The brain is required for normal muscle and nerve patterning during early Xenopus development

We conclude that the absence of a brain provokes a dramatic and specific change of the branching of the internal nerve net in the whole animal body (aberrant neural network).Fig. 5The absence of a brain generates an abnormal neural network in the entire animal body. a, b Acetylated-tubulin (Tub) immunoexpression for Ctrl (a) and BR− (b) animals. There three types of nerve fibers: (i) commissural fibers (dorsoventral axis, long arrows); (ii) longitudinal fibers (anteroposterior axis, short arrow); and (iii) internal neuropil (no defined axis, unfilled triangles). Animals developed without a brain show normal commissural and longitudinal nerve fibers (turquoise long arrows in b), with some alterations (magenta long arrow), but a dense internal neuropil (yellow unfilled triangles). c, d Tub-immunoexpression for BR− treated with cholinergic drugs: scopolamine (c) and carbachol (d). Scopolamine treatment was not able to rescue the aberrant internal network (yellow unfilled triangles in c), and carbachol-treated animals exhibited a chaotic nerve patterning (magenta and yellow arrows in d). e Ectopic HCN2-WT expression (injected in both cells at two-cell stage) fixed the BR−–induced internal nerve branching. f Quantification of the mean OD of internal neuropil and statistical comparisons among untreated/uninjected Ctrl and untreated/uninjecetd BR− (BR−, without drug treatment nor ion channel misexpression), scopolamine-treated BR− (BR− + scopolamine), carbachol-treated BR− (BR− + carbachol), HCN2-WT both-sides injected BR− (BR− + HCN2 WT), and HCN2-WT LR side-injected BR− (BR− + 1/2 HCN2 WT) embryos (one-way ANOVA, P < 0.01). No significant differences after a posteriori analysis were detected among the different Ctrl groups. Data represent the mean OD units and s.d. of two independent replicates. Number in bars indicates n or number of animals analyzed for each group.

P values after post-hoc Bonferroni’s test are indicated as **P < 0.01, *P < 0.05, ns P > 0.05. g, h. Ectopic HCN2 expression in only one LR side fixes the BR−-induced internal nerve branching. Tub-immunoexpression on β-gal-reacted sections (dark deposits) in a 1/2 HCN2-WT BR−, showing both the contralateral uninjected side (g) and the injected (h) of the same embryo. Aberrant neural network was completely rescued (turquoise arrows), exhibiting a similar nerve pattern to the Ctrl group in both sides. a-e, h: Rostral is upper right and dorsal is up. g: Rostral is upper left and dorsal is up. Scale bar, 100 μm

To test whether the effect of brain removal was due to lack of an endogenous pruning phase, we analyzed the motoneuron axonal patterning in early stage embryos (stages 31–41; Supplementary Fig. 5) using the antibody znp1. This antibody labels primary motoneuron axons40, and is widely used for many different animal models during embryogenesis26. We found no significant difference in the density of motor axonal branches (internal neuropil) among stage 31, 35 and 45 Ctrl embryos (after post-hoc Bonferroni’s test). Conversely, both errors in axonal establishment (magenta arrows in Supplementary Fig. 5B, D, F showing lack of trajectory compared to turquoise arrows in Supplementary Fig. 5A, C, E) and ectopic/aberrant branching (yellow unfilled triangles in Supplementary Fig. 5B, D, F pointing branches projecting off the main axons; compare to turquoise unfilled triangles in Supplementary Fig. 5A, C, E) were detected from the onset and during the progression of the development in BR− embryos. Thus we conclude that the brainless phenotype involved ectopic growth of neural tissue, not a failure of normal pruning.

Having detected that ectopic branching was a BR−-induced specific effect on peripheral nerve structure, we asked next if the pharmacological treatment and/or ectopic ion channel expression used to rescue the muscle phenotype would have similar effects on the BR−-aberrant neural network (Fig. 5c–f). Results derived from cholinergic-drug treatment revealed that the BR−-induced dense internal neuropil was not fixed by scopolamine or carbachol. Unlike the rescue effects on muscle phenotype, scopolamine was not able to prevent the massive internal neural branching (Fig. 5c; OD mean value of 31 ± 5 units; P < 0.05 compared to Ctrl group, after post-hoc Bonferroni’s test). Carbachol-treated BR− embryos showed a completely disorganized nerve structure, more aberrant than those in drug-untreated BR−, with aberrations in the three different types of nerve fibers (Fig. 5d; OD mean value for internal neuropil of 37 ± 9 units; P < 0.01 compared to Ctrl group, after post-hoc Bonferroni’s test). However, analysis of HCN2-injected embryos showed that the nerve sprouting consequent to developing without a brain was efficiently rescued by expression of HCN2 WT (Fig. 5e; OD mean value of 25 ± 11 units; P > 0.05 compared to Ctrl group, after post-hoc Bonferroni’s test). We conclude that the aberrant neural network in BR− can be fixed by ion channel misexpression providing additional channels, but not by the pharmacological treatment targeting existing ones.

Given that the BR−-induced aberrant neural network was rescued by HCN2 overexpression, we tested the spatial signaling between HCN2-expressing cells and the responding PNS, by means of quantitative evaluation of the nerve patterning in the contralateral uninjected side of 1/2 HCN2-WT overexpressing embryos (see text above and Fig. 3b for experimental injection details). 65% of uninjected sides (low HCN2 expression) had internal neuropil similar to those in both injected side (high HCN2 expression), and Ctrl animals (compare internal neuropil in a typical contralateral uninjected side (Fig. 5g) to that one in injected side (Fig. 5h) and Ctrl animal (Fig. 5a)), displaying an OD mean value of 20 ± 5 units (P > 0.05 compared to Ctrl group, after post-hoc Bonferroni’s test; the quantitative evaluation of the intensity of Tub protein signal on injected side was not performed because OD values could not be comparable to the other analysis, due to the characteristic dark precipitate in the cells expressing β-gal). Our results suggest that the rescue effects of HCN2 are not only mediated by the cells expressing this specific channel (autonomous cell behavior), but that, in absence of brain, the alteration of bioelectrical state could promote accurate nerve patterning via long-distance signals.

Next we investigated the possible pathway by which brain acts on distant tissues: electrical efferent pathway, via spinal cord, vs. alternative or exo-spinal pathway, by severing the spinal connection between brain and the rest of the body (Fig. 6a and Table 3). We analyzed the muscle and nerve patterning in embryos developed with a brain, but with a cervical fragment of spinal cord resected at stage 25 (SC−; Fig. 6a top panel). The mean angle of the myotome fibers in SC− (128 ± 16°) differed significantly from the typical chevron-shape angle in Ctrl animals (114 ± 10°, P < 0.05 after post-hoc Dunn’s test), leading to an overall altered muscle organization (Fig. 6a middle panel, magenta dashed arrowhead-like line). However, SC− presented less severe muscle phenotype than BR−: most of the myofibers presented normal fine structure (with non-significant differences in the mean length of the myotome fibers compared to Ctrl, after post-hoc Bonferroni’s test) with unfrequented structural aberrations. SC− neural patterning exhibited some degree of organization for commissural and longitudinal fibers, but frequent errors were present (magenta arrows in Fig. 6a bottom panel). Internal neuropil was, nevertheless, profoundly altered (yellow unfilled triangles), displaying the typical BR− aberrant or ectopic nerve branching. We conclude that while muscles can develop moderately well without direct spinal cord-dependent brain signaling, the observed brain effects on nerve patterning require an intact spinal cord.Fig. 6Brain effects on muscle and nerve patterning are partially mediated via spinal cord and mimicked via the dorsal expression of HCN2. a-c Upper row, a Lateral view of a stage-33 embryo following spinal cord resection (SC−) at stage 25. Site of injury is indicated by magenta arrow.

b, c Embryos were microinjected with HCN2 (wild-type channel) and lacZ mRNA either in the two ventral cells (b, blue arrows) or two dorsal cells (c, blue arrows) at the four-cell stage. Animals were evaluated at stages 42–48. HCN2-ventral embryos were β-galactosidase negative (β-gal−, white arrow) for brain (center image in b, dorsal view) and SC (right image in b, lateral view) and β-gal+ (blue arrow) for ventral myotomes (right image in b). HCN2-dorsal embryos were β-gal+ for brain (center image in c) and SC (right image in c) and β-gal− for ventral myotomes (right image in c). For lateral views, rostral is left and dorsal is up. Scale bar, 500 μm. Middle row, Typical muscle phenotype for SC− (left panel), HCN2-ventral injected BR− (center panel), and HCN2-dorsal injected BR− (right panel), as seen under polarized light. Muscle patterning (angle of the myotomes, magenta dashed arrowhead-like line) in SC− was altered compared to Ctrl. SC− presented a less severe phenotype than BR− displaying myofibers with normal structure (turquoise arrow) and some incorrect patterning (magenta arrow). BR− + HCN2 ventral embryos presented profound defects in muscle structure, both in angle (magenta dashed line) and in length/organization (magenta arrow) of the myotome fibers. Ectopic or aberrant patterning was also present (yellow arrow). BR− + HCN2 dorsal embryos presented an organized myotome, with normal myofiber structure and organization (turquoise dashed line and arrows). Lower row, typical nerve patterning (commissural fibers indicated by long turquoise arrow, longitudinal fibers indicated by head arrows, and internal neuropil indicated by unfilled triangles) for SC− (left panel), HCN2-ventral injected BR− (center panel) and HCN2-dorsal injected BR− (right panel), shown on anti-acetylated alpha-tubulin antibody staining.

SC− exhibited some degree of organization for commissural and longitudinal fibers (turquoise arrows), but frequent errors were present (magenta arrows). Internal neuropil was, nevertheless, profoundly altered, displaying the typical BR−aberrant or ectopic nerve branching (yellow). Nerve patterning in BR− + HCN2 ventral was markedly altered for all the different fiber types. Conversely, HCN2-mRNA injections in dorsal cells lead to an entirely well-organized nerve phenotype, indistinguishable from controls. Rostral is upper right and dorsal is up. Scale bar, 100 μm. d-f. Quantification of the mean angle (d Kruskal–Wallis, P < 0.01) and length (e one-way ANOVA, P < 0.01) of central myotome fibers and Tub-positive internal neuropil (f one-way ANOVA, P < 0.01), along with statistical comparisons for each experimental group vs. Ctrl (P values above the bar). Data represent the mean OD units and s.d. of two independent replicates (n = 50 animals per group). P values after post-hoc analysis are indicated as **P < 0.01, *P < 0.05, ns P > 0.05 Table 3Muscle and nerve measurements for Ctrl, BR−, SC−, BR− with ventral HCN2 injections, and BR− with dorsal HCN2 injectionsMuscleNerveAngle (°)Length (μm)Branching (OD units)Ctrl114 ± 11°159 ± 1313 ± 4BR− 115 ± 12°* 136 ± 14**32 ± 11*SC− 128 ± 16°* 145 ± 17ns 28 ± 7*BR− + HCN2 ventral138 ± 17°** 132 ± 17**32 ± 12*BR− + HCN2 dorsal109 ± 8°ns 156 ± 11ns 12 ± 6ns Ctrl: control; BR−: brainless; SC−: spinal cord resected; OD: optical densityValues are presented as mean and s.d. Statistical comparisons to Ctrl group are indicated for each group. P values are indicated as **P < 0.01, *P < 0.05, ns P > 0.05 (black labels after post-hoc Bonferroni’s and blue labels after post-hoc Dunn’s test)

Having seen the differential effect of brain signaling on muscle and nerves, and considering the HCN2 rescue effects, we next specifically targeted the dorsal (neural) regions or ventral (somatic muscle) regions of brainless animals with HCN2 mRNA. Embryos were microinjected with HCN2 (wild-type channel) and lacZ mRNA either in the two ventral cells (Fig. 6b upper panel) or in the two dorsal cells (Fig. 6c upper panel) at the four-cell stage. Embryos with HCN2 ventral injections that developed without a brain (BR− + HCN2 ventral) presented profound defects in muscle structure, both in angle and in length/organization of the myotome fibers (Fig. 6b middle panel). Ectopic or aberrant patterning was also present, as seen in BR−. Conversely, embryos with HCN2 dorsal injections that developed without a brain (BR− + HCN2 dorsal) presented a perfectly organized myotome, with normal myofiber structure and organization, indistinguishable from typical Ctrl-muscle patterning (Fig. 6c middle panel). Nerve patterning in BR− + HCN2 ventral animals was markedly altered for all the different fiber types, with an extensive and mispatterned intermyotomal nerve branching (Fig. 6b lower panel). Conversely, HCN2-mRNA injections in dorsal module of the embryo lead to a well-organized nerve phenotype, indistinguishable from what occurs in Ctrl embryos (Fig. 6c lower panel). Quantification is shown in Fig. 6d–f. We conclude that in order to rescue the BR−-induced effects, HCN2 needs to be expressed in dorsal structures (neural tube).

Here we show that the early morphogenesis and patterning of trunk muscle structure and innervation in animals developing without a brain are highly abnormal. Brainless (BR−) animals’ peripheral neural network is profoundly disorganized, with fibers chaotically oriented through the animal body, while the muscle organization was adversely affected both at the microscopic tissue organization level (length/definition and angle of myotomes), as well as at the animal morphological level (aberrant phenotype). The effect is brain-specific, as removal of other body regions does not induce this effect. The brain is not only required for normal development, but also exerts a protective effect, ameliorating the effects of teratogenic drugs which are made notably worse in brainless embryos.

We gained insight into the mechanism of brain-dependent, long-range patterning effects by rescue assays. Ectopic expression of a hyperpolarization-activated cyclic nucleotide-gated ion channel (HCN2) was sufficient to prevent muscle and nerve mispatterning in brainless animals. The HCN2 rescue effect only occurs when CNS-fated blastomeres are targeted, suggesting that bioelectrical signals, when acting within neural tissues, can mimic the endogenous effects of the brain. Future work will determine the relative contributions, to the HCN2 rescue, of modulating spiking-encoded activity in the nervous system41, and alteration of non-neural distributions of resting potential that have likewise been implicated in developmental patterning42.

We also started exploring the potential therapeutic implications of our findings, by recapitulating the protective effects using pharmacological agents targeting endogenous channels (not requiring exogenous misexpression). Drugs targeting the cholinergic system differentially affected BR−-induced outcomes. A dual nicotinic and muscarinic agonist exacerbated the defects in muscle structure; in contrast, suppression of muscarinic pathway, by means of scopolamine treatment, rescued it. The fact that scopolamine can partially rescue the defects in muscle, but not the aberrant nerve phenotype, and that spinal cord-transected animals develop a partially normal muscle phenotype, suggest that the brain could play a direct role in muscle development that may not involve spinal pathway and peripheral nerves (Fig. 7a, b). Taken together, these data reveal an essential role for brain-derived signaling during embryogenesis, long before its involvement in behavior, and show that the patterning effects of the brain can be largely mimicked by available reagents targeting cells’ bioelectric state.Fig. 7Brain signaling for muscle and nerve development and patterning a. Schematic representation drawing of a Xenopus embryo, showing the main components of our experiments: brain (blue), spinal cord-peripheral nerves (pink) and somites-muscle (brown). Brain effects on nerve patterning could occur directly (2), by using efferent spinal pathway. Brain effects on muscle patterning could occur indirectly (3a), by acting on neurons, or directly (3b), by acting on muscle. b A spinal mechanism, for coding the information about patterning and morphogenesis, could occur via direct signaling from the brain to the neurons in the spinal cord (pink circle).

According to our results (different treatments are indicated with purple labels), the effects of the peripheral innervation on muscle cells can be partially explained in terms of developmental bioelectricity or changes in Vmem excitability. We hypothesize that at these stages, the brain is in part controlling the bioelectric state of peripheral tissues, and a correct balance (turquoise triangle) of brain activity (long-range instructive cues or top-down perspective) and local signals (bottom-up perspective) is necessary for correct morphogenesis. Both an excess of tonic activity (induced after carbachol or RS treatment) and an excess of slow If gradients through membrane lead to mispatterning. The extra-spinal pathway by which the brain is acting on muscles can be mimicked pharmacologically, with pharmacological agents targeting bioelectricity (i.e., scopolamine). We hypothesize that scopolamine is acting at presynaptic/synaptic level, blocking the inhibitory ACh actions (via mAChRs) on slow ion flows, and leading the Vmem to appropriate values for muscle patterning. c, d. Schematic representation of neuromuscular specificity in normal development (c, with brain) and in absence of the brain (d, BR−). Our results suggest that ectopic branching detected in the absence of a brain is not due to deficits in early pruning or target retrograde signaling. Pathfinding behavior at the onset of Xenopus development starts at the spinal cord level, as early patterned electrical gradients in SC cells is required for the correct axon guidance. The different treatments applied in our experiments (purple labels and circles) are placed on the cellular/subcellular domains where they are probably acting

X. laevis is uniquely suited for the study of biophysical mechanisms underlying pattern regulation. Somites in X. laevis are comprised of myotome fibers43 and embryonic myogenesis in Xenopus involves intricate interplays between several MRFs: MyoD, Myf5, Myf6 (also called Mrf4), Myogenin and Myf6. Dynamic temporal and spatial expression patterns of these factors orchestrate the main steps of muscle development: lineage specification of muscle cells, differentiation of myocytes, fusion into myofibers, and formation of muscle groups (reviewed in ref. 23). Future developments integrating in vivo physiological monitoring with transcriptional reporters will address the interaction of neurotransmitter and bioelectric signals from the brain with the transcriptional control of these and other important factors.

Our results show that brain input is important for patterning and myotome organization (Fig. 1c–h) rather than the early events of myogenesis, such as fate or induction of cell lineages. Somite segmentation is not altered by brain removal, as demonstrated by the normal number of somites in BR− animals at late stages, suggesting that the periodicity of somite formation and initiation of myogenesis (mediated mainly by MyoD and Myf5) do not require early brain-derived signaling. Brain inputs might be acting in later events, when the myotome fibers have been already established in the somite. At these differentiation steps, expression of Mrf4 and Myogenenin might be susceptible to brain signaling. In Xenopus, Mrf4 has been showed to be the main myogenic factor subject to nerve influence. This evidence is, however, conflicting and some authors postulate that the initial activation of Mrf4 is nerve-independent in embryonic Xenopus 44. However, different studies demonstrate a neural influence on Mrf4 expression. In fact, muscle denervation leads to decrease Mrf4 levels, in both development and regeneration19. Each myogenic factor might have a distinct role in the regulation of nerve-regulated genes, such as different subunits of a neuromuscular junction (NMJ) receptor: the nicotinic acetylcholine receptor nAChR45.

A variety of studies have demonstrated the relationship between innervation and correct anatomical development of muscle structures, in different vertebrates. Denervation of rat skeletal muscles in utero provokes degeneration and myofiber fragmentation, as well as a slowing down of myofiber growth46. Similar findings were seen in frogs, where the denervation of the hind limb leads to a ~ 12% reduction in growth, and for the development of limb transplants in chick embryos47. Abnormalities in overall patterning (size and shape) of the limbs in absence of nerve influence have been also described in salamanders, being the muscle the most sensitive to nerve absence48. Such data are usually thought to be explained by ‘trophic’ or permissive effect of nerves49. Our results, where brainless embryos have more innervation than the control ones (Fig. 5a, b), suggest an additional and ‘instructive’ role, mediated by signals normally originating in the brain.

Our results reveal a long-distance role of brain-derived signaling at both organ level (overall muscle phenotype; Fig. 2a, b) and tissue level (myofibrillar structure; Fig. 2g, h). The partially fixed muscle phenotype after the spinal cord resection or after scopolamine treatment seems to indicate that in addition to the spinal pathway, muscle is susceptible to a long-distance action of the brain, perhaps via diffusion of neurotransmitter signals (schematized in Fig. 7).

Some of the most tantalizing data reveal a role for the brain and CNS in patterning of distal structures. Older studies have postulated a role of the brain-derived signals, which are conducted along the spinal cord, on morphogenesis in Xenopus tail regeneration; the subcommissural organ was suggested as the source of this signaling50. Recent experiments using point ablation in the spinal cord14 showed that patterning of the final regenerated tail is influenced for both the injury position along the AP axis and a non-linear combinatory effect when two different AP injury levels are performed. Hence, shape-instructive long-distance signaling is not explained by simple presence of nerve (trophic effects) but appears to generate distinct information along different positions in the spinal cord. Our ‘hypercurvature’ phenotype in BR− animals (Fig. 4e) is comparable to Mondia et al.’s most severe one, suggesting that similar signals from brain might are acting on development and regeneration.

We report the first steps to identify the mediators of the early brain’s effect by showing that the effects on myopatterning derived from the absence of a brain can be completely rescued by ectopic alteration in neurotransmitter (Fig. 2) and ion channel (Figs. 3 and 6c middle panel) signaling.

Even in the absence of a brain, muscarinic AChR (mAChR) suppression (via scopolamine treatment) led to close-to-normal muscle development, while nicotinic AChR (nAChR) and mAChR activation (carbachol treatment) provoked a more aberrant muscle phenotype. Exogenous application of acetylcholine modulates the intrinsic properties of spinal motoneurons after SC transection in the juvenile salamander through the mAChR51. Likewise, functional nAChR have been reported in spinal motoneurones of X. laevis embryo52. These studies support the reorganization in spinal cord circuits in absence of brain as the main target for actions of scopolamine and carbachol on muscle cells. The alteration of membrane potential (Vmem), by a direct action of these drugs on receptors in the early muscle cells, could be an important part of the mechanism. We hypothesize that scopolamine is acting on NMJ, specifically at presynaptic/synaptic level, blocking the Ach actions by mAChRs on slow ion flows52 (If; Fig. 7d). This If disinhibition could counter the excess of excitability induced by the presence of ectopic branching in BR−. Our results suggest a strategy to pharmacologically target muscular defects, by means of regulating the balance between slow If and action potentials at NMJ level.

Prior work revealed patterning roles of several neurotransmitters37, 53, and suggested that neurotransmitter drugs could be potent teratogens36. We found that an NMDA agonist provoked severe phenotypes in BR− animals (RS, Fig. 3). NMDA-glutamate receptors (NMDAR) in muscles are also excitatory and depolarize the muscle cells leading to their contraction (such as carbachol does on nAChR). Given recent work on the importance of steady-state developmental voltage gradients during Xenopus muscle patterning31, our findings are consistent with the hypothesis that maintaining the balance between developmental bioelectricity (resting potential gradients) and discrete action potentials at NMJ may be important during muscle development and inhibitory signals from the brain (probably via extra-spinal alternative pathways) may be acting to shield the developing muscle cells from such excitatory stimulus that might cause muscle mispatterning (Fig. 7b, brown rectangle representing muscle membrane). Recently, it has been showed that activation of NMDAR impairs the myogenic differentiation in C2C12 cells through mTOR/MAPK signaling pathway54. The protective effect of brain detected in the RS-treated control animals (vs. the devastating effects observed in RS-treated BR−-animals) evidences the key role of the brain signaling for the correct morphogenesis. The possibility of exploiting and perhaps strengthening brain-derived protective signals represent an exciting area of research for future efforts in the field of birth defects.

In addition to the muscle effects, the absence of a brain during development generates an abnormal patterning and organization of the peripheral innervation or PNS of the animal (Fig. 5; somatic component of the PNS). We used immunohistochemistry with an antibody to acetylated α-tubulin39 to analyze the somatic neural processes coming from both primary motoneurons and sensory neurons, in order to study the instructive role of brain inputs in patterning and global organization of the nervous system.

Cell fate and differentiation for primary motoneurons and sensory neurons start early in embryogenesis. Rohon-Beard (RB) neurons cells originate during gastrulation and present electrical excitability as early as stage 2024. Neural crest (NC) material segregates at gastrula stage (around stage 15) and starts migrating around stage 2025. In our assay, brain removal is done after the onset of the migration, at stage 25, and hence, when the fate of the trunk NC cells has been already specified55.

In neural morphogenesis, after differentiation and migration, and once the progenitors have reached the final location, new steps are necessary for axon growth and guidance (pathway selection), formation of initial connections (target selection), and connection remodeling and pruning (address selection). The early neural morphogenesis, differentiation and migration of X. laevis, both for sensory and motor somatic neurons56, earlier than the time point for the brain removal, lead us to suggest that aberrations detected in neural patterning in BR− animals might be due, to the later steps on pathfinding and synapse formation. Thus, the peripheral innervation pattern is an epigenetic outcome that depends on complete development of the functional physiology of the brain; in brain removal, not all of the signals required for a correct peripheral pattern are conveyed.

The peripheral innervation formed in brainless animals is mispatterned throughout the whole animal body (Fig. 5). How neuromuscular specificity arises during embryonic development has been a controversial issue. Are correct connections pre-established from the outset or do motor axons project randomly into the developing muscles followed by extensive pruning of incorrect connections? Our analysis of brainless animals (Supplementary Fig. 5) shows that nerves are altered very early in development (as soon as stage 31), with clear errors in finding the correct trajectory (actually in BR− nerve fibers fail in turning to create the correct intermyotome division, see Supplementary Fig. 5B, ventral magenta arrow). Our results suggest that brain-derived signals are important for correct early patterning, not for maintenance of a default pruning program.

HCN2 targeted to dorsal part (brain and spinal cord) rescues both muscle and nerve. Spinal cord-transected embryos show the same aberrant ectopic branching than BR−, suggest that pathfinding behavior at the onset of Xenopus development starts at the spinal cord (or CNS) level (Fig. 7c; as previously was showed in zebrafish57 or axolotl58). Moreover, in line with recent evidence59, we show that the early patterned electrical in spinal motoneurons (as HCN2 dorsal is able to rescue the whole phenotype and brain effects on nerve patterning require an intact spinal cord) is required for the correct axon guidance. While previous work did not explain the mechanisms by which the rhythmic activity in spinal neurons affects early nerve development, our data are consistent with a role for the brain in determining patterned morphogenesis by controlling slow ion flows on primary-neuron fast firing (Fig. 7b).

Interestingly, we discovered a rescue effect on the neural phenotype only by means of reagents that target membrane potential (Vmem), induced by the ectopic expression of HCN2 (Figs. 5e–h and 6c bottom panel). While targeted misexpression of ion channels has been demonstrated to rescue patterning of the brain itself32, we report here that using ion channel expression can overcome developmental defects stemming from brain damage. Future work will exploit the emerging advances in optical imaging of neural activity in vivo60 to characterize the temporal properties of brain-derived signals and their modification by HCN2.

One remarkable aspect is that HCN2 expression can rescue patterning of cells on the other side of the animal – cells that do not themselves express HCN2 (Fig. 5g). Such long-range bioelectrical signaling has been observed in bioelectric tumor suppression61 and control of apoptosis/proliferation32. Our data reveal that expression of HCN2 at a remote location can induce repair of peripheral neural structures in a damaged background, suggesting a range of therapies where easily-accessible tissues are targeted to induce repair in a difficult-to-reach site. Future work extending closed-loop optogenetic control to neural and non-neural somatic tissues62 will refine the specific bioelectric state that facilitates normal network structure and test these interventions in adult disease models.

Scopolamine-treated BR− animals display an aberrant neural network but normal muscle phenotype, suggesting distinct regulatory mechanisms. We do not claim that correction in nerve patterning (by HCN2 injection; Fig. 5e–h) is the only factor responsible for muscle correction by that treatment (Fig. 3b). The direct action of ion channels in muscle cells and their indirect action on muscle patterning through peripheral nerves are not mutually exclusive possibilities. Future work will be directed towards understanding their mutual contributions to the repair process. The fact that pharmacological modulation of Ach transmission does not affect the neural patterning (Fig. 5c, d) but bioelectrical modulations do (as in vitro studies seems also to indicate63), suggests that both long-distance brain input and developmental bioelectricity are potential targets for future applications in the area of muscle-nerve communication pathologies.

Here, we establish an experimental model for the study of long-range patterning control, and the discovery of pre-behavior functions of the nascent brain. This model is amenable to optical, biophysical, genetic, and chemical approaches, and offers the unique opportunity to target diverse spatial sites (due to the Xenopus fate-map) to test non-cell-autonomous mechanisms of brain-dependent instructive patterning signals. Future development of transgenic promoters will allow tissue-specific tests, while targeted ablation technology may enable finer dissection of brain regions responsible for various patterning outcomes.