Using CRISPR/Cas9 to investigate the role of candidate human disease gene orthologs in Ciona

Abstract

The tunicate Ciona robusta offers a tractable non-vertebrate chordate model for probing gene function via tissue-specific, CRISPR/Cas9-mediated mutagenesis in F0. Building on Arcadia Science’s Zoogle platform, which identifies and ranks orthologs of human genes from various non-traditional model organisms, we carried out a pilot project to probe the developmental roles of three notochord- and endoderm-expressed candidate orthologs of human disease genes (Fcho, Pgm3, and Nckap1) alongside a fourth gene (Plastin) implicated in papilla cell elongation. This preprint compiles and updates a series of research project milestones previously posted episodically on Zenodo. Here we summarize the full results and our conclusion about this pilot project. Using CRISPR/Cas9, we found that tissue-specific knockout of Pgm3 and, to a lesser extent, Fcho caused significant defects in larval tail elongation. Separately, CRISPR knockout of Plastin, an actin-bundling gene expressed throughout the sensory-adhesive papillae of the larva, caused a subtle reduction in papilla cell elongation when combined as a duoble knockout with another actin-bundling protein-encoding gene, Villin. These results identify Pgm3 as the most promising candidate for further development as a Ciona-based model of human disease and demonstrate the utility of tissue-specific CRISPR screening for prioritizing candidate disease gene orthologs identified through comparative genomics platforms like Zoogle.

INTRODUCTION

The tunicate Ciona robusta (formally intestinalis Type A) serves as a convenient chordate model for investigating gene function during embryogenesis (Satoh, 2013). Its simplified body plan, rapid development, and experimental tractability make it an attractive system for exploring fundamental or potentially chordate-specific mechanisms of gene regulation and cellular morphogenesis (Bernadskaya and Christiaen, 2016). CRISPR/Cas9-mediated mutagenesis in Ciona has enabled targeted disruption of single genes with high efficiency in F₀ embryos, supporting both basic research and pedagogical applications (Johnson et al., 2023; Pennati et al., 2024).

With support from Arcadia Science, our group has used CRISPR/Cas9 in Ciona to investigate the developmental roles of candidate genes identified by Arcadia Science’s Zoogle platform (https://zoogle.arcadiascience.com/about) as Ciona orthologs of human disease-associated genes (Chou et al., 2025; Chou and Sun, 2025b). This report combines four installments of that ongoing project, previously published on Zenodo (Hernandez et al., 2025; Hernandez and Stolfi, 2026; Johnson et al., 2025; Johnson and Stolfi, 2025). The first three describe our investigation of three notochord- and endoderm-expressed genes (Fcho, Pgm3, and Nckap), from sgRNA design through phenotypic analysis of tail morphogenesis. The fourth describes a related, parallel investigation, inspired by an Arcadia Science collaboration with David Booth’s group at UCSF in the choanoflagellate Salpingoeca rosetta (Chou and Sun, 2025a), of the gene Plastin and its role in elongation of the sensory-adhesive papilla cells of the Ciona larva.

Targeting this small set of genes was intended as a pilot project to identify potential candidate genes for which Ciona might serve as a useful human disease model. The first step was to identify loss-of-function phenotypes by tissue-specific CRISPR. Here we describe in detail our CRISPR approach and report the initial characterization of these knockouts, which might inform future plans for using Ciona as a model to study the associated human disease genes. Of note, both Fcho and Pgm3 appear to be required for proper morphogenesis of the Ciona larval tail, while Plastin might be required in combination with another gene, Villin, for proper cell elongation.

METHODS, RESULTS, AND DISCUSSION

Fcho, Pgm3, and Nckap1 identified as target candidate genes

Fcho, Pgm3, and Nckap1 were identified as Ciona orthologs of human disease-associated genes and intriguing candidates for further investigation by Arcadia Science’s Zoogle platform (Figure 1A)(Chou et al., 2025; Chou and Sun, 2025b). In vertebrates, Fcho1 (FCH And Mu Domain Containing Endocytic Adaptor 1) encodes an endocytic adaptor involved in clathrin-mediated vesicle formation (Calzoni et al., 2019; Henne et al., 2010; Łyszkiewicz et al., 2020). Pgm3 (Phosphoacetylglucosamine mutase 3) participates in carbohydrate metabolism and glycosylation pathways critical for extracellular matrix integrity (Greig et al., 2007; Stray-Pedersen et al., 2014; Winslow et al., 2022; Yang et al., 2024). Nckap1l (Nck-Associated Protein 1-Like) is part of the WAVE regulatory complex, linking signaling pathways to actin cytoskeleton remodeling (Castro et al., 2020; Cook et al., 2020; Park et al., 2008). Disruption of these genes in humans is associated with diverse developmental and immune disorders, but we lack a more detailed tissue-level understanding of their cellular functions.

Figure 1.

A) Comparison of human disease proteins FCHO1, PGM3, and NCKAP1L and their Ciona orthologs, figure adapted from Chou and Sun 2025b, courtesy of Arcadia (Chou and Sun, 2025b). Trait distance and portfolio ranks are generated by Zoogle, with detailed methods explained in Chou et al. 2025. “Portfolio rank” is the relative rank of the Ciona ortholog compared to all other species’ orthologs (and sometimes paralogs). TM-score is a standard measure of protein structural similarity. See Chou and Sun 2025b for more detailed information about protein structure predictions. B) Single-cell RNAseq (scRNAseq) expression plot comparing the three genes’ expression across different tissue clusters in the whole developmental dataset from Cao et al. 2019, as visualized by Single Cell Portal. C) Similar plot comparing expression in the endoderm across developmental stages. D) Similar plot as in C, but for notochord expression across developmental stages. For plots B-D, the relative size of the dot is proportional to % cells expressing the gene, while the blue-red gradient indicates scaled mean expression from 0 (blue) to 1 (red).

As highlighted previously by both bulk and single-cell RNAseq (scRNAseq), all three appear to be expressed in the notochord and endoderm during Ciona embryonic development (Figure 1B–D)(Cao et al., 2019; Chou and Sun, 2025b). Only Pgm3 expression has been characterized by in situ hybridization, showing strong and specific upregulation in the endoderm and notochord (Reeves et al., 2017). Hence, we started focusing on a plan to knock these genes out in endoderm and notochord cell lineages and assay the effects on the proper morphogenesis of these cells and surrounding tissues during embryogenesis.

Designing sgRNAs to target Fcho, Pgm3, and Nckap1

For each target gene, three single-chain guide RNAs (sgRNAs) were selected (Table 1) using the CRISPOR website (https://crispor.gi.ucsc.edu/)(Concordet and Haeussler, 2018), following our established Ciona-specific strategies that have been previously published (Popsuj et al., 2024). CRISPOR is ideal for researchers working on Ciona due to its incorporation of single nucleotide polymorphism (SNP) data, allowing us to avoid potentially polymorphic target sites; Ciona populations are known to harbor extreme genetic diversity (Nydam and Harrison, 2010), and any mismatches between the sgRNA and the target sequence in the actual specimens used in the lab might result in low mutagenesis efficacies.

Table 1.

List of novel sgRNAs designed to target Fcho, Pgm3, and Nckap1, and their efficacy and specificity scores as predicted by CRISPOR.

sgRNA	Doench ’16	Doench-RuleSet3	MIT specificity
Fcho 1.34	57	23	100
Fcho 4.63	59	33	100
Fcho 5.112	63	−25	99
Nckapl 2.74	68	63	100
Nckapl 7.36	66	43	100
Nckapl 8.103	60	71	100
Pgm3 1.83	72	81	33*
Pgm3 2.27	60	24	50*
Pgm3 3.31	59	61	100

Asterisks denote “false” cases of low specificity due to confirmed genome assembly errors.

To search for potential sgRNAs, we entered each gene’s exons one at a time, starting from exon 1 upwards, until we obtained sgRNAs with high enough predicted efficacies and specificities. High efficacy is typically predicted by high “Doench ’16” score (Doench et al., 2016), specificity by an “MIT specificity” score >99 (Hsu et al., 2013). We have previously shown correlation between high, actual efficacy and predicted Doench ’16 efficacy scores over 50. Thus, our informal predicted efficacy (Doench ’16) score cutoff is typically ~50. A new predictive algorithm has been recently incorporated into CRISPOR, named “Doench RuleSet 3” (RS3)(DeWeirdt et al., 2022). There is some evidence that RS3 might be modestly better at predicting effective sgRNAs for CRISPR in Ciona (Popsuj et al., 2026), though this remains to be rigorously shown.

To ensure specificity, we aimed for MIT specificity scores >99. As the Ciona robusta genome is quite small and compact (~120 Mbp), scores of 99 and 100 are quite common and would rule out most potential off-target effects. We have previously detected a low (~5%) mutagenesis rate at a site with perfect sgRNA seed-target match but a non-canonical/alternative protospacer-adjacent motif (PAM)(Popsuj et al., 2024). However, CRISPOR did not allow us to automatically screen for those.

Candidate sgRNAs were chosen to maximize predicted mutagenesis efficiency while minimizing potential off-target activity. We also prioritized targeting sequences encoding the N-terminal region of the resulting protein, to maximize the likelihood of generating null alleles. We must balance the likelihood that a null, or “near-null” allele will result from our CRISPR-mediated indels, with the chance that cutting too close to the start codon will allow for translation of a mostly functional protein from a new, alternate start codon further into the coding sequence.

For Pgm3, we noticed abnormally low MIT specificity scores for sgRNAs targeting the first two exons, but we ascertained that this was due to older genome assembly errors that had erroneously duplicated those Pgm3 sequences. We were able to confirm that these earlier assembly errors were not present in the latest updates “HT” version of the Ciona robusta genome assembly (Satou et al., 2019), thus restoring our confidence in the specificity of these sgRNAs. Indeed, all the predicted off-targets flagged initially were in the incorrect assembly duplicate sequences.

Validating sgRNAs

All sgRNAs were tested according our established protocol (link to protocols.io)(Figure 2)(Popsuj et al., 2024). Briefly, each sgRNA is expressed using the U6 small RNA promoter, from plasmid co-electroporated into synchronized Ciona zygotes along with Eef1a>Cas9 plasmid. In this setup, Cas9 and the sgRNA are expressed in most cells. Larvae are then collected after hatching, comprising pools of hundreds or even thousands of F₀ CRISPants. Genomic DNA is extracted from each pool (representing a single sgRNA test), and amplicons centered on the sgRNA target site are obtained by PCR from the resulting genomic DNA. Amplicons (150-450 bp in size) are then sequenced by Illumina sequencing by Genewiz/Azenta, which provides automated indel analysis including estimated mutagenesis efficacy rate. Often the efficacy percentages presented are only approximated due to confounding naturally occurring SNPs and indels. This is because of the high genetic diversity in the wild Ciona populations we use for our studies.

Figure 2.

Diagram depicting our sgRNA validation workflow. Synchronized Ciona robusta zygotes (1-cell-stage embryos) are electroporated with plasmid DNAs to drive near-ubiquitous expression of an sgRNA and Cas9. The sgRNA is transcribed by RNA polymerase III via the small RNA promoter, while Cas9 expression is driven by the Eef1a promoter. Co-expression is denoted by purple coloring of cells. Mosaic uptake and/or retention of the plasmids results in some cells that do not express either. Embryos are then left to grow (typically to the larval stage) before pooling and genomic DNA extraction and downstream PCR and sequencing of target site amplicons (see text and linked protocol for details).

For Fcho, we tested three sgRNAs targeting exons 1, 3, or 5, surrounding the region that encodes a conserved FCH domain near the N terminus (Figure 3). The most efficacious sgRNAs by indel analysis were sgRNAs 1.34 and 5.112. The efficacy of sgRNA 1.34 was measured at 43%, which is on the higher end of the range of efficacies that we have measured for other sgRNAs in our lab. For sgRNA 5.112, an exact efficacy percentage was not obtained due to a confounding natural SNP, but it was well over 15%. Comparing to 1.34, it was likely to be closer to 30-40%. As such, these two sgRNAs were selected for future use, and sgRNA 4.63 was all but abandoned. When used in combination, sgRNAs 1.34 and 5.112 are also predicted to delete the entire FCH domain at the N terminus, which would further increase the odds of producing a null allele.

Figure 3.

Fcho sgRNA validation results

For Pgm3, sgRNAs 1.83 and 2.27 had the highest efficacies (Figure 4). Again, the exact efficacy for 1.83 was not measured due to a naturally occurring indel in the amplicon region, but was well over 12%. By looking at the height of the major CRISPR-induced deletion peak, its efficacy appeared roughly equivalent to that of sgRNA 2.27, which was measured at 23%. Both of these sgRNAs target the first two overlapping phosphoglucomutase/phosphomannomutase (PGM/PMM) domains, suggesting that using them in combination will likely produce null alleles. We therefore went ahead and selected these two for further use.

Figure 4.

Pgm3 sgRNA validation results

Finally, for Nckap1 (Figure 5), the presence of extensive natural indels and SNPs prevented an exact measurement of the efficacy of any of our three sgRNAs, but based on comparisons to the Pgm3-targeting sgRNAs, we believe that 2.74 and 8.103 were also in the 20-30% range. These target the 2^nd and 8^th exons of Nckap1, respectively. Unlike Fcho or Pgm3, the Nckap1 protein does not have any conserved domains other than the entire protein representing its own domain family. As such, it is not clear what parts of this large protein might be key for its functions, but using the above sgRNAs in combination, a large deletion at the N terminus should likely result in null alleles.

Figure 5.

Nckap1 sgRNA validation results

Phenotyping Fcho, Pgm3, and Nckap1 CRISPR knockouts in endoderm and notochord

Because all three targeted genes are expressed in the Ciona notochord during embryogenesis (Figure 1B.D) we first asked if any are required for Ciona notochord convergent extension, an essential process that occurs early during notochord development (Figure 6). This process involves the mediolateral polarization and migration of 40 post-mitotic cells that form two rows of 20 cells on either side of the midline in the developing embryo (Di Gregorio, 2020). These two rows of cells then intercalate to form a single row of cells. Eventually, through additional steps of elongation, vacuolization, and tubulogenesis, the notochord transforms into an extracellular fluid-filled tube that serves as a hydrostatic skeleton for the swimming Ciona larvae (Jiang and Smith, 2007).

Figure 6:

schematic of notochord morphogenesis in Ciona during embryogenesis.

With validated sgRNAs in hand, we used specific combinations of the most effective sgRNAs for each gene to target each one as follows. To knock out Pgm3, we used the combination of sgRNAs 1.83 and 2.27. For Fcho, we used the 1.34 and 5.112 sgRNA combination, and for Nckap1, we combined sgRNAs 2.74 and 8.103. These were previously identified as having the highest efficacies in creating insertion/deletion in our desired target regions (Johnson et al.). These data are summarized in Table 2 below.

Table 2:

Estimated efficacies of selected sgRNAs. See validation above for details.

sgRNA	Mutation efficacy (indel rate)
Pgm3.1.83	23%
Pgm3.2.27	>12.5%
Fcho.1.34	43%
Fcho.5.112	>15%
Nckap1.2.74	>15%
Nckap1.8.103	>15%

To assess the potential role of these genes in notochord cell intercalation, we performed CRISPR/Cas9-mediated mutagenesis in the Ciona early cell lineages that give rise to the notochord and endoderm. We used the dechorionation and electroporation protocol previously described by Christiaen et al (Christiaen et al., 2009a, b). Synchronized zygotes were electroporated with gene-specific combinations of plasmids encoding the sgRNAs described in Table 2, together with the Foxa.a>Cas9::Geminin^N-terminus plasmid to induce mutations early in endoderm and notochord lineages as previously described (Di Gregorio et al., 2001; Popsuj et al., 2023). We allowed the embryos to develop to the mid tailbud I stage (Hotta Stage 21, roughly 9.5 hours post-fertilization when reared at 20°C) when intercalation is complete in normally developed embryos. The fluorescent reporter plasmids Brachyury>CD4::GFP (encoding cell membrane-bound GFP), Brachyury>Unc-76::GFP (encoding GFP tagged with a portion of the Unc-76 protein from C. elegans, which promotes its cytoplasmic localization), and/or Brachyury>H2B::mCherry (encoding histone-fused mCherry) were also used to visualize the notochord cell shape and nuclei, respectively (Corbo et al., 1997). A control experiment was performed using the same plasmids, using a U6>control sgRNA plasmid targeting no Ciona sequence instead of the sgRNA plasmid combination for each gene. Per 700 μl of electroporation volume, we used 80 μg of sgRNA plasmid (40 μg of each when combined), 35 μg of Cas9 plasmid, and 40 μg of each green or red reporter plasmid.

Embryos in control and knockout conditions were imaged using an inverted epifluorescence microscope and scored as either normal intercalation, minor defect in intercalation (1-2 cells out of midline) and severe defect in intercalation (3+ cells out of midline). Examples of normal and severe defects in three different conditions are shown in Figure 7. Experiments were performed in three separate biological replicates using different batches of dissected gametes, with 75 individuals scored per condition and replicate. Raw images and scoring data for Pgm3 and Fcho CRISPR embryos (but not Nckap1 CRISPR) can be found here: https://zenodo.org/records/17857562

Figure 7:

Representative epifluorescence microscopy images of early tailbud embryos showing examples scored as “normal” intercalation (top) or those with severe intercalation defects (bottom). Control = negative control CRISPR condition. Dashed outlines indicate embryo outline. Anterior always to the left or top/bottom-left. Dorsal always over ventral. All scale bars = 50 μm. See text for details about type and amount of plasmids electroporated.

Three replicates of the Pgm3 knockout were performed, and the scoring data are summarized in Figure 8. In the first replicate, we observed severe notochord intercalation defects in 51% of embryos in the Pgm3 knockout condition, compared to 23% of negative control embryos. This initially suggested that Pgm3 could play a role in proper notochord intercalation. We performed an additional two replicates of this experiment. However, in these additional experiments, we observed roughly equal percentage of defects in both control and knockout conditions (~50% and ~30% defects in replicates two and three respectively). Since we could not replicate our initial results showing an increase in intercalation defects in Pgm3 knockout condition compared to the control, we cannot conclude that Pgm3 is vital for proper intercalation of notochord cells.

Figure 8:

Scoring of intercalation defects in mCherry+ tailbuds in control and Pgm3 knockout conditions in three separate replicates. Replicate 3 used Unc-76::GFP instead of CD4::GFP.. n = 75 for each condition and replicate. Control = negative control CRISPR condition.

Similarly, for Fcho CRISPR we performed the same intercalation assay over two replicates to determine if this gene was playing a role in this process of notochord morphogenesis. Scoring results are summarized in Figure 9, showing a slight increase in defects in the knockout in replicate one, though barely any increase in replicate two. Therefore we cannot conclude that Fcho is required for proper intercalation either. Finally, in a single replicate for Nckap1 CRISPR as well, and in the first replicate, we also saw similar percentage of defects in the control vs knockout conditions, and these data are summarized in Figure 10.

Figure 9:

Scoring of intercalation defects in mCherry+ tailbuds in control and Fcho knockout conditions over two replicates. n = 75 for each condition and replicate. Control = negative control CRISPR condition.

Figure 10:

Scoring of intercalation defects in mCherry+ tailbuds in control and Nckap1 knockout condition in one replicate. n = 75 for either condition. Control = negative control CRISPR condition.

Overall, these results do not suggest that these genes play a significant role during the early stage of intercalation in the notochord. However, we cannot rule out minor roles either. As can be seen from the variation in the incidence of severe defects among replicates, even in negative control embryos, it is clear that intercalation is a sensitive process that is especially susceptible to non-specific effects. These effects can arise from the myriad steps involved in generating tissue-specific CRISPR knockouts in F0 embryos, namely the chemical and enzymatic removal of the chorion (egg shell), or the transfection with plasmid DNAs by whole zygote electroporation. Intercalation may also be sensitive to environmental and/or seasonal conditions, as gametes are obtained from adult specimens collected weekly from various locations in the wild. Nonetheless, even with a higher baseline incidence of intercalation defects in some control samples, we would expect an increase in defects in the parallel CRISPR samples, if these genes were heavily involved in this process.

Pgm3 and Fcho are required for proper elongation of the larval tail

Next, we sought to assess the potential role of these genes in notochord elongation and tubulogenesis, using a similar tissue-specific CRISPR/Cas9-mediated mutagenesis approach as for the intercalation experiments above (Popsuj et al., 2024). Per 700 μl of electroporation volume, we used 80 μg of sgRNA plasmid (40 μg of each when combined, using same combinations as for intercalation assays), 35 μg of Cas9 plasmid, and 20 μg of Brachyury>H2B::mCherry. Instead of a green fluorescent reporter plasmid, we aimed to use fluorescent phalloidin to label the outline of cells to more easily assay vacuolization.

Larvae were raised to 17.5 hpf at 20°C, fixed, stained with fluorescent phalloidin (conjugated to AlexaFluor 647, from ThermoFisher), and mounted on glass slides for microscopy. Larvae were imaged using a scanning point confocal microscope (Zeiss), as well as using an epifluorescence microscope (Leica) for scoring and tail length measurements. Confocal image Z-stacks were processed in Fiji/ImageJ, while tail measurements were performed in Leica LASX software. Tails lengths were measured roughly along the anterior-posterior axis, and larvae were also qualitatively scored for wild-type or truncated tails. Raw data and images can be found at https://zenodo.org/records/18623201.

When compared to negative control larvae, Pgm3 CRISPR and Fcho CRISPR larvae had visibly shortened, truncated tails. In contrast, Nckap1 CRISPR larvae tails did not appear to be significantly different from that of negative control larvae. This was observed both through measurements of individual tail lengths (Figure 11A), as well as qualitative assessment of tails that appeared truncated (Figure 11B). Notochord cells were still arrayed more or less linearly in Pgm3 and Fcho CRISPR larvae, suggesting that earlier intercalation was not affected as much as later processes of notochord cell or lumen elongation, and/or elongation of surrounding tissues such as tail muscles or endodermal strand (Figure 12). Pgm3 CRISPR appeared to have the largest effect, as the reduction in tail length was most pronounced in this condition (Figure 11A), though qualitatively both Pgm3 and Fcho CRISPR conditions were comparable as far as frequency of tail elongation defects (Figure 11B).

Figure 11:

Effect of Pgm3/Nckap1/Fcho CRISPR on tail elongation.

A) Plot showing tail measurements of individual larvae. n = 40 for each condition. Statistical analysis was performed as ordinary one-way ANOVA with Dunnett’s multiple comparison tests. ****: p < 0.0001, *: p = 0.0324, ns: p = 0.0789. B) Plot showing scoring same larvae as in panel A, for normal/wild type tails and for truncated tails. n = 40 for each condition. Fisher’s Exact Test was performed comparing each CRISPR condition to the negative control. ****: p < 0.0001, *: p = 0.0016, ns: p = 0.4391. Only larvae expressing Brachyury>H2B::mCherry were imaged/scored.

Figure 12:

Pgm3 CRISPR larvae.

Representative confocal image slices of Pgm3 CRISPR larvae of varying phenotype severity. Inset: notochord cells have intercalated but cell elongation and vacuole/lumen expansion is blocked. In green is AlexaFluor-647::phalloidin.

Overall, these suggest the Pgm3 and Fcho might be required for tail elongation during Ciona embryogenesis, while Nckap1 appears mostly dispensable (pending biological replicates). Our findings do not rule out a role for Nckap1 in other developmental processes not assayed here, or in Foxa.a-negative tissues. The more pronounced effect of Pgm3 CRISPR relative to Fcho CRISPR could be due to one or multiple factors, regardless of the gene products’ biological functions. Pgm3 is the more highly and widely expressed gene according to single-cell RNA sequencing (scRNAseq) data (Figure 1B–D) and is upregulated in the endoderm and notochord during tail elongation according to mRNA in situ hybridization (Reeves et al., 2017). While the expression pattern of Fcho has yet to be determined by in situs, by either scRNAseq or bulk RNA sequencing its expression appears lower and relatively sparse (Figure 1B–D) Therefore, Pgm3 might be rate-limiting in more cells crucial for tail morphogenesis in Ciona.

Future work will be needed to determine the tissue-specific requirements for these genes, especially Pgm3, which is highly expressed in both endoderm and notochord. The key will be to knock out Pgm3 only in the endoderm or the notochord and assay tail elongation. It is possible that the severe defect we describe here might be due primarily to disrupting Pgm3 in the endoderm, or in the notochord, for instance. To do this we can use endoderm- or notochord-specific promoters to drive Cas9 expression in only one tissue at a time.

Future work is also required to elucidate the molecular mechanisms of Pgm3 and/or Fcho in tail elongation. In mammals, Pgm3 is vital for protein glycosylation. Previous work has shown that protein glycosylation is required for proper notochord lumen expansion in Ciona (Wang et al., 2023). We hypothesize that Pgm3 in the endoderm and/or notochord may be required for glycosylation of extracellular matrix and/or cell surface receptors involved in tail elongation and lumen expansion. Proteomic analysis of Pgm3 CRISPR larvae might reveal such downstream glycoproteins. In fish, Fcho1 interacts with BMP receptors, and fcho1 morphants show notochord and tail elongation defects as result of BMP signaling perturbation (Umasankar et al., 2012). BMP plays several roles in patterning the Ciona embryo, particularly in the tail (Christiaen et al., 2010; Lanoizelet et al., 2024; Liu et al., 2023; Pasini et al., 2006; Roure et al., 2023). Other roles for BMP or TGF-family signaling pathway in specifically notochord development and tail elongation have yet to be fully demonstrated, though Fcho could be generally affecting other receptors in an analogous manner in Ciona.

Investigating a role for Plastin in papilla morphogenesis

While we were focused on knocking out Fcho, Pgm3, and Nckap1 in the notochord and endoderm, we were inspired by a parallel collaboration between Arcadia and David Booth’s group at UCSF (Chou and Sun, 2025a). Zoogle identified the gene Plastin 1 (PLS1) as an exceptional candidate to study in the choanoflagellate Salpingoeca rosetta based on predicted structural similarity. This gene encodes an actin-bundling protein, and associated mutations are implicated in hearing loss in humans, likely through effects on mechanosensitive hair cell developmental defects (Morgan et al., 2019; Taylor et al., 2015).

In Ciona, we previously identified a PLS1 homolog (KH gene ID KH.S521.5, KY21 gene ID KY21.Chr1.1966) as being upregulated in the developing sensory-adhesive papillae of the larva (Figure 13)(Johnson et al., 2024), which is a set of three protruding clusters of cells that are elongated along the apical-basal axis (Zeng et al., 2019). Of note, we found that this gene (which we call herein Plastin) is upregulated by the transcription factor Islet, which is crucial for the elongated cell morphology of the papillae (Johnson et al., 2024; Wagner et al., 2014). We therefore had previously hypothesized that Ciona Plastin might be important for regulating actin dynamics during papilla cell elongation.

Figure 13.

Whole-mount, fluorescent mRNA in situ hybridization for Plastin in Ciona robusta embryos, revealing strong expression in the developing papilla region (arrowheads). Expression can also be seen in the endoderm of the head/trunk and tail (endodermal strand), especially at mid-tailbud stage.

Zoogle did not suggest Ciona Plastin as a particularly close structural match for human Plastin 1, coming in at a very middling portfolio rank of 29 (out of 50). However, we noticed that the particular Uniprot sequence of Ciona Plastin used by Zoogle came from an ENSEMBL gene model that is incomplete, missing a substantial portion of the exons encoding the conserved C-terminus (Figure 14). We we therefore curious whether, in reality, Ciona Plastin might have scored higher and could serve as a good model for studying conserved Plastin functions. Here we describe our first attempt to knock out Plastin in the papillae of the Ciona larva to potentially reveal resulting defects in cell elongation.

Figure 14.

A) Diagram of KY21 (top, blue) and ENSEMBL (bottom, orange) gene models, showing that the ENSEMBL model is likely incomplete. B) Diagram comparing the predicted domain structures of the KY21 Plastin protein sequence (top) and the UniProt sequence (bottom, based on the ENSEMBL gene model). The third Calponin Homology (CH) domain is truncated, and the fourth CH domain is missing, in the UniProt sequence. EFh = EF-hand, calcium-binding motif. Protein domains analyzed by SMART (Letunic et al., 2021).

Plastin CRISPR approach

Several single-chain guide RNAs (sgRNAs) targeting Plastin were designed and validated by Illumina-based amplicon sequencing followed our established routine protocol (Popsuj et al., 2024). Of these, we selected the sgRNAs shown in Figure 15 for use in combination to knock out Plastin by tissue-specific CRISPR/Cas9-mediated mutagenesis. Figure 3 plots indicating successful targeting.

Figure 15.

Top: Plastin sgRNAs selected after validation. Bottom: Insertion/deletion (indel) plots generated by Azenta/GeneWiz’s “Amplicon-EZ” illumina sequencing service. Black arrows indicate indels generated by CRISPR/Cas9 using the indicated sgRNAs. Red asterisks indicate naturally-occurring indels.

Tissue-specific CRISPR was carried out similarly as above, but using the Foxc promoter to drive expression of Cas9::Geminin^N-terminus in the papilla lineage (Johnson et al., 2024; Pennati et al., 2024). Embryos were co-electroporated with Foxc>H2B::mCherry reporter plasmid to label the cell nuclei of the CRISPR’d lineage, as well as a CryBG>Unc-76::GFP reporter plasmid to label the cytoplasm of the axial columnar cells (ACCs), the central, most elongated cell type of the papillae (Johnson et al., 2020; Shimeld et al., 2005). The electroporation recipes (per 700 μl of electroporation volume) were thus: 10 μg Foxc>H2B::mCherry, 35 μg Foxc>Cas9::GemininN-terminus, 35 μg CryBG>Unc-76::GFP, and 80 μg total of sgRNA plasmid (40 μg each Plastin sgRNA, or 80 μg of the single negative control sgRNA).

Larvae were fixed at 17 hpf (20°C) and imaged using a Leica DMI8 inverted epifluorescence microscope. Raw images can be found at https://zenodo.org/records/17892564. ACC lengths were measured along the apical-basal axis using LASX software, following the previously established method (Johnson et al., 2024). Raw lengths were plotted and analyzed in GraphPad Prism.

We observed that ACC length was not significantly reduced upon Plastin CRISPR in the papilla lineage (Figure 16A,B). This was not entirely surprising, as previous monogenic knockouts of other genes encoding cytoskeletal regulators downstream of Islet (e.g. Villin, Tubulin alpha) failed to result in significant, consistent shortening of the papillae (Johnson et al., 2024). This in spite of Islet knockouts or dominant-negatives showing significantly shortened papillae (Johnson et al., 2024; Wagner et al., 2014).

Figure 16.

A) Images of the papillae in 17 hpf (20°C) larvae, comparing the negative control CRISPR and Plastin CRISPR conditions. Yellow arrow indicates axial columnar cell (ACC) length along their apical-basal axis, which was measured for this assay. Larvae counterstained with DAPI (white overlay). B) Quantification of ACC cell lengths, showing a slight but not statistically significant drop in average length upon tissue-specific Plastin CRISPR knockout. ns = not statistically significant, Welch’s t-test (two-tailed). C) In contrast there was a statistically significant (p = 0.0032 by two-tailed Welch’s t-test) difference in cell lengths between double Plastin/Villin CRISPR.

We hypothesize that multiple genes downstream of Islet may be acting in a partially overlapping or redundant manner to regulate the cytoskeleton during papilla cell elongation. For instance, Villin (KY21.Chr9.368) encodes another actin-bundling protein and therefore might be compensating for loss of Plastin. In fact, Villin is expressed later, specifically in elongating central Islet+ cells that give rise to the ACCs (Johnson et al., 2024). The expression of Plastin during tailbud stages throughout the entire papilla territory (Figure 14) suggests it may be important for the earlier thickening of the papilla epithelium. Thus, it is possible that Villin and Plastin may combine in partially overlapping ways to carry out different steps in papilla morphogenesis. In mammals, a triple knockout of actin-bundling factors (Villin, Plastin1, and Epsin) did not prevent intestinal microvillus formation but reduced their length (Revenu et al., 2012).

With this in mind, we sought to perform tissue-specific, double CRISPR knockouts in the papillae of Ciona. For this, we used 80 g of the above Plastin-targeting sgRNAs combined with 35 g each of previously validated and published Villin.4.74 and Villin.5.105 sgRNAs (Johnson et al., 2024). When we combined both Villin- and Plastin-targeting sgRNAs (raw data accessible here: https://zenodo.org/records/21840646), we obtained a statistically significant reduction in ACC length compared to the negative control (Figure 16C). This was also in contrast to the previous attempt at knocking out Villin with three sgRNAs, which did not reproducibly shorten cell length (Johnson et al., 2024). Taken together, these data suggest Plastin and Villin may be acting in partially redundant ways to support cytoskeletal dynamics underlying the elongated axes of protruding papilla cells in Ciona.

CONCLUSIONS

Here we have described the initial characterization of three genes in Ciona pinpointed by Zoogle as highly similar to their human candidate disease gene orthologs (Fcho, Pgm3, Nckap1), and one that we believe may be a good match based on updated, full-length protein sequences (Plastin). Tissue-specific CRISPR knockouts of three out of these four resulted in a developmental defect. Of these three, we believe Pgm3 to be the most promising one as its disruption had the most profound, obvious, and penetrant effect, resulting in a robust phenotype that would be amenable to higher-throughput screening. Fcho CRISPR resulted in a similar, but less profound defect, while an even more subtle phenotype for Plastin CRISPR was only unmasked by concurrent knockout of Villin. Using Ciona as a model to study Pgm3 function may be further enhanced by its potential role upstream of protein glycosylation. In the future, it may be attractive to use proteomics to identify candidate targets of Pgm3 in the Ciona notochord or endoderm, and devise rescue experiments aimed at identifying potential therapeutics for treating Pgm3-related human diseases. Our lab continues to use Zoogle to identify other promising disease gene candidate orthologs that may be amenable to studying in Ciona embryos.