All animal husbandry and staging was carried out according to Kimmel et al. (1995). Embryos used for microscopy were treated with 0.003% phenylthiourea to reduce pigmentation (for transgenic fish lines used in this study, see Supplementary Table S1).

Standard mapping methods using bulk segregant and meiotic recombination analysis of microsatellite markers (Green et al., 2009; Cox et al., 2014) were used to map the sl19 mutation to a critical interval on chromosome 14. Total RNA was then extracted from mutant and wild type 4 days post fertilization (dpf) larvae and cDNA synthesized with Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA). cDNAs encoding candidate genes within the critical interval were sequenced using gene specific primers.

RNA (100 pg/nl) or morpholino oligonucleotides (MO; Gene Tools, 1–2 ng/nl, Supplementary Table S2) were injected as previously described (Cox et al., 2014).

Imaging was carried out as described previously (Cox et al., 2014). Final brightness and/or contrast values of images were adjusted in Adobe Photoshop CC 2017.

This was performed as described in a previous article (Cox et al., 2014). Images were obtained by confocal microscopy as described above.

In situ hybridization (ISH) was performed as previously described (Kucenas et al., 2006). Primers were designed from the predicted sequence of the zebrafish adam19b mRNA (Supplementary Table S3) and a 600 bp fragment was obtained by PCR from 4-day-old zebrafish cDNA. This fragment was cloned into pBSII(+; Promega) and used to prepare digoxigenin-labeled RNA.

An RNA oligo complementary to a sequence in exon 2 of adam19b was designed using http://chopchop.cbu.uib.no/ (Supplementary Table S4). Alt-R oligos were synthesized by IDT. A complex consisting of Alt-R oligos and Cas9 protein (IDT) was injected into single cell zebrafish embryos according to the method of Essner as reported on the IDT website. To validate CRISPR/Cas9 targeting, genomic DNA was isolated at 4 dpf from control and injected larvae, exon 2 was amplified by PCR (see Supplementary Table S4 for primers) and the products sequenced.

A forward genetic screen investigating zebrafish cranial sensory circuit formation was performed using the reporter strain 3.2:gfp (Kucenas et al., 2006; Figure 1A, Supplementary Figures S1). 3.2:gfp^sl19 (sl19) was one of several mutant lines detected. sl19 showed a recessive Mendelian inheritance: mutants constituted 27% ± 6 of larvae in a clutch (n = 235/870 from nine clutches). By 4dpf, the sl19 line contained defects in the central projections of the epibranchial ganglia (gVII, gIX and gX; Figure 1B, Supplementary Figures S1); in all cases the axons projected to the hindbrain as normal, but once in the hindbrain, the axons showed aberrant pathfinding to their targets. While there was considerable variability in the misrouting of the axons, gVII axons typically failed to turn caudally towards the afferent plexus in the viscerosensory column and instead continued in a dorsal direction with no discernable terminal field (Supplementary Figures S1D,F). In some mutants, gIX axons turn rostrally instead of caudally towards the plexus and stop anteriorly to gVII nerve (Supplementary Figures S1C,F); sometimes gIX axons pathfind rostrally, then join gVII and turn towards the plexus Supplementary Figure S1E. In most cases, gX axons do not branch to form a normal hindbrain plexus (Supplementary Figures S1C,D,F). The gV projections do not appear to be affected in this mutant (Figure 1B, Supplementary Figure S1). These phenotypes were never observed in wild-type larvae (n > 5,000).

Linkage mapping indicated that the sl19 allele was located on Chromosome 14, and high-resolution positional cloning identified a C > T mutation located within exon 17 of adam19b (Figure 1C) as the potential causative lesion. This nucleotide switch converts an arginine to a cysteine at amino acid 673. adam19b encodes a member of the A Disintegrin and Metalloproteinase (ADAM) family, and is a 923 amino acid Type I transmembrane protein composed of a series of conserved domains that include: a signal sequence, a pro-domain, a metalloproteinase domain, a disintegrin domain, a cysteine-rich domain (CRD), an EGF-like domain, a transmembrane region and a C-terminus containing signaling-related motifs (e.g., SH3 sequences; White, 2003; Figure 1C). The sl19 mutation is within the CRD of the protein, a region that has been shown to be critical for the protein’s metalloprotease activity (Smith et al., 2002; Kang et al., 2004). Supplementary Figure S2 shows an alignment of several vertebrate CRDs.

Whole mount ISH revealed adam19b mRNA in the CSG (Figure 1D). Expression was detectable in the ganglia early, present in the trigeminal (gV) and both lateral line ganglia at 24 hpf. In 48 hpf embryos, expression was detected in the CSG and also in CNS structures such as the cerebellar plate and optic tectum. By 72 hpf, adam19b expression included all of the CSG as well as cells in the retina, olfactory epithelium and hindbrain: this was also the pattern seen in 4 dpf larvae.

One approach to validating the identity of a candidate gene is to rescue the phenotype of the mutant by injection of the wild type full-length RNA. When full length adam19b RNA was injected into embryos from sl19^+/– in-crosses (three separate experiments), we unexpectedly observed a significant increase in the number of 4 dpf larvae exhibiting the mutant phenotype: 47% (57/121) of injected embryos had a mutant phenotype compared to 25% in un-injected clutchmates (n = 26/102; p < 0.0013, Fisher’s exact test; Figure 1E). This suggested that over-expression of adam19b induced perturbations in afferent circuits. To validate this hypothesis, full length adam19b RNA was injected into single-cell wild type 3.2:GFP embryos. This resulted in disruption of the epibranchial afferents, phenocopying sl19, in 24% of the 4 dpf embryos (9/37; Figure 1F). These findings indicate that appropriate guidance molecule processing by adam19b is critical for epibranchial axon pathfinding.

We predicted that knock-down of adam19b would phenocopy the sl19 mutant. To test this, we used a morpholino targeting the start codon of adam19b (MO1). Injection of MO1 (2–4 ng) resulted in 19% of injected 4 dpf larvae exhibiting a phenotype resembling sl19: misrouting of the projections of the VIIth, IXth and Xth ganglia, but not those of the Vth (Figure 2A; n = 29 out of 156, six clutches vs. 0/127 of un-injected controls, p < 0.0001, Fisher’s Exact Test). Injection of a second morpholino (MO2), designed against the e17i17 splice junction, gave similar phenotypes (Supplementary Figure S3). To further validate adam19b as the affected gene and to address known concerns with the use of morpholinos, CRISPR/Cas9 targeting of exon 2 in adam19b in wild-type 3.2:GFP embryos was performed. These 4 dpf larvae phenocopied sl19 (n = 11 out of 49, three clutches, vs. 0/54 un-injected controls, p < 0.0001 Fisher’s Exact Test; Figure 2B). Confirmation of CRISPR/Cas9 targeting by sequencing is shown in Supplementary Figure S4. Combined, these results strongly support a role for adam19b in epibranchial afferent pathfinding.

To determine if adam19b is required for the formation of other peripheral cranial circuits, we again turned to CRISPR/Cas9 targeting of exon 2 in transgenic fish. In 4 dpf larvae exhibiting defects in epibranchial circuits (Supplementary Figures S5, S6), we did not detect any perturbations in either lateral line circuits [Tg(p2rx3.2GR:nsfB-mcherry;cfos:gfp.sill); Figures 2C,D; Supplementary Figure S5; n = 0/41 or branchiomotor projections (Tg(p2rx3.2:gfp;hsp70:nsfB-mcherry.crest)); Figures 2E,F; Supplementary Figure S6; n = 0/32]. These results suggest that adam19b is utilized selectively by the epibranchial sensory neurons in their pathfinding to hindbrain targets.

The configuration of the branchiomotor neurons and their axons was not affected at 4 dpf by adam19b CRISPR/Cas9 targeting (Figures 2E,F). This suggests that adam19b does not have a critical role in the segmental organization of the hindbrain. To verify this, we used immunostaining of acetylated tubulin, which labels axons throughout the CNS and periphery. As seen in Figures 2G,H, there are no gross perturbations in the CNS in sl19 mutants at 4 dpf (n = 52): the organization of the optic tectum, the cerebellum and commissural neurons appear normal.