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

Possible roles of brain-derived signals in the regulation of embryogenesis are unknown. Here we use an amputation assay in Xenopus laevis to show that absence of brain alters subsequent muscle and peripheral nerve patterning during early development. The muscle phenotype can be rescued by an antagonist of muscarinic acetylcholine receptors. The observed defects occur at considerable distances from the head, suggesting that the brain provides long-range cues for other tissue systems during development. The presence of brain also protects embryos from otherwise-teratogenic agents. Overexpression of a hyperpolarization-activated cyclic nucleotide-gated ion channel rescues the muscle phenotype and the neural mispatterning that occur in brainless embryos, even when expressed far from the muscle or neural cells that mispattern. We identify a previously undescribed developmental role for the brain and reveal a non-local input into the control of early morphogenesis that is mediated by neurotransmitters and ion channel activity.

The brain and central nervous system (CNS) generate information that controls muscle activity to implement behavior in adult organisms. However, the CNS may also provide an instructive influence over the behavior of multiple cell types during the establishment, repair and maintenance of complex anatomical patterns in vivo. For example, the patterning disorder known as cancer has been linked to neural control1. Tumors are readily induced by denervation in cockroach2, and are more readily induced by chemical carcinogens in denervated rabbit ears compared with contralateral controls bearing normal innervation3. The same has been observed with sarcomas implanted into normal or denervated frog limbs4. Even normal tissue can disorganize in the absence of neural signaling, as occurs in the papillae of the mammalian tongue after denervation5. Thus, the presence of growth control signals between the CNS and target cells has led to suggestions of the CNS as a potential approach to treat tumor progression in the clinic6.

Likewise, pattern regulation during regeneration requires neural signals. Appendage regeneration has long been known to be dependent on local neural supply7, although this is apparently an acquired addiction to developmental presence of innervation8. Recent studies have implicated roles for neural signaling in bone regeneration9 and epidermal de-differentiation10. The role of CNS in regeneration is instructive, not merely permissive. The polarity and contiguity of the CNS determines head/tail specification in a number of invertebrate models, including planaria11, 12. In salamander, the presence of nerve in a non-native location can induce the formation of ectopic limbs at wound sites13. Moreover, in the frog model, focal damage to the spinal cord induces alterations in the specific pattern of tail regeneration, with distinct morphological changes resulting from spinal cord interruptions at different locations14. Despite recent advances15, the molecular connection and reciprocal influence between nerves and patterning remains poorly understood16, especially with respect to the ultimate source of neutrally mediated patterning signals.

One of the most interesting contexts for neural pattern control is embryogenesis, when anatomical structures are first established. Parasympathetic innervation controls morphogenesis of the submandibular gland, influencing the branching and general patterning of the organ17. Muscle cells are also affected by interactions with nerves18, especially in terms of gene expression of myogenic regularity factors (MRFs)19. Complementing molecular and cell-level readouts, regenerative medicine approaches to birth defects and bioengineered organs requires addressing neural inputs into large-scale patterning. However, possible roles of the brain (or brain-derived signals) for early patterning, long before behavior begins, remain unknown.

In Xenopus, somitogenesis occurs in an anterior-posterior, bilaterally symmetric manner. In the clock and wavefront model20, 21, the presomitic mesodermal cells dynamically oscillate between permissive and nonpermissive conditions for segmentation (clock) as the embryonic development proceeds from anterior to posterior direction (wavefront along with tissue shortening-mediated Doppler effect) resulting in formation of segmented somites. Wnt, Shh, BMP, Fgf and Notch pathways are involved in regulating the clock with rhythmic activation of these pathways conserved across vertebrates. The intersection of antagonistic gradients of retinoic acid (anterior to posterior) and Fgf and Wnt (posterior to anterior) determine the wavefront22, 23. Given the tandem physiology of nerves and muscles, might innervation help orchestrate the symphony of signals that leads to somitogenesis? Does the presumptive nervous system release instructive signals that might be involved in somitogenesis and muscle development? Is there any element of this process co-opted into muscle regeneration? Similarly, the peripheral innervation forms a stereotypic and complex neural network, which also must be precisely patterned in a way that integrates size and positional information across the whole body24.

Most of the existing knowledge about neural inputs focuses on the local microenvironment; very few studies have examined the contribution of the CNS in providing long-range instructive cues to muscle or peripheral innervation. To probe the role of distal neural structures and to search for tissues that could provide complex nerve-mediated information to downstream patterning targets, we focused on the brain. Here, we establish a brainless Xenopus embryonic model to reveal a role for the brain in the patterning of muscle (somite) and peripheral innervation. Crucially, we show that defects induced by extirpation of the brain, or by teratogenic agents in the context of normal embryogenesis, can be largely rescued in the absence of brain by the dorsal overexpression of an exogenous ion channel or exposure to a muscarinic-receptor antagonist. The characterization of defects and their repair by global (pharmacological) or spatially targeted (molecular-genetic) reagents reveal the first molecular components of the signaling cascade by which the brain directs complex embryonic patterning processes.

To investigate the possible role of brain-derived signals for muscle morphogenesis and patterning, we established an assay in Xenopus laevis: brain removal at stage 25 (Fig. 1a, b), performed at a time when its main subdivisions (forebrain, midbrain and hindbrain) and the rostral-caudal and dorsal-ventral axis are already defined25. By removing the brain at the early tailbud stage (i.e., when somitogenesis is starting), we were able to study muscle structure development in completely brainless developing animals ( > 85% of the microsurgeries we performed resulted in viable animals). To determine whether the brain is required for the onset and/or patterning of myotomes, we evaluated the muscle phenotype at two relevant time points, corresponding to the different myogenic waves (reviewed in ref. 23): early- (stages 30–41; first and second waves completed) and late- (stages 42–48; third wave completed) stages after brain removal, respectively.Fig. 1The absence of the early brain leads to abnormal muscle development and patterning. a After fertilization, the brain was removed from stage 25 embryos to generate BR− animals. Morphological evaluation of muscle phenotype was performed at early- (stages 30–41) and late- (42–48) stages. b Lateral views of stage 25 embryos before (left) and after (right) brain removal. The area occupied by the developing brain is marked with a white-dashed line. (left) rostral is left and dorsal is up. Scale bar, 250 μm. cg, cement gland; e, eye; fb, forebrain, hb, hindbrain, sm, somites. c–h The brain is required for normal muscle development and patterning, as shown after quantitative evaluation of collagen density (short arrows), length of myotome fibers (double-headed arrows), central body axis and myotome angle (overlaid dashed axis and arrowhead-like lines) at early c, d and late f–h stages.

At the onset of development, BR− embryos possessed a lower collagen density in myotome fibers (magenta arrow in c compared to turquoise arrow in b), a significantly more open central angle along the rostra-caudal axis e and shorter somites than control (Ctrl) embryos. During development, defects in the organization of central body axis and muscle patterning were not corrected at any anatomical level in BR− (magenta dashed lines in g compared to turquoise dashed lines in f. The mean angle for BR− is significantly displaced to 180°, compared to those in Ctrl (H). c, d, f, g Photomicrographs taken under polarized light. Rostral is upper right and dorsal is up. Turquoise and magenta arrows indicate correct and incorrect anatomical pattern, respectively. Scale bar, 500 μm. e, h Graphic representation of the mean angle of myotome fibers at rostral, central and caudal levels (blue squares) of Ctrl (white) and BR− (gray) embryos. Data represent the mean and s.d. of three independent replicates (n = 75 animals per group). P values after t (equal variances, black labels) or Mann–Whitney (unequal variances, blue labels) tests are indicated as **P < 0.01, *P < 0.05, ns no significant difference

Soon after brain removal (stages 30–41), animals developing without a brain (BR−) began to display a notable decrease (−43 ± 7%) in the collagen density of the myotomes, compared to the control animals (Ctrl) (OD mean value of 64 ± 8 units for BR− group compared to 113 ± 13 units for Ctrl group; t-test P < 0.01; n = 79) (Fig. 1c, d, turquoise and magenta short arrows). Analysis of the muscle structure revealed that the somites were 25% shorter in BR− than those found in control animals (101 ± 24 μm vs. 148 ± 13 μm, t-test P < 0.01; n = 75) (Fig. 1c, d, double-headed arrow indicates the length of one myotome). Moreover, the spatial organization of the somatic muscle was also perturbed. The body axis and mean angle of the muscle fibers along the anteroposterior axis were significantly altered in BR−, with a mean of 16 ± 2° more-opened angles compared to Ctrl ones (t-test P < 0.01 for rostral and caudal levels, Mann–Whitney test P < 0.01 for central level levels; n = 75) (Fig. 1e and Table 1).Table 1Mean angle of the muscle fibers along the rostro–caudal axis in Ctrl and BR− embryosEarly stageLate stageCtrlBR− CtrlBR− Rostral132 ± 15°148 ± 13°**123 ± 12°133 ± 20°* Central115 ± 12°134 ± 16°** 114 ± 10°123 ± 18°* Caudal110 ± 10°122 ± 12°**111 ± 11°118 ± 15°*Ctrl: control; BR−: brainlessMean angle values for Ctrl and BR− groups are given for the rostral, central and caudal level of the embryo body, respectively, at both early- and late-stages. Values are presented as mean angle ± s.d. Statistically significant intragroup differences after unpaired and two-tailed Student’s t-test are highlighted by *P < 0.05, **P < 0.01 (blue labels are for P values after two-tailed Mann–Whitney test)

We then asked whether the early defects in muscle patterning were also present during subsequent development in BR− embryos. To address this question, we analyzed the number of somites (indicator of segmentation) and the fine muscle structure (angle and length of the myotome fibers) at late stages (stages 42–48). No significant differences were detected for the mean number of somites between Ctrl and BR− embryos (33 ± 3 vs. 32 ± 4; t-test P = 0.74; n = 75), suggesting that the brain was not functionally implicated in segmentation per se, and that our assay does not generally (nonspecifically) impair embryogenesis of the somites. In contrast, the analysis of the fine muscle structure confirmed that the defects in myotome organization were not repaired during subsequent development. Myotomes in late-staged BR− embryos were 10% shorter (136 ± 14 μm vs. 160 ± 13 μm, t-test P < 0.01; n = 75) than in control animals. Likewise, the central axis in the late-stage BR− was significantly displaced, somites lacked the typical chevron-shape, and a difference of −9 ± 1° in the mean angle of the myotome fibers was detected when compared to Ctrl animals (Mann–Whitney test P < 0.05 for rostral and central levels, t-test P < 0.05 for caudal level; n = 75) (Fig. 1f–h, overlaid dashed axis and arrowhead-like lines, and Table 1). Control animals subjected to sham surgeries where either yolk mass or tailbud was resected (Yolk− or Tail− embryos, respectively) displayed normal muscle architecture, indistinguishable from control myotome fibers, both in terms of length/definition and angle of the central myotomes (at distance from the initial resection, in the case of Tail− sham-embryos) (Supplementary Fig. 1A–G), demonstrating that surgery per se (and subsequent regenerative responses) do not induce muscle mispatterning.

We conclude from these data that while the brain is not required for the somite segmentation (partitioning of the presomitic mesoderm into somites), it has a key role in both the onset and establishment of correct muscle patterning and structure.

Muscle organization was adversely affected both at the microscopic tissue organization level, as well as macroscopically – at the animal morphological level (incidence of aberrant phenotype within both Ctrl and BR− population; Table 2). A macroscopic evaluation of embryo morphology revealed a higher percentage of abnormal embryos in the BR− population, both at early (66 ± 6% vs. 13 ± 1%; z-test, P < 0.01) and at later stages (80 ± 1% vs. 16 ± 2%; z-test P < 0.01) compared to Ctrl animals. Muscle organization was affected by brain absence rather than any indirect detrimental effect of tissue removal surgery, as demonstrated the percent of aberration in Yolk− population (14 ± 2% and 15 ± 5% at the onset and later in development, respectively; z-test compared to Ctrl P > 0.05 for both cases; Supplementary Fig. 1H).Table 2Proportion of individuals with aberrant phenotype in Ctrl and BR− populations (pi/ni) after drug treatmentEarly StageLate StageCtrlBR− CtrlBR− No drug30/22080/120**38/228132/164**Scopolamine6/4124/56**6/389/37 ns Carbachol14/4045/60**29/6145/47**Ctrl: control; BR−: brainles; ns: not significantNumber of aberrant embryos (pi) and the pooled-sample size (ni) is given for untreated (no drug), scopolamine- and carbachol-treated Ctrl and BR− populations, respectively, at both early- and late-stages. Statistically significant intragroup differences after z-test are highlighted by *P < 0.05, **P < 0.01, ns P > 0.05

Taken together, these results clearly indicate the importance of the early embryonic brain for normal development of muscle structure occurring at considerable distance.

To characterize the brain-dependent signals that were necessary for normal muscle development, we first asked if the aberrant BR− muscle phenotype could be rescued by pharmacological treatment. Neurotransmitters, such as acetylcholine, are conserved and ubiquitous mediators of the brain’s electrical activity on other organs and tissues in the body. We reasoned that similar mechanisms might be at work before behavior, in the developmental process, as neurotransmitters not only mediate adult physiological function downstream of bioelectrical events but also play a developmental role in the patterning and formation of the synapses that they subserve26. Therefore, we targeted this pathway to attempt to understand and recapitulate the brain’s role in embryogenesis. We tested several cholinergic drugs that were known to target muscarinic (mAChRs) and nicotinic cholinergic receptors (nAChRs) and are standard tools for altering brain performance, especially in terms of memory, attention, and relevant-stimulus processing27.

Ctrl and BR− embryos were treated with scopolamine (a muscarinic-receptor antagonist)28 or carbachol (a dual muscarinic- and nicotinic-receptor agonist)29 (Fig. 2a–h). When comparing percent of abnormal embryos (macroscopic phenotype) within BR− and Ctrl populations for each pharmacological treatment, we found that the absence of brain leads to a higher incidence of abnormalities at the onset of the development, irrespective of the drug used (Fig. 2a and Table 2). When intergroup analysis is performed, however, significant differences between scopolamine-treated BR− and untreated BR− were detected (38 ± 18% vs. 66 ± 8%; z-test P < 0.01), indicating that scopolamine could have a protective effect on the overall morphology when the brain is absent. Our analysis at late stage (Fig. 2b and Table 2) revealed that, while carbachol treatment increased the incidence of aberration in the BR− population (reaching almost 100% of embryos with abnormalities), scopolamine treatment abrogated those effects completely. Consequently, no significant differences were found between scopolamine-treated BR− and scopolamine-treated Ctrl (25 ± 9% vs. 15 ± 5%; z-test P = 0.36), though significant differences were observed between scopolamine-treated BR−and untreated BR− (25 ± 9% vs. 80 ± 2%; z-test P < 0.01).Fig. 2Scopolamine rescues the BR− muscle phenotype. a, b Quantification of the mean percentage of abnormal embryos and statistical comparisons among Ctrl and BR− populations under normal conditions and after drug treatment, at early- (a) and late- (b) stages after brain removal. Values are plotted as mean % ± s.d. (no-pooled data from, at least, three different replicates). c–f. Typical muscle phenotype for Ctrl (c) and BR− (d), and BR− after scopolamine (e) or carbachol treatment (f), as seen under polarized light. Rostral is upper right and dorsal is up.

Turquoise, magenta and yellow arrows indicate correct, incorrect and aberrant formation, respectively. Scale bar, 100 μm. g, h Quantification of the mean length of myotome fibers and statistical comparisons among untreated Ctrl and untreated BR−, scopolamine-treated BR− and carbachol-treated BR− at early- (g, one-way ANOVA, P < 0.01) and late- (h, Kruskal–Wallis test, P < 0.01) stages after brain removal. No significant differences after a posteriori analysis were detected among the different Ctrl groups. Data represent the mean and s.d. of three independent replicates. i. Scopolamine exposure and rescue effects on BR− phenotype. (left) Graphical representation of the different exposure times to scopolamine in BR−, after brain removal (t = 0, magenta arrow; white band means no drug and blue bands means drug treatment) and for a 2-week (2w) period. First-week experimental group was exposed to scopolamine immediately after brain removal and consecutively for the first week. Second-week animals were exposed to the drug 1 week after the brain removal, for 1-week period. First- and second-week animals were exposed to scopolamine immediately after the brain removal and for the 2 next consecutive weeks. (right) Quantification of the mean percentage of embryos with abnormal phenotype within each BR− group. Values are plotted as mean % ± s.d. (no-pooled data from three different replicates). For all panels, number in bars indicates n or number of embryos analyzed for each group. P values after z-test a, b, i and post-hoc Bonferroni’s g or Dunn’s test h are indicated as **P < 0.01, *P < 0.05, ns P > 0.05

We then microscopically analyzed the structure of the myotomes in the different drug-treated groups (Fig. 2c–f for comparative micrographs and Fig. 2g, h for statistical comparisons). Evaluation of somitogenesis and fine somatic muscle structure revealed that both the organization and size of myotomes in BR− treated with scopolamine resembled the typical Ctrl muscle phenotype (turquoise arrows in Fig. 2c, e), and differed clearly from the typical BR− muscle phenotype (represented in Fig. 2d). Both the mean number of somites (32 ± 2) and length of muscle fibers at early- (141 ± 7 μm) and late- (157 ± 17 μm) staged scopolamine-treated BR− were statistically similar to the values for the untreated Ctrl population (with mean number of 32 ± 2 somites and mean lengths of 148 ± 13 μm at early stage and mean lengths of 160 ± 13 μm at late stage; Fig. 2g, h). Conversely, carbachol treatment in BR− had dramatic negative consequences for muscle formation and patterning, which was especially clear at the later stages, deviating the organization of somites and myotomes from the typical BR−-induced aberrant muscle phenotype. At early stages, the size of the myotomes (mean length of 137 ± 11 μm) was similar to what was measured in the Ctrl embryos. However, later on in development, a significantly lower number of somites (24 ± 4; P < 0.01 after post-hoc Bonferroni’s test), longer myotome fibers (196 ± 35 μm; P < 0.01 after post-hoc Dunn’s test), and complete asymmetric and disorganized muscle patterning were observed in the carbachol-treated BR−(magenta and yellow arrows in Fig. 2f).

Taken together, we conclude that the brain may inhibit the muscarinic pathway to achieve correct organization of the somatic muscle system and that the absence or prevention of muscarinic signaling (for example, the pharmacological antagonism mediated by scopolamine) is able to rescue the aberrant BR− muscle phenotype.

We next sought to determine when during muscle development the rescue effect of scopolamine occurs. BR− embryos were exposed to scopolamine for different lengths of time (Fig. 2i left, for a schematic representation of drug-exposure timing). We observed that scopolamine treatment is able to fix the BR− muscle phenotype if the drug exposure starts immediately after the brain removal (Fig. 2i, right), regardless of whether BR− are treated for one (first-week experimental group) or 2 weeks (first- and second-week experimental group). In either case, treatment was able to significantly decrease the percentage of abnormal tadpoles (51 ± 13% and 55 ± 7%, respectively; z-test P < 0.01 for both groups) when compared to the BR− population that had not been exposed to drug (87 ± 11% of abnormalities). Our results indicate that scopolamine acts on events during the first week (from stages 25 to 37) after brain removal.

Considering the many examples of muscle formation and patterning mediated by bioelectrical signaling in both mammals30 and amphibians31, as well as the ability of some channels to rescue profound embryonic defects32, we wondered whether the exogenous expression of a specific ion channel, hyperpolarization-activated cyclic nucleotide-gated ion channel 2 (HCN2), could counter the effects of brain removal. The HCN2 channel is known to be an important modulator of functional bioelectric state33 and can hyperpolarize cells, and it has been recently shown to be implied in firing and rhythmic properties of the cholinergic neurons in both CNS and gastrointestinal tract34, 35. We found the HCN2 channel to be endogenously expressed in the developing neural tube along the base and lateral regions and in the perisomitic area (Supplementary Fig. 2). To address the role of bioelectric signaling during muscle development in organisms incapable of relaying signals from their brains to other tissue, embryos were injected at two-cell stage, in both blastomeres, with mRNA encoding wild-type HCN2 (Fig. 3a, turquoise arrows, HCN2-WT group). Uninjected and water-injected embryos served as controls. In addition, in order to understand the signaling between HCN2-expressing cells and the structural muscle outcome (either a local or long-distance effect), we evaluated the muscle phenotype after injection of HCN2-WT mRNA in only one side (left-right) of the embryo (1/2 HCN2-WT group). Co-injection with lacZ mRNA (as a reporter of injected cells’ progeny; Fig. 3b, left, blue arrow) and detection of the β-galactosidase (β-gal) distribution were used to select and evaluate quantitatively those embryos with strongly unilateral expression (Fig. 3b, right, blue arrow indicates high β-gal expression and, thus, ipsilateral injected side).Fig.

3Ectopic expression of HCN2 rescues the BR− muscle phenotype. a Embryos were microinjected (Inj) with HCN2 mRNA (wild-type channel, WT) either in the two cells (HCN2 WT-group, turquoise arrows) or in one cell (1/2 HCN2 WT, see b blue arrow) at the two-cell stage. Brain was removed at stage 25, and animals with and without brain (Ctrl and BR−, respectively) were analyzed for muscle structure and patterning at early- (stages 30–41) and late-stage (stages 42–48). b 1/2 HCN2-WT injection: embryos were microinjected with HCN2 and lacZ mRNA in one of the cells at two-cell stage (blue arrow). The injection side was confirmed by enzymatic detection of β-galactosidase, β-gal (dorsal view of one Ctrl animal is showed on the right). Rostral is up. Scale bar, 1 mm. c Quantification of the mean percentage of abnormal embryos (macroscopic phenotype) and statistical comparisons between uninjected BR− embryos (No Inj) and the different injected-BR− populations (Water, black arrows: water-injection in the two cells; HCN2: HCN2-WT mRNA injection in the two cells; 1/2 HCN2 + β-gal: co-injection of HCN2-WT and lacZ gene reporter in one LR side). Values are plotted as mean % ± s.d. (no-pooled data from two different replicates). d, e Typical muscle phenotype for uninjected BR− (d) and HCN2-WT injected BR− (e), as seen under polarized light. Rostral is upper right and dorsal is up. Scale bar, 100 μm. f, g Quantification of the mean length of myotome fibers and statistical comparisons among uninjected Ctrl and uninjected BR− (BR−), HCN2-WT injected BR- and 1/2 HCN2-WT (measured on uninjected contralateral side) at early- (f one-way ANOVA, P < 0.01) and late- (g one-way ANOVA, P < 0.01) stages after brain removal. No significant differences after a posteriori analysis were detected among the different Ctrl groups. Data represent the mean and s.d. of two independent replicates.

For all panels, number in bars indicates n or number of animals for each group. P values after after z-test c or post-hoc Bonferroni’s test f, g are indicated as **P < 0.01, ns P > 0.05

First, we quantified the proportion of embryos with abnormal phenotype within each BR− population (untreated or uninjected, water-injected, HCN2-WT and 1/2 HCN2-WT; Fig. 3c). Our results revealed that the expression of a WT HCN2 mRNA in BR− clearly reduced the onset of abnormalities in the macroscopic morphology and significantly decreased the percent of abnormal embryos within the population (from 85 ± 7% in non-injected BR− embryos to 45 ± 7% in HCN2-WT BR− group; z-test P < 0.01). We observed the same rescue effect of HCN2 channel when only one side of the animal was injected. The incidence of aberrant phenotypes in the 1/2 HCN2-WT BR− group was significantly lower than that for the regular (or uninjected) BR− population (65 ± 3% of aberrant individuals in the 1/2 HCN2-WT BR− group; z-test P < 0.01 compared to the non-injected BR−group).

We then microscopically analyzed the somatic-myotome patterning in BR− embryos under the different ion channel misexpression conditions (Fig. 3d–g). Analysis of fine muscle structure revealed that the HCN2-WT BR− mutants exhibited somites and myotomes that were perfectly organized, differing clearly from the typical BR− muscle phenotype (Fig. 3d, e for comparative micrographs). The differences in the number of somites (30 ± 2) and the mean length of myofibers (134 ± 11 μm at early stage and 165 ± 15 μm at late stage) were not statistically significant from those measured in uninjected Ctrl embryos (after post-hoc Bonferroni’s test). The unilateral HCN2 expression in animals developing without brain had a protective effect on the muscle organization of both the local side and the uninjected contralateral side. Like the results for the both sides HCN2-injections, the size of myotome fibers in the uninjected contralateral side were similar to the Ctrl embryos, both at the beginning (early stage: 139 ± 13 μm) and during the development course (late stage: 170 ± 14 μm). Taken together, these results indicate that the ectopic expression of the HCN2 channel counteracts the effects of a missing brain during the somitic myogenesis, even when the ectopic HCN2 channel is present on the other side of the animal (a strongly non-local effect).

Having seen that the brain provides a beneficial patterning influence for aspects of development, we sought to test possible interactions of this effect with teratogenic agents. Could the adverse effects induced by drugs on macroscopic phenotype and tail patterning be prevented if the brain were present? Teratogenic agents targeting the GABAergic, glutamatergic, adrenergic and dopaminergic pathways have been shown to perturb pre-nervous functions of ion channels and neurotransmitter receptors36, 37. Thus, we performed a loss- and gain-of-function screen on our BR− animals (brain removal at stage 25, followed by immediate-drug exposure) with different neuroactive agents, in order to identify drugs that provoked the most severe phenotype in the BR− population (i.e., those that induce more highly aberrant phenotypes than the ones induced by brain removal). We observed that introduction of (RS)-(Tetrazol-5-yl)glycine (RS), an agonist of the NMDA-glutamate receptor38 significantly increased the occurrence of aberrant tail phenotypes within BR− embryos (P < 0.01 for X2 (0.05,6); Fig. 4) after 2 weeks of treatment (stages 42–48).Fig. 4The brain can prevent drug-induced abnormalities of body patterning from occurring. a–d Lateral view of stage-45 tadpoles with brain (left column; Control, BR+) or without brain (right column; Brainless, BR−) after housing in normal conditions (top row; no drug treatment, Drug−) and after continuous treatment with 10 μm (RS)-(Tetrazol-5-yl)glycine (RS, an NMDA receptor agonist), respectively. Drug treatment in Control animals (c, Drug+ BR+) did not produce alterations in tail patterning (turquoise arrows in c similar to a), and there was no incidence of aberrant or hypercurved phenotypes.

Drug treatment in BR− animals (d, Drug+ BR−) lead to a completely aberrant population, with highly curved phenotypes (different to those in BR− without drug treatment, yellow arrows in d compared to magenta arrows in b). Rostral is to the left and dorsal is up. Turquoise, magenta and yellow arrows indicate correct, incorrect and aberrant tail modules, respectively. Scale bar, 1 mm. e Analysis of the phenotype distributions within each experimental group showed that RS is able to induce a significantly aberrant body patterning (a ‘hypercurvature’ phenotype) only if the brain is absent. Data represent the pooled distribution of three replicates (n = 75 animals per group). P < 0.01 for X 2 (0.05, 6). f, g Evaluation under polarized light of drug-treated animals, with brain (f) and without brain (g), revealing clear muscle defects, both in structure and overall patterning, when the brain is not present (yellow arrows in g). This disorganization is not present in drug-treated control animals, exhibiting normal somites and myotome fibers (turquoise arrows in f, see Fig. 2c for similarity to Ctrl group). Turquoise, magenta and yellow arrows indicate correct, incorrect and aberrant muscle structure, respectively. Scale bar, 100 μm

The most frequent tail phenotype in (untreated) BR− embryos is characterized by a single lateral bending, starting approximately in the one-third posterior of the tail (see the most anterior arrow drawn on Fig. 4a–d photomicrographs). Continuous treatment with RS in BR− caused a 99 ± 3% of highly bent tails (including curvature in notochord and spinal cord and spiraling of the tail; see Fig. 4d, yellow arrows, for a representative profile). Strikingly, these RS-induced effects only occurred in animals developed without brain. The RS treatment of Ctrl animals (normal development with brain, BR+) had no effect on tail patterning, and severe phenotypes were not detected. The analysis of the frequency of distribution of the different phenotypes within each population revealed significant differences among the experimental conditions (Fig. 4e; X2 (0.05,6) = 370.8; P < 0.01). The macroscopically identifiable changes in tail patterning, induced by RS treatment in BR− embryos, were also accompanied by clear qualitative alterations in the fine muscle structure and somite organization (Fig. 4f, g). We conclude that presence of the brain helps embryogenesis resist the disrupting effects of otherwise strongly teratogenic agents.

To identify possible pathways mediating this brain-protecting effect, we tested whether scopolamine (which rescues BR−-muscle phenotype) could also protect against the effects of teratogens in BR− embryos. We found that in the absence of a brain, scopolamine treatment partially counteracted the teratogenic effects of RS (Scopo+, Supplementary Fig. 3A–C). Scopolamine-treated RS-BR− embryos (Drug+BR−Scopo+) are much healthier than BR− embryos, but they do not exhibit an entirely recovered Ctrl-like muscle phenotype (Supplementary Fig. 3D). Macroscopic tail phenotype and muscle structure analysis revealed that scopolamine significantly decreases the occurrence of highly aberrant phenotypes (from 92 ± 2% in Drug+BR−Scopo− to 15 ± 1% in Drug+BR−Scopo+, X2 (0.05, 4) = 136.3; P < 0.01), leading to a phenotype distribution similar to what occurs in the regular BR− population (31 ± 9% in BR− without drug treatment or Drug−BR−Scopo−). We conclude that scopolamine treatment prevents the severe deformities caused by this teratogen, but its effect is not sufficient to fully prevent the muscle defects caused in BR− by the presence of the drug.

Having seen the profound effects of the brain on the developing musculature, we next asked what type of alterations/reorganizations could have occurred on the remaining nervous tissue after brain removal (peripheral innervation).

We visualized the body-wide neural network at late-stage embryos by immunolabeling them with acetylated alpha-Tubulin antibody (Tub) and quantifying the Tub-immunolocalization by OD measurements (ranging from 0 (black, no expression) to 255 (white, maximal expression)). This antibody is a widely-accepted marker for nerve fibers because stabilized microtubules, such as those found in neuronal processes, contain important amounts of acetylated tubulin39. The typical neural pattern in the peripheral nervous system (PNS) of Ctrl embryos revealed that, at late stages, three types of fibers can be clearly identifiable after Tub-immunostaining (Fig. 5a): (i) commissural nerve fibers, running along dorsoventral axis (long arrows); (ii) longitudinal nerve fibers, along anteroposterior axis (short arrows); and (iii) internal network or neural network underlying the space between two consecutive segments defined by the commissural ones (unfilled triangles). The internal neuropil in Ctrl animals consisted of a thin network, barely detected by OD measurements (OD mean value of 15 ± 7 units). After brain removal, and similarly to errors detected for segmentation, commissural and longitudinal fibers in BR− were mispatterned (note some incorrect commissural nerve distribution coincident with defects at the level of the somitic myogenesis, magenta arrow in Fig. 5b). Strikingly, embryos developed without a brain exhibited a robust ectopic branching (internal neuropil), with nerve fibers chaotically orientated through the animal body (yellow unfilled triangles in Fig. 5b, OD mean value of 32 ± 11 units; P < 0.01 compared to Ctrl group, after post-hoc Bonferroni’s test). Results from sham-surgery embryos (extirpating non-brain regions) confirmed that this nerve misspatterning is specifically due to brain removal, as demonstrated after Tub immunolocalization (Supplementary Fig. 4).