which allows nerve growth to respond to bioelectrical properties instead), or, more broadly, that tissue becomes more sensitive to bioelectric signals when it is located in an anatomically incorrect location.

These findings highlight a novel role (regulation of innervation paths) for bioelectric signaling via control of transmembrane resting potential (distinct from prior work using electrodes to understand cell responses to externally applied electric fields [39, 40]) and identify a novel target for studies in molecular medicine seeking to implant and connect bioengineered sensory structures.

Hyperinnervation can be inhibited through multiple distinct methods of hyperpolarization, suggesting the effect is a more general response to membrane voltage depolarization itself and not limited to the action of a single ion channel or ion type. This finding confirms similar results in the molecular bioelectricity of cancer [56, 100], left–right asymmetry [58, 84, 101], planarian regeneration [102], and tail/limb regeneration [103, 104]. In all of these cases (including those that were host-specific, such as K+ channel mRNA overexpression), it was shown that a wide variety of distinct methods and ion fluxes could be used to induce the same patterning and cell behavior response provided the resting potential and its distribution were altered appropriately. Thus, hyperinnervation is likewise a true bioelectric effect in which cells sense their neighbors’ V mem.

Experiments with reagents targeting serotonergic signaling implicated a role for 5-HT in ivermectin-induced hyperinnervation. In mammalian systems, 5-HT has been increasingly recognized as a key player in both axon guidance and neurogenesis. Decreases in 5-HT levels have been shown to inhibit neurogenesis in the brain of adult rats [105], and retinogenesis in amphibians [106]. In addition, 5-HT signaling has been shown to affect the patterning of both the developing visual system and the isothalamus [107, 108]; as in our results, these studies also implicated the 5-HT1 family as key signaling cascade member. This is supported by several recent studies showing that serotonergic signaling modulates neural guidance directly [107, 109, 110]. Here we show that extracellular 5-HT availability can drive hyperinnervation and present data that implicate membrane potential as an upstream regulator of 5-HT availability developing nervous system. It is not yet clear if this hyperinnervation is achieved through increased axon outgrowth, retinal ganglion cell proliferation, or decreased cell death, and future studies will be necessary to differentiate between these possibilities.

The current findings also support a growing body of literature linking gap junction to CNS development; a number of studies have shown that gap junctions are key players in regulation of neural connectivity and recognition among neurons [111–114]. Here, we report gap junctional communication as a mediator of ivermectin-induced hyperinnervation. This result was consistent with a role for the intracellular transport of 5-HT between cells according to voltage gradients, where it is then stored in negatively charged cells. In many ways this model mirrors that of left–right axis specification, which requires 5-HT transport, followed by localization, in early Xenopus development [83]. Beyond the results presented, additional evidence also supports the role of gap junction expression and communication in the proper patterning of the developing CNS, as connexins and innexins have been reported to regulate neural proliferation in the neocortex and network morphology of cells in culture [115–119], as well as determining self versus not-self in neural connectivity [120]. 5-HT has also been shown to have a role in gap junction signaling. In addition to moving through gap junctions in response to voltage difference, 5-HT also appears to have a feedback mechanism by which it can affect the open and closed state probabilities of gap junctions between adjoining cells [121, 122]. The feedback between 5-HT movement among cells (via gap junctional communication), as well as the overall regulation of intracellular 5-HT levels (via SERT), establishes the possibility of forming signaling loops with rich emergent dynamics. An understanding of these complex dynamics will likely be required to gain control of nerve growth in biomedical applications.

Integrating all of the data, we suggest one possible model that accounts for all of the observed results (Fig. 6). In untreated animals, 5-HT synthesis begins following depletion of maternal 5-HT stores [83], and is distributed across the embryo. As a positively charged molecule that can pass through gap junctions, 5-HT then accumulates in strongly negative (hyperpolarized) cells, which sequester the molecule from the surrounding tissue (mirroring the normal reuptake function of SERT). In the absence of extracellular 5-HT, ectopic axons receive no signaling from 5-HT receptors on their surface and as a result exhibit very limited growth cone extension (Fig. 6a). In animals treated with ivermectin, the 5-HT sequestering cells are depolarized in response to glycine-gated chloride channel activation and the subsequent loss of chloride ions [56]. In the absence of their normal hyperpolarization 5-HT translocates into extracellular space via SERT, where it binds 5-HT1/2 receptors on the surface of donor retinal ganglion cells, inducing growth cone elongation and hyperinnervation of host tissue (Fig. 6b). This model in many ways matches what has been reported in the developing visual system of mice, where retinal ganglion cells must uptake specific amounts of 5-HT from the extracellular environment for normal patterning to proceed [108].Fig. 6Model of ectopic eye innervation in response to membrane voltage changes. (a) In untreated animals, serotonin (5-HT) is produced across time space during development. The positively charged 5-HT moves between cells via gap junctions and accumulates in negatively hyperpolarized cells, which act as sinks for the molecule. In the absence of extracellular 5-HT, the 5-HT receptors of ectopic eye neurons are not activated and the growth cones show minimal extension.

(b) In treated animals 5-HT also accumulates in negatively hyperpolarized cells. However, exposure to ivermectin (IVM) activates glycine-gated chloride channels (GlyCl), allowing chloride to exit the cell along its concentration gradient, depolarizing the cell. In response to depolarization, the accumulated 5-HT diffuses out of the cell via the 5-HT transporter (SERT). Extracellular 5-HT then binds 5-HT1/2 receptors on the surface of ectopic eye retinal ganglion cells, leading to extension of the growth cone

The response of nerves to V mem of cells in their environment (i.e., a nonautonomous effect arising from neighbors or distant cells rather than the neurons themselves) is particularly exciting because membrane potential could be regulated with extreme temporal precision, for example using optogenetic stimulation to set up standing patterns of resting potential in surrounding tissue [104, 123, 124]. This would enable bioengineers to refine technologies for sculpting neural connections in biomedical applications or in synthetic bioengineered constructs (innervated biobots [125]). Moreover, all host cells express an array of ion channels that could be modulated to produce a desired membrane potential. Unlike diffusible signals, changes in V mem do not require gene transcription, translation, or chemical gradients; cells can rapidly depolarize or hyperpolarize with only a change in the open or closed state of specific ion channels, which can be induced by any convenient pharmacological agent (many of which are already approved for human medical use and are increasingly recognized as “electroceuticals” [52, 53]). Thus, our data suggest a new class of ion channel drug-based strategies for directing transplanted or regenerating neurons in biomedical applications via sculpting of bioelectrical patterns in surrounding tissues.

Below is the link to the electronic supplementary material.Supplementary Figure 1Chloride channel activation does not alter native Xenopus innervation. (a) Immunohistochemistry to visualize the neuronal marker acetylated tubulin shows wild-type optic nerve morphology (white arrow), the characteristic chevron pattern of the somites, and the lateral line (dotted line) running anterior to posterior in Xenopus tadpoles. (b) Treatment of animals with the glutamate-gated chloride channel ivermectin (IVM) throughout development does not alter normal development of the optic nerve, somites, or lateral line. (c) Presence of an ectopic eye arising from eye primordium transplantation does not alter native innervation in the host animal. (d) While IVM exposure results in hyperinnervation arising from donor tissue following transplantation, host innervation remains unchanged. (e) Concentric circles with increasing radii of 50 μm were applied to images of wild-type and IVM-treated tails for innervation comparison by Sholl analysis. (f) Sholl analysis reveals no differences in axon number between control and IVM-treated animals (n =6 for each treatment, 2-way analysis of variance P =0.43) (JPEG 927 kb) (PDF 1.19 mb)