Monoclonal antibodies DP311 and DP312, raised against Drosophila Prd (Davis et al., 2005) were kindly provided by Nipam Patel. Embryos were fixed in MEM-FA (3.7% formaldehyde, 0.1 M MOPS pH7.4, 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, 0.05% Triton X-100), washed/quenched in 1X PBS, 0.35% Triton X-100, 50 mM NH4Cl, and washed in 1X PBS, 0.05% Triton X-100. Embryos were then blocked for 30 min at room temperature in “Blocker” Buffer (1X PBS, 0.05% Triton X-100, 1% Thermo Scientific “Blocker” BSA). Embryos were incubated in antibodies 1:100 in “Blocker” Buffer overnight at 4°C. Embryos were washed three times in 1X PBS, 0.05% Triton X-100 (rocking 15 min at room temperature, each time). Samples were incubated in AlexaFluor-488 anti-mouse IgG secondary antibody 1:500 in Blocker buffer for 1 h at room temperature. Washes were performed as in the previous step before mounting and imaging.

Gravid adult Ciona robusta (intestinalis Type A) were collected and shipped by M-REP from San Diego, CA. Embryo dechorionation and electroporation was performed as previously established (Christiaen et al., 2009a, Christiaen et al., 2009b). Fluorescent reporter plasmids were electroporated at concentrations of 10–35 µg/700 µl electroporation volume for histone fusions, and 70–100 µg/700 µl for other reporters. Leech H2B::mCherry and Unc-76::GFP/YFP reporter plasmid backbones have been previously published (Gline et al., 2009; Imai et al., 2009). Tyrp. a>2XGFP was a kind gift from Filomena Ristoratore (Racioppi et al., 2014). Embryos were fixed and washed as for immunofluorescence above, without blocking or incubating. Embryos were mounted in Mounting Solution (1X PBS, 2% DABCO, 50% Glycerol) and imaged on compound epifluorescence or scanning-point confocal microscopes. All relevant sequences can be found in Supplementary Material S1.

Pax3/7 sgRNA expression cassettes on the “F + E” optimized scaffold (Chen et al., 2013; Stolfi et al., 2014) were designed by CRISPOR (Haeussler et al., 2016) and cloned together with the U6 promoter (Nishiyama and Fujiwara, 2008) by One-Step Overlap PCR (OSO-PCR) as previously described (Gandhi et al., 2017; Gandhi et al., 2018). OSO-PCR cassettes were individually tested for mutagenesis efficacy by the “peakshift” method (Gandhi et al., 2018) before cloning into plasmids, by electroporating 25 µl of unpurified PCR product and 25 µg of Eef1α>Cas9 (Sasakura et al., 2010; Stolfi et al., 2014) per 700 µl volume (Supplementary Table S1). Amplicons representing alleles from pools of embryos were PCR amplified and Sanger-sequenced as detailed in Gandhi et al., 2018. All sgRNA, Cas9 vector, and primer sequences and electroporation recipes are in Supplementary Material S1.

To visualize Pax3/7 expression in the neural plate borders of C. robusta (also known as intestinalis Type A), we performed immunofluorescence (IF) staining of neurula- and tailbud-stage embryos using monoclonal antibodies raised against the Drosophila Pax3/7 ortholog (Paired), which show broad specificity for pan-metazoan Pax3/7 proteins (Davis et al., 2005). Both antibody clones DP311 and DP312 showed staining of nuclearly-localized Pax3/7 expressed in anterior-posterior rows of cells at the mid-neurula stage, which will contribute to dorsal neural tube and dorsal midline/dorsolateral epidermis later on (Figures 1C,D, Supplementary Figure S1) (St. 15; 6.5 h post-fertilization or hpf at 20°C), consistent with its expression by mRNA in situ hybridization as previously reported in Ciona (Stolfi et al., 2015) and in the nearly identical embryos of another tunicate species, Halocynthia roretzi (Ohtsuka et al., 2014). Although DP311 showed qualitatively less background, DP312 staining was brighter overall, so we proceeded with the latter clone. At late neurula stage (St. 16; 7 hpf at 20°C), staining started to appear in the progenitors of the posterior sensory vesicle (PSV) region of the brain, and anterior MG (Figure 1E). At mid-tailbud stage (St. 22; 10 hpf at 20°C), Pax3/7 IF signal was observed in the lateral rows of the anterior MG (Figure 1F), while slightly later at St. 24 (12 hpf, 20°C) signal was stronger in the PSV and “neck”, which are neural progenitors situated between the neurons of the brain and the MG (Figure 1G). These patterns are also consistent with those detected by Pax3/7 in situ hybridization previously (Imai et al., 2009; Stolfi et al., 2011).

Because the early (neurula) and late (tailbud) domains of Pax3/7 expression are not connected by descent, arising from distinct blastomeres of the 8-cell stage embryo, we hypothesized that separate cis-regulatory modules might control their activation in the different lineages. Previously, a ∼3.5 kb sequence immediately upstream of the Pax3/7 transcription start site had been shown to drive reporter gene expression in the neural plate borders (Horie et al., 2018) (Figure 2A). We tested a longer sequence stretching further upstream, comprising -5877 bp upstream of the Pax3/7 start codon, alongside the Fgf8/17/18 reporter which labels the A9.30 lineage that gives rise to most of the MG (Imai et al., 2009) The extended fragment was sufficient to drive GFP reporter gene expression in both the neural plate border derivatives and the lateral rows of the neural tube in the brain, neck, and anterior MG (Figure 2B). Expression in the Neck was often not as bright as in the brain or MG, reflecting perhaps the apparent downregulation in this compartment as observed by in situ previously (Imai et al., 2009). When we isolated just the -5877 to -4121 fragment and placed this in front of the basal promoter of the Friend of GATA gene (bpFOG), which is routinely used in Ciona as a minimal promoter, (Rothbächer et al., 2007), this was sufficient to drive GFP expression in the brain/neck/MG but not in the neural plate borders (Figure 2C). Expression was seen in both A11.120 and A11.119 left/right pairs of neural progenitor cells in the MG. Although expression of Pax3/7 in A11.119 appears to be downregulated when observed by in situ hybridization (Stolfi et al., 2011), it has been reported as initiating in the mother cell (A10.60). Thus, this likely represents GFP protein accumulation, and is consistent with the IF staining observed at 10 hpf (Figure 1E).

We also identified separate cis-regulatory elements in introns 1 and 4 that were also sufficient to drive expression in neural plate border derivatives (Figures 2D,E), hinting at complex regulatory control of Pax3/7 expression through partially overlapping “shadow enhancers” (Hong et al., 2008). Taken together, our results suggest that the different domains of Pax3/7 expression in the neurectoderm (neural plate borders and lateral rows of the brain/neck/MG) are largely regulated by distinct cis-regulatory elements (Figure 2F).

To study the functions of Pax3/7 in neural development in the tunicate embryo, we sought to use tissue-specific CRISPR/Cas9 as previously adapted to Ciona (Sasaki et al., 2014; Stolfi et al., 2014; Gandhi et al., 2017). We tested two candidate single-chain guide RNA (sgRNA) constructs (Pax3/7.2.1 and Pax3/7.4.1), targeting exons 2 and 4 respectively (Figure 3A). We validated their mutagenesis efficacies following the “peakshift” method of Sanger-sequencing PCR amplicons of targeted sequences (Gandhi et al., 2018). Efficacies for Pax3/7.2.1 and Pax3/7.4.1 were 17% and 34% mutagenesis, respectively (Figure 3B) (Supplementary Table S1).

We next tested the ability of these sgRNAs, when combined, in eliminating Pax3/7 expression from the neural plate borders. To do this, we co-electroporated the sgRNAs with FOG>Cas9, which drives Cas9 expression in all animal pole-derived cells including the neural plate borders. A FOG>CD4::mCherry reporter plasmid was also co-electroporated to reveal transfected cells’ outlines, and DP312 antibody IF was used to assay Pax3/7 expression. Pax3/7 sgRNAs were compared to a negative control sgRNA (“Control”) that does not target any sequence in the Ciona genome (Stolfi et al., 2014). While in the negative control neurula embryos (St. 16) the Pax3/7 antibody clearly labeled the nuclei of both transfected and non-transfected cells (Figure 3C), we detected substantial loss of Pax3/7 IF signal in cells transfected with Pax3/7 CRISPR constructs. Non-transfected cells in the same embryos served as a clean internal control for Pax3/7 IF, showing that the loss of Pax3/7 was visibly confined to only transfected cells. In some embryos, no Pax3/7+ cells were seen at all in the transfected half, suggesting biallelic knockout of Pax3/7.

To score this, we performed Pax3/7 IF on Pax3/7 CRISPR and negative control embryos co-electroporated FOG>H2B::mCherry to visualize the nuclei of transfected cells at stage 16 (Figure 3D; Supplementary Figure S2). Cells were scored for H2B::mCherry expression and Pax3/7 IF signal, independently on either left or right borders of the neural plate to account for mosaicism (Figure 3E). Embryos with no H2B::mCherry expression in the neural plate borders (untransfected) were not included, as were embryos oriented in a way that obscured the view of the neural plate borders. In 52 such Pax3/7 CRISPR embryo halves, 40 were H2B::mCherry+, but only 7 of those were also positive for Pax3/7 while 33 were negative, indicating loss of Pax3/7 in 82% of transfected neural plate borders. In negative control embryos, 42 of 43 H2B::mCherry + halves were positive for Pax3/7. Pax3/7 was observed in 100% of untransfected (H2B::mCherry-negative) neural plate border cells in both Pax3/7 CRISPR (12/12) and negative control embryos (10/10). In sum, these results confirmed the specificity and high efficacy of Pax3/7 CRISPR knockout in this system.

In Ciona and vertebrates, neural tube closure is driven in part by epithelial “zippering” involving the formation of cellular “rosettes” in which cells undergo sequential apical contraction and cell junction exchange (Hashimoto et al., 2015; Hashimoto and Munro, 2019; Mole et al., 2020). In some of our CRISPants, we noticed a lack of such rosettes and epithelial zippering, implying that perhaps loss of Pax3/7 might impair this process (Figure 3C; Supplementary Figure S3). In vertebrates, Pax3/7 factors are essential for neural tube closure (Epstein et al., 1991). Thus, we sought to determine if CRISPR knockout of Pax3/7 might induce similar neural tube defects in Ciona.

We carried out tissue-specific knockout of Pax3/7 as described above, targeting Cas9 and therefore CRISPR activity to the animal pole (neural plate borders, not brain/neck/MG). We then imaged early tailbud embryos (St. 20) looking for any neural tube closure defects, not differentiating between mild or severe defects (Figure 3F). When we scored Pax3/7 CRISPR embryos (Figure 3G), we observed defects in 40% of embryos (n = 100). 36% had seemingly normal neural tube closure, while 22% of embryos were “unclear” due to orientation of embryo on the slide. In contrast, in embryos electroporated with the same components except a negative control sgRNA (“DenhT2”) in the place of Pax3/7-specific sgRNAs, only 16% embryos showed a neural tube defect of any severity, likely due to non-specific effects of dechorionation/electroporation. 59% of negative control embryos were normal, while 25% were unclear. These data suggest that, as in vertebrates, Pax3/7 is required in the neural plate borders for normal neural tube closure.

To test whether Pax3/7 regulates conserved gene expression and/or specifies conserved cell types derived from the neural plate borders in Ciona, we focused on the melanin-containing pigment cells that give rise to the ocellus and otolith pigment cells. These two cells arise from the neural plate borders and might be evolutionarily linked to neural crest-derived melanocytes in vertebrates (Abitua et al., 2012; Olivo et al., 2021). However, Pax3/7 CRISPR did not significantly affect their specification, as assayed by expression of the Tyrp.a>2XGFP reporter (Racioppi et al., 2014). Therefore, we cannot conclude whether Pax3/7 is required or not for the specification and differentiation of neural plate border-derived cell types in Ciona.

Although its role in specifying cells derived from the animal pole-derived lineages of the neural plate borders of Ciona remains unclear, Pax3/7 was previously shown to specify anterior MG fates derived from more medial cells of the neural plate. Pax3/7 is required for the specification of a single pair of descending decussating neurons (ddNs, A12.239 cell pair), which was shown through a combination of morpholino knockdown (Imai et al., 2009), overexpression of full-length Pax3/7 and dominant-repressor (Pax3/7::WRPW) as well as cis-regulatory analyses (Stolfi et al., 2011). However, a genetic knockout of Pax3/7 was never attempted in the Ciona MG. We used the Fgf8/17/18 promoter (Imai et al., 2009) to drive expression of Cas9 in the A9.30 lineage, which gives rise to most of the core MG including the ddNs (Figure 4A). Fgf8/17/18>Cas9 was co-electroporated with Pax3/7-targeting sgRNAs and a Dmbx reporter plasmid (Figure 4B) (Stolfi and Levine, 2011), and larvae were assayed for reporter gene expression in ddNs. As predicted, an average of 36% of Pax3/7 CRISPR larvae showed Dmbx reporter expression across two replicates, compared to an average of 78% of negative control CRISPR larvae (Figures 4C,D). These results further confirm the requirement of Pax3/7 in the regulation of Dmbx expression and ddN specification in the Ciona MG.