An in vivo brain-bacteria interface: the developing brain as a key regulator of innate immunity

Next, we extracted the specific genes with both the highest response (up- or downregulation) and unique (exclusively expressed) within each comparison (see Supplementary Data 1 for the complete list of genes; Supplementary Tables S2–S8 and Supplementary Note 2 for Summary of the top differentially expressed genes), concluding that most of the genes implicated in the bidirectional communication brain-infection belong to TNFR-I signaling pathway and encoding of ligands and receptors specific of immune/myeloid cells. Next, we grouped the differentially enriched elements according to their “large-scale functions” to reveal the types of processes mainly regulated within each brain-infection condition. The more predominant gene networks were related to blood, bone, ion transport or transmembrane potential (bioelectric signaling),55,56 apoptosis, development and regeneration, neural, immune response and bacteria (Supplementary Fig. S4a–d). The bioelectric-related genes (ion transport) occupied 3% of the total DEGs in control embryos under infection (Supplementary Fig. S4a), while this percentage was clearly lower in the rest of groups (around 1%), consistent with the known role of bioelectric signaling in the innate immune response.9 The removal of the brain (in absence of infection) did not induce any changes in the transcription of genes related to bone (0%), with the neural-related functions the most affected (10% of the total of the transcriptome for Ctrl NI vs. BR– NI; Supplementary Fig. S4c). How the absence of a brain affects the response to infection (Ctrl UTI vs. BR– UTI; Supplementary Fig. S4d) was reflected by considerable changes (respect to Ctrl NI vs. Ctrl UTI) in the percentages of immune-related transcripts (−8%), neural-related (+6%), bone (−4%), and developmental and regeneration-related (+2%) functions.

To reveal the gene regulatory networks (GRN) and motifs and features that were statistically overrepresented, we performed SNEA (Fig. 5c–f, Supplementary Figs S4 and S5, and Supplementary Data 1 and 2 for the complete list of pathways). In animals developed with a brain, 80 pathways were controlled by the infection (66 of them were unique to Ctrl NI vs. Ctrl UTI; black labels in Fig. 5b). Conversely, in animals developed without a brain (RNA isolation took place after brain removal), infection induced the differential regulation of 151 pathways (116 were unique to BR– NI vs. BR– UTI; orange labels in Fig. 5b), suggesting that about 70% of the response to infection depends on the presence of brain.

In grouping the neural-related pathways, we determined that three elements were shared by Ctrl UTI and BR– UTI (innervation, neurogenesis and brain microcirculation), five sub-networks were included exclusively in infected embryos with brain (Ctrl UTI: adrenergic transmission, CNS function, glia proliferation, brain blood flow and cerebrovascular circulation; Fig. 5c) and eleven were unique to infected embryos without brain (BR– UTI: dopaminergic system, glial cell reaction, glial cell response, hindbrain development, hippocampus plasticity, nerve discharge, neural tube closure, neural tube patterning, neuron development, neuron differentiation, and peripheral-nerve function; Fig. 5d). Intriguingly, in the latter category, networks related to bacteria were upregulated by 20%, overall as a group (Supplementary Fig. S4e). The only common pathways regulated in both response to brain removal and tail removal were microtubule- and cell division-related (cell cycle regulation, cell fate, chromatin remodeling, chromosome movement, epithelium development, microtubule bundling, microtubule cytoskeleton organization, microtubule sliding, and microtubule/kinetochore interaction). Some of the unique pathways differentially regulated after brain removal only, include apoptosis of neutrophils, apoptotic chromosome condensation, DNA annealing, DNA damage excision, DNA damage recognition, DNA end joining repair, DNA strand breakage, neurogenesis, neuron development, and neuron differentiation. The sub-network related to innate immune response was one of the most affected by the presence/absence of brain. Specifically, the innate immunity for brainless embryos with infection exhibited an upregulation of up to 11% (median change in the network of 1.11-fold; Fig. 5e). Some of these unique immune-pathways regulated in absence of brain were complement activation (classical pathway, Fig.

5f), immune complex clearance and activation, macrophage-focused (adhesion, apoptosis, fusion) and neutrophil-focused (activation, chemotaxis, extravasation and recruitment; all genes within a differentially expressed pathway are provided in Supplementary Data 2).

Thus, our analysis revealed profound changes in the transcriptional networks responsive to infection induced by absence of brain, suggesting a key role of the early brain in producing a robust innate immune response in response to systemic infection and implicating a variety of factors related to immune and neural pathways such as migration or dopaminergic transmission.

Considering the molecular candidates and pathways revealed by RNA-seq, we next targeted one of the regulatory networks exclusively affected after infection in the absence of the brain. From the 11 neural pathways unique to BR– UTI, we decide to validate the possible role of the dopaminergic signaling in mediating the increased susceptibility to infection, as it has been also implicated in macrophage migration.57

First, we functionally tested whether the DA levels were quantitatively different between Ctrl− and BR– infected embryos. Embryos were analyzed by LC–MS/MS at early st. 35 (~20 h post-surgery, 44 h post infection), immediately before the peak of infection-induced death in BR– animals and the significant differences both in survival rate and apoptosis are reached respect to Ctrl embryos (Fig. 2a, b). DA turnover rates were significantly different between Ctrl (35.56 ± 10.49 pg/uL) and BR– (14.90 ± 0.99 pg/μL) embryos (r = 3, n = 30; unpaired t-test P = 0.0274; Fig. 6a, Supplementary Fig. S6), confirming that the DA concentration in the tail region of the embryo after infection is differentially affected in the presence vs. absence of the brain.

Next, we performed pharmacological assays targeting the type-1 or type-2 family of dopamine (DA) receptors (D1R, D2R). We used specific agonists and antagonists of each dopamine receptor,58 such as the following: for D1R,57,59 SKF-3839360 (SKF) and SCH-2339061 (SCH) were used as an agonist and antagonist, respectively; for D2R, Quinpirole57 (Quin) and L-741,62662 (L741) were used as an agonist and antagonist, respectively. In order to understand whether we could mimic the BR– phenotype (decreased survival rate) in Ctrl embryos and/or mimic the Ctrl phenotype (increased survival rate) in BR– embryos, we exposed Ctrl and BR– infected embryos to the dopaminergic drugs from st. 25 (immediately after brain removal) to st. 48, where we evaluated the survival rate after each treatment (r = 3, n = 30, N = 90 embryos per treatment; Fig. 6b). For Ctrl infected embryos, no significant differences in the survival rates were found after exposure to any of the drugs. For BR– infected embryos, exposure to SCH (D1R antagonist) rescued embryos from the increased death (14.8 ± 8% with no drug) to a survival rate of 36 ± 6%, displaying significant differences respect to the values reached for the rest of no-drug/drug-treated BR– embryos (two-way ANOVA P < 0.01; Bonferroni’s posttest for BR–: SCH vs. MMR P = 0.0049; SCH vs. SKF P = 0.0015; SCH vs. Quin P = 0.0006; SCH vs. L741 P = 0.0071) and no differences respect to no-drug Ctrl embryos (48 ± 7%, Bonferroni’s posttest P = 0.1878 for BR– SCH vs. Ctrl. MMR). These results confirmed, using functional assays, that the dopaminergic regulatory network is affected in BR– infected embryos, and that antagonism of the D1R signaling pathway can rescue the induced-infection death detected in absence of brain, mimicking the protective effects of the brain signals on the immune response to infection.

Immunology, developmental biology, neuroscience, and regenerative medicine are all converging in an emerging interdisciplinary field of profound significance for biomedicine as well as basic biology.63 While the brain–immune axis is now beginning to be characterized,9,17,64 many gaps exist in our understanding of the functional links between the brain, response to infection, and regenerative processes triggered by wound healing.65 This is especially true for the earliest developmental events that shape the future interactions of these subsystems during health and disease. Here, we exploited the highly tractable Xenopus laevis model to identify unrevealed aspects of the innate immune response that rely on the presence of brain, and we characterized the cell-level and transcriptional machinery that underlies the effects.

We challenged Xenopus embryos with UTI E. coli load of 4.5 × 103 cfu/ml; this dose is up to five orders of magnitude smaller than some model organisms and human organs tolerate,66–69 providing a comfortable dynamic range within which to evaluate susceptibility and tolerance. Crucially, we show that the specific presence of the brain strongly impacts the ability of embryos to survive bacterial challenge (Fig. 1). The early brain is protecting the embryo by inducing cellular and physiological responses (decreasing the infection-induced damage and apoptosis; Fig. 2, and promoting macrophage migration; Figs 3 and 4) and molecular mechanisms (suppressing transcriptional consequences of the infection; Fig. 5) that help to overcome the bacterial threat. Pharmacological and functional assays revealed that absence of brain during infection leads to decreased levels of peripheral DA. Targeting the DA-D1R signaling pathway in brainless animals rescues them from infection-induced death (Fig. 6). These results suggest that brain-derived DA signaling is key in mediating the protective effects of the brain signals in response to infection.

The brain is not necessary for the early development and proliferation of primitive myeloid precursors (Supplementary Fig. S2). However, in response to infection, signals from the brain are required for proper macrophage migration (Fig. 3), revealing that brain-dependent signals regulate the behavior of immune cells. Without infection, at later stages with a more mature immune system, the absence of a brain leads to aberrant and ectopic distribution of macrophages and neutrophils (Fig. 4), which occurs in the context of an ectopic neural network which could be a factor guiding the abnormal distribution of immune cells (Fig. 4m–o). Importantly, removing the brain does not induce the migration of immune cells to the injury site (the head region), which clearly occurs with other parts removed from the body, such as tail amputation (Supplementary Fig. S3n).

The protective effects of the brain do not require spinal cord contiguity, an important fact for the design of immune-enhancing therapies. Survival rate of SC– embryos after infection is similar to Ctrl (Fig. 1i), and mature immune cells in SC– animals have similar numbers and general distribution as the control group (Supplementary Fig. S3a, b). Conversely, the peripheral-nerve phenotype in SC– animals showed the same ectopic growth and sprouting of neural network than in BR– animals (Supplementary Fig. S3o–q). Thus, unlike the peripheral-nerve distribution, immune cell behavior is not controlled via the spinal cord. Likewise, neural crest ablation did not recapitulate effects of brain removal (Supplementary Fig. S1e–g), consistent with the essential role of the brain per se.

Important information was provided by two additional groups: tailless (a control for general surgical damage), and the Simvastatin-treated group (Simv, a control for general tissue stress). Simvastatin causes severe myotoxic effects in humans and zebrafish embryos,39,40 leading to death from continuous exposure or above μM concentration, indicative of a strong drug-derived stress. In Xenopus embryos, Simv treatment led to severe aberrant muscle phenotypes, without altering the brain morphology (Supplementary Fig. S1h–j). Survival after infection (Fig. 1f) and the immune phenotype (Supplementary Fig. S3) of the Simv group are not significantly different than the Ctrl group, indicating that muscle alterations are not a primary cause of immune defects, and the combination of infection + strong stressor does not reduce survival. Thus, the absence of brain when a bacterial infection is present leads to dysfunctional macrophage behavior and high susceptibility to infection that is not recapitulated by even severe general stressors or tissue damage.

To more fully understand the link between brain and immunity, we assessed the bacterial load at 48 h after infection and the relation between survival and pathogen load, using a metric that combines these two variables in a single number: Host-Pathogen Response Index (HPRI; Fig. 1i). Comparing survival percentage across the control infected vs. brainless infected vs. brainless noninfected (Fig. 1f, g) under constant pathogen load confirmed brain-dependent susceptibility, which includes a tolerance component (Fig. 1h, i). In addition to the brain, it is likely that other components of the circuit could be discovered in the future. The cellular and morphological mechanisms behind the higher susceptibility in absence of the brain include increased apoptosis and inflammation induced by infection (Fig. 2), which promote an earlier peak of death and hamper the recovery from infection and normal development to st. 48.

Analysis of the embryos’ robust transcriptional response shed light on the pathways related to brain → body, bacteria → body, and bacteria → brainless body signaling. Infection when brain is present induces the differential expression of immune-related elements such as antigen recognition, leukocyte cell adhesion, lymphoid differentiation, T-cell proliferative response and tolerance, basophil activation, disease resistance, thrombocyte aggregation, and virus morphogenesis. Infection. Similarly, some of the more significantly upregulated genes, include genes encoding proteins (ligand-receptor) for immune cell functioning (Fig. 5, Supplementary Fig. S5), such as the gene encoding colony-stimulating factor 3 (CSF 3, a granulocyte growth factor necessary for the differentiation of bone marrow cells to granulocyte-lineage70), the gene encoding for receptor-interacting serine/threonine-protein kinase 3-like (RIP, a component of the Tumor Necrosis Factor-receptor I (TNF-R1) signaling complex,71 and RAG complex genes involved in maturation of the antibody repertoire of adaptive immunity. The differential regulation of the TNF-signaling pathway is consistent with recent mammalian data showing that innate immune response in mice to Listeria monocytogenes results in upregulation of TNF in cerebrovascular fluid.2

The deficient response to infection detected in brainless animals, with low survival rates and defects in immune cell location, was transcriptionally reflected, with an increased activity of the innate immune sub-network by ~11–12% (Fig. 5e) compared with animals with a brain. The ineffective upregulation of transcripts may indicate compensatory transcriptional responses in the body from other systems in the absence of a brain when challenged with a pathogen.72 RNA-seq revealed the most significant immune and neural pathways and transcripts affected by absence of brain such as complement activation, macrophage-focused (adhesion, apoptosis, fusion), neutrophil-focused (activation, chemotaxis, extravasation and recruitment), or overexpression of BAG-4 related genes (or TNF-R1 silencers) and cell adhesion proteins (VCAM-1). Networks functionally related to bacteria are inhibited by 20% and eleven neural pathways are uniquely affected after infection when brain is not present.

The dopaminergic transmission is one of the most affected regulatory networks in presence of infection when the brain is absent, and levels of peripheral DA are decreased in absence of brain. Macrophage migration is one of the targets for the protective role of the brain and an increasing number of recent studies relate dopamine and inflammation and immunity.57,73,74 Based on our cellular, molecular and pharmacological results, a possible mechanism explaining the protective effects of the brain in presence of bacteria is illustrated in the drawings-models of Fig. 6c, d. In presence of bacteria, the immune response of the embryo is initiated in the early brain (st. 25) and communicated to the periphery by modulating DA-signaling pathways in macrophages (trough D1R antagonism) during the first 48 h post infection (or before st. 35). The DA-activated macrophages migrate and act systemically decreasing apoptosis and infection-induced inflammation. As a consequence, embryo’s tolerance to infection is harnessed and by st. 40, induced-infection death is entirely stabilized. In absence of brain, low levels of peripheral DA cannot activate immune cells and promote them to initiate the migratory response. Consequently, in absence of a macrophage network, brainless embryos become more susceptible to the lethal effects of the bacterial infection, which induces high levels of inflammation and apoptotic events that lead to an earlier and massive peak of death by st. 35.

Brainless animals showed an aberrant overexpression of central- (nigrostriatal dopamine) and peripheral- (peripheral-nerve function) neural networks (Fig. 5d), as well as bacteria-related pathways (Fig. S4e), revealing responses of the host to microbial presence that could explain the differences in the HPRI and, consequently, the lower survival rates detected in brainless animals after infection. Specifically, the TNF-R1 pathway is significantly upregulated in intact embryos (developed with brain) in presence of infection, and it seems to be deregulated when brain is absent. Brainless infected animals overexpress genes related to BAG-4 or silencer of death domains (SODD)—a widely expressed 60-kiloDalton protein that associate with the death domain of TNF-R1, silencing it when overexpression is detected.75 The accumulation of immune cells in niche in absence of brain (Fig. 3m, n) is also detected in transcriptome changes, with the significant upregulation of the vascular cell adhesion protein 1 (VCAM-1) that mediates the adhesion of immune cells. Cell cycle and chromatin-expression regulators, such as ubiquitination of proteins (mediated by tnip1 or cbl), methylation (via downregulation of euchromatic histone-lysine N-methyltransferase 1L that methylates the lysine-9 position of histone H3 and tags it for transcriptional repression) or uridylation of mRNAs appear significantly and exclusively regulated in brainless animals (with or without infection). In infected animals, absence of brain induces the specific networks for ubiquitization of CSF-1R (macrophage receptor for growth and proliferation). This mechanism, along with high expression of adhesion proteins, could be the responsible for the attenuation in the macrophage proliferation detected in brainless animals.

In conclusion, a unique vertebrate model that develops without a brain enabled morphological, transcriptional and functional evidence for brain-mediated modulation of the immune system reactivity. Specifically, we demonstrate that the absence of brain makes embryos more susceptible to pathogen, lowering tolerance41,42 and increasing apoptosis and inflammation. The influence of the endogenous brain seems to be mediated by control of cell localization, especially affecting macrophage migration to fight infection. RNA-seq revealed the most significant immune and neural pathways and transcripts affected by absence of brain. Overall, 70% of the response at the cellular network level is different based on the presence/absence of brain. Our functional assays reveal that DA signaling could be key in the bacteria–brain–immune crosstalk, suggesting that modulation of dopamine receptors could be an important strategy for mimicking the protective effects of brain signals on the immune response to infection in biomedical settings.

The brain is an active component of the innate immune response. Future work will focus on decoding the bioelectric and biochemical signals that mediate its effects on distant immune cells and identifying heretofore unrecognized cellular targets of these signals. More broadly, beyond the brain–body–bacteria axis, the understanding of the influence of the brain over cell- and molecular-level processes is an interesting frontier, with numerous potential applications across basic biology and medicine. A full understanding of brain effects on cellular behavior (complementing neuroscientists’ focus on whole animal behaviors) is likely to not only shed light on the evolution of neural and immune systems but also to facilitate the development of intervention strategies in biomedical settings. We speculate that combinations of appropriate bioelectrical and neurotransmitter signals can become a useful tool for addressing infectious and other disease states.

Xenopus laevis wild-type and transgenic embryos were fertilized in vitro according to standard protocols76 in 0.1X Marc’s Modified Ringer’s solution (MMR; 10 mM Na+, 0.2 mM K+, 10.5 mM Cl–, 0.2 mM Ca2+, pH 7.8). The transgenic Xenopus, expressing Green Fluorescent Protein (GFP) under the control of the lurp1 gene promoter, xlurp::GFP, was obtained from the Marine Biological Laboratory (MBL, National Xenopus Resource RRID:SCR_013731).77 Xenopus embryos were housed at 18–21 °C and staged according to Nieuwkoop and Faber.78 All experimental procedures involving Xenopus embryos were approved by the Institutional Animal Care and Use Committees and Tufts University Department of Laboratory Animal Medicine under protocol M2017-53.

All bacterial strains were generously provided by Dr. Matthew A. Mulvey (University of Utah, Salt Lake City, Utah, USA) and have been previously described.79 We utilized an uropathogenic Escherichia coli (E. coli) strain UTI89 to elicit a pathogenic response. Bacteria were grown at 37 °C in LB medium (MP Biomedicals, LLC) supplemented with 50 mg/mL ampicillin (Fisher). One milliliter of the UTI89 stock (1 × 109 colony forming units per ml (cfu/ml)) overnight bacterial culture was pelleted and resuspended to a stock concentration of 1 × 1010 cfu/mL in phosphate-buffered saline before injections in embryos. Gastrula stage embryos were injected using borosilicate glass needles calibrated for a bubble pressure of 25–30 kDa and 150 ms pulses. All known variables were kept consistent across biological and technical replicates. To calculate the injected cfu, embryos from different biological replicates were harvested immediately after injection (t = 0 h) for subsequent lysis and plating. Later, embryos infected with E. coli UTI 89 (UTI condition) and not-infected (NI condition) were incubated at 21 °C (Fig. 1a for general diagram of experimental design). Surgeries were performed 18–24 h after infection (st. 25). Survival within each experimental group was evaluated at several stages until embryos reached st. 46–48, or 4–5 d after infection. Even though bacteria were grown and concentrated following a constant methodology, and embryos were collected and grown under the same conditions, variation in survival rates was observed between experiments. This is likely due to differences in bacterial growth in situ and in the genetic background of each individual from different egg clutches.

For these reasons, every individual comparison within an experiment was conducted using eggs from a single fertilization and a single suspension of bacteria, and controls were used within each study. At least 120 embryos were used for every treatment being compared, and each assay was tested in triplicate. Three additional injection assays were performed in order to calculate the number of bacteria per embryo at 48 h after infection. Bacterial load was determined after lysis of live embryos (five independent embryos/experimental group; TissueLyser for 30 s at 30 Hz) and following standard protocols of microbial plating analysis. Timing selection was established following previous published work,9 which demonstrated that the lethal effect of UTI89 infection on Xenopus embryos can be evaluated at 72 h and the survival rates maintain constant after 5 d. The Host-Pathogen Response Index (HPRI = % survival/(1 + log10(CFU + 1)) was developed to compare embryo survival with respect to pathogen load in order to provide a quantitative metric of susceptibility and tolerance that comprises both pathogen load and survival. HPRI scores of 0 indicating complete susceptibility and 100 indicating complete embryo survival and clearance of pathogen. The slope of HPRI over time or following a perturbation, reflects tolerance phenotypes.

At 18–24 h post infection, st. 25 infected (UTI condition) and not-infected (NI condition) embryos were randomly allocated in one of the four experimental groups, respectively (Fig. 1b–e, Supplementary Fig. S1): Control (Ctrl; Fig. 1b, c), Brainless (BR–; Fig. 1d, e), Spinal Cord resection (SC–; Supplementary Fig. S1c, d) or Tail-bud resection (Tail–). The SC– group was used to test whether the effects are mediated by neural communication going through the spinal cord. We used Tailless embryos as control for invasiveness of removal of large amounts of tissue and as a positive control, since we have previously demonstrated this assay increases resistance to infection.9 Brain removal was performed under dissecting microscope, using a dissecting knife (FST #10055-12), via a single cut of the antero-dorsal region corresponding to the three main subdivisons of the developmental brain (see Supplementary Fig. S1a, b for images of stage-48 embryos showing that no brain regeneration takes place after removal at st. 25). For SC–, a consistent piece of the most cervical levels, sizing 50–100 μm in length, was completely removed by using two forceps with super-fine tips (Dumont #5ST, FST 11252-00). Tail buds were amputated from the most posterior fifth portion of their body using a scalpel blade (additional details of microsurgeries can be found in ref. 36). As an additional experimental control for the brain ablation, we included the group Neural Crest ablation (NC–; Supplementary Fig. S1e–g). Published protocols for ablation of the migrating cranial NC in Xenopus80 were followed at st. 17 embryos, ~6–8 h after infection.

After surgery, animals were allowed to heal in 0.75X MMR for 1 h, then raised in regular 0.1X MMR at 21 °C and scored and analyzed for percentage of survival rate at st. 26–27 (2–4 hps), st. 30 (8 hps), st. 35 (24 hps), st. 40 (36 hps), st. 42 (48 hps), and st. 46–48 (4 days ps). At each time point, embryos were harvested and prepared for morphological analysis (in situ hybridization or immunofluorescence). Molecular assays, RNA extraction and liquid chromatography-coupled tandem mass spectrometry (LC–MS/MS), were performed on whole embryos harvested at 3 and 20 hps, respectively.

Stage-25 UTI and NI Xenopus embryos were treated with 0.4 μM Simvastatin (Supplementary Fig. S1h–j; Tocris Cat. No. 1965/50) for 18 h at 22 °C. After 18-h exposure, embryos were transferred to fresh normal media (0.1X MMR) and allowed to develop up to the same stages as the rest of the groups for harvesting and posterior analysis. Stock solution of the drug was created by dissolving the compound in ethanol to a final concentration of 0.24 mM. Further dilution was made in 0.1X MMR.

In situ hybridization was performed as previously described9 on UTI and NI embryos belonging each experimental group: Ctrl, BR–, SC–, and Tail–. Briefly, specimens were washed in 0.1% Tween-20/phosphate-buffered saline (PBS-T) and dehydrated through increasing concentrations of methanol. Xenopus laevis spib-a (spiba; GE Dharmacon) and Xenopus tropicalis mmp7 (matrix metalloproteinase-7) were used as probes and kindly provided by Enrique Amaya (University of Manchester). spib- is an ETS transcription factor, highly conserved with mammals,10,47,81 necessary for myeloid specification. spiba expression at st. 28 embryos marks the earliest primitive myeloid population.3,10 mmp7 is a secreted metalloproteinase involved in extracellular matrix remodeling and extensively used as macrophage differentiation and migration markers.3,10,11,48,82 mmp7 probe was used to label migrating embryonic macrophages on st. 35–36 embryos, as relative to other myeloid markers, mmp7 is expressed considerably later.10 Probes were generated in vitro from linearized templates using a digoxigenin (DIG)-labeling mix (Sigma-Aldrich, #11277065910 Roche).

Quantification of spiba- (Supplementary Fig. S2) and mmp7- (Fig. 3) positive cells was performed using ImageJ software on bright-field images of whole-mount ISH st. 28 and st. 35 embryos, respectively. Images and counts were performed on the left side of the animals belonging each experimental group (Ctrl, BR–, SC–, Tail–) and for each condition (NI vs. UTI). Firstly, each image was carefully subdivided and cropped (to avoid overlapping) in four independent (face, tail, dorsal and ventral) using six reference points or landmarks (Fig. 3a, b, Supplementary Fig. S2a, b, s–w). Clearly identifiable landmarks for Xenopus embryos were orderly used (from 1 to 6) to decrease variability and bias during the blind counting, such as the following: (1) beginning of spinal cord (SC), most anterior portion or cervical levels, (2) intersection between posterior edge of the IV branchial arch (ba) and ventral edge for the first and most anterior somite (Sm); (3) anterior to the heart (h); (4) end of hindgut (hg); (5) beginning of the tail bud (tb); (6) ventral line delimited by somites. The face region is composed of the entire anterior area including the olfactory bulb, the eye area, and the otic vesicle, the tail region is composed of the posterior area including tail fin (posterior to the hindgut); the dorsal region is composed of the spinal cord, notochord, and somites (it begins directly posterior to the face region and ends directly anterior to the tail region), and the ventral region is composed of the entire gut area (it is directly inferior to the dorsal region, begins posterior to the face region and ends anterior to the tail region, before hindgut and anus). Since chromogenic ISH staining can contain some background enzyme activity, we standardized the cell counting and were able to differentiate one cell vs. a group of cells (red asterisk vs.

white-dashed circle, respectively, in Supplementary Fig. S2d) using a custom ImageJ macro applied to images of embryos processed in the same ISH session. First, on clearly identifiable independent cells, we measured the smallest and biggest size (in pixels2) for one single positive cell. Then, we applied a threshold and adjusted the saturation to increase the contrast and segment cells from potential background. Finally, we indicated the minimal and maximal size for one independent cell on ‘Analyze Particles’ tool. The generated results were overlaid on the original image to remove possible artifacts. Number of positive cells per each region was normalized to the area in pixels2. The total number of cells per embryo was obtained as a sum of the four independent regions. Per each marker, spiba and mmp7, inter- and intragroup comparisons were analyzed. Intergroup comparisons were performed among the different experimental groups (Ctrl, BR–, SC–, Tail–) for each condition (NI vs. UTI) and per body region (face, tail, dorsal, ventral and total; Fig. 3c–l, Supplementary Fig. S2e–n). Intragroup comparisons were done between NI vs. UTI embryos, for each body region, within the same experimental group (Fig. 3e, o–r; Supplementary Fig. S2o–r). Particular care was taken to ensure that embryos from all the different groups per each condition were processed in the same batch at the same ISH session.

Cleaved Caspase-3 (Asp175; CC3), XL-2 (anti-Xenopus leukocytes,8 and acetylated alpha-tubulin (Tub) antibodies were used to detect apoptotic cells,43 leukocytes,8,43,83 and peripheral nerves,36,84 respectively, on wild-type and xlurp::GFP embryos. Ctrl, BR–, SC–, Tail–, NC– and Simv-treated embryos were fixed 1 h in MEMFA76 and washed three times in PBS. They were then permeabilized in PBS-T for 30 min at room temperature (RT), followed by a 1 h blocking at room temperature in PBST supplemented with 10% heat-inactivated goat serum. The embryos were then incubated overnight at 4 °C with the primary antibody (monoclonal mouse anti-XL2 at 1:1000, kindly provided by Makoto Asashima’s Lab at Tokyo University; polyclonal rabbit anti-CC3 at 1:300, Cell Signalling 9661; monoclonal mouse anti-Tub at 1:500, Sigma T7451). On the next day, they were washed six times in PBS (1 h each time, RT), before being incubated overnight with the secondary antibody (goat anti-mouse IgG conjugated with Alexa-Fluor 555; Invitrogen) at 4 °C. The following day, animals were photographed using an Olympus BX-UCB microscope under ×4 and ×10 magnification, controlled by Metamorph software. Particular care was taken to ensure that embryos from all the different groups were processed in the same batch at the same immunofluorescence session.