We obtained the zygotes of sterlet (Acipenser ruthenus Linnaeus, 1758) from the adults kept and regularly bred at the Research Institute of Fish Culture and Hydrobiology, Faculty of Fisheries and Protection of Waters, University of South Bohemia in České Budějovice, Vodňany, Czech Republic (RIFCH). The husbandry, animal conditioning, gamete collection, and fertilization were described in detail by Chebanov and Galich (2011) and Saito et al. (2014). The adult breeding fish were handled under anesthesia in 0.05% tricaine. Briefly, both female and male adult sterlets, aged five to 9 years, were transferred from the outdoor ponds into 4,000 L indoor tanks with water temperature kept at constant 15°C. Spermiation in males was induced by intramuscular injection of carp pituitary extract at 4 mg/kg body weight in 0.9% NaCl solution. Sperm was collected 48 h later via a 0.6 mm catheter into cell culture flasks and kept on ice until in vitro fertilization. Female ovulation was induced in a similar way, but with two doses of the hormone instead of only one injected 12 h apart −0.5 and 4.5 mg/kg of body weight, respectively. Females ovulated 18–20 h after the hormone was administered. Ovulating females were placed in dorsal recumbency position and oviduct incision was performed using an eye microsurgery scalpel. This minimal invasive procedure allows accessing the ovulated eggs in sturgeon oviduct that is physiologically folded (see Chebanov and Galich, 2011). Then the female’s abdomen was massaged in anterior-to-posterior direction and eggs were collected into bowls (Figures 1B,C), sealed with aluminum foil and subsequently fertilized with sperm at 15°C in dechlorinated tap water. We only used sperm with the spermatozoa motility assessed at >80% (Chebanov and Galich, 2011). The zygotes were rinsed in 0.04% tannic acid to make them less adherent (Saito et al., 2014), and then kept in water at 15°C before they were used in experiments. After the injections, we transferred the embryos to 48-well plates, each in 1 ml volume of E2 zebrafish medium (Brand et al., 2002) containing antibiotics (120 ng/ml of penicillin and 200 ng/ml of streptomycin) kept at 15°C. The medium was changed daily or more often, if necessary. Embryos selected for raising to later developmental stages were transferred to well-oxygenated tanks with E2/antibiotics at 15–17°C shortly after they hatched, which typically happened 7 days post fertilization (st. 35). We staged the embryos and larvae using the staging system of Detlaff et al. (1993). When the embryos and larvae reached the desired developmental stage, they were anesthetized by tricaine (MS-222; Serva) and fixed in 4% PFA in PBS overnight at 4°C. After several washes in PBS, we gradually dehydrated the embryos and larvae through a series of PBS/methanol solutions and stored in 100% methanol at −20°C until further use.

Animal care and all experiments were approved by the Ministry of Agriculture of the Czech Republic (MSMT-12550/2016-3), followed the principles of the European Union Harmonized Animal Welfare Act of the Czech Republic, and Principles of Laboratory Animal Care and National Laws 246/1992 “Animal Welfare”, and were conducted in accordance with the Animal Research Committee of RIFCH. Authors of the study own the Certificate of professional competence for designing experiments and experimental projects under Section 15 d (3) of the Czech Republic Act no. 246/1992 Coll. on the Protection of Animals against Cruelty.

We used similar F0 mutagenesis strategy as was described for sea lamprey (Petromyzon marinus) and African clawed frog (Xenopus laevis) (Square et al., 2015; Square et al., 2020) (Figure 1D). Using the spotted gar (Lepisosteus ocelatus) and zebrafish (Danio rerio) tyrosinase and sonic hedgehog protein sequences as queries we searched sterlet Tyr and Shh homologs in our de novo assembled transcriptomes of sterlet pharyngulae (available online at https://www.researchgate.net/profile/David-Jandzik/projects). The identity of recovered sequences was checked with BLAST (Altschul et al., 1990). We targeted the protein-coding exons at loci showing high evolutionary conservation across vertebrate taxa. In both tyr and shh we identified four putative exons. We selected the best CRISPR/Cas9 target sites with sequence 5′-GG (18N) NGG-3′ and with no off-target matches to our transcriptomes. The individual sequences were visually checked and identified as off-targets if they showed more than 85% similarity to our candidate sequence including PAM by BLAST (0–3 mismatches) and the mismatches were close to PAM site (more than one mismatch closer to PAM than 10 bp). The sequences of target sites were as follows: tyr sgRNA 3: GGT​TAG​AGA​CTT​TAT​GTA​AC (GGG), tyr sgRNA 4: GGC​TCC​ATG​TCT​CAA​GTC​CA(AGG), shh gRNA1: (CCC)CAA​TGT​GGC​CGA​GAA​GAC​CC, shh gRNA2: GGG​CCA​GTG​GCA​GAT​ATG​AA(GGG) with PAM sites in parentheses and the PCR primers used to amplify the target sequences in Table 1. The amplified fragments were annealed and in vitro phosphorylated with T4 Polynuclease Kinase (NEB M0201S) at 37°C for 1h, and ligated into the DR274 plasmid (Hwang et al., 2013) pre-digested with BsaI (NEB R0535). Single guide RNAs were in vitro transcribed using T7 High Yield Kit (New England Biolabs) and purified by phenol-chloroform extraction followed by precipitation in 70% ethanol with 0.3M sodium acetate. The precipitate was resuspended in nuclease-free water.

We prepared a fresh injection mix on ice shortly before each injection session. The commercially produced recombinant Cas9 protein (PNA Bio Inc.) was resuspended per manufacturer’s instructions to the stock concentration of 1 mg/ml, aliquoted, and stored at −80°C. For a 6 μl injection mix we first incubated 1.6 μg of diluted Cas9 with 800 ng of total sgRNA for 10 min on ice, then brought the total volume up to 5.5 μl with nuclease-free water. Approximately 0.6 μl of 50 μg/μl lysinated Rhodamine-dextran (LRD; Invitrogen) in nuclease-free water was then added, resulting in a final LRD concentration of 5 μg/μl. One-cell-stage embryos were manually dechorionated using Dumont forceps and positioned in shallow holes in modeling clay in a Petri dish to facilitate their proper orientation and stability. The microinjection was performed either with mouth or manual injector (set to 100 hPa for 1 s) using microcapillary needles (Drummond Microcaps) pulled in a Narishige pc-10 puller (58°C with two weight elements; diameter 1.02 mm). The needle tip diameter was adjusted to allow to produce a ∼20 nl drop (approximately 1/7 of the sterlet zygote in diameter) containing around 2.67 ng of sgRNA and 5.3 ng Cas9 protein (in 1:2 wight ratio) in total. While injecting, we targeted the animal pole of the embryo at a ∼45° angle (Figure 1D). After injection the embryos were transferred to 48-well plates with E2/antibiotics at 15°C. We screened the embryos for LRD at the neurula stage (ca. st. 21). To control for mortality rates, we injected in parallel a few batches of embryos with a non-functional sgRNA at the same mix composition and concentrations as the experimental mixes. Each sgRNA was injected multiple times, in eggs from multiple clutches and by several authors of this study.

We assessed the efficiency of tyr mutagenesis in injected larvae raised approximately to 16 mm of total length, when pigmentation is visibly present and conspicuous on the head and body of the wild-type individuals. Based on the severity of pigmentation reduction we scored the observed phenotypes in four categories as 0–25% reduction, 25–50% reduction, 50–75% reduction, and 75–100% reduction. We also checked the larvae for any non-specific malformations and deformities. The embryos injected with shh sgRNAs with Cas9 protein were fixed at the pharyngula stage (st. 28), when several organs regulated by hedgehog signaling have formed. These include heart, pre-oral gut, olfactory and optic placodes, first and second pharyngeal pouches, fore-, mid-, and hindbrain, somites and pronephros. Rather than scoring the global phenotypes of the embryos and comparing the severity of the phenotypes among different characters, we scored each structure separately, recording whether it was present or reduced/deformed. We also used in situ hybridization to visualize changes in expression patterns of neural crest markers foxd3 and twist1 and pharyngeal pouch formation marker ripply3 in mutant embryos.

To confirm mutations of targeted loci, we genotyped selected sterlet embryos and larvae showing variable phenotypes; at least two individuals per sgRNA. Due to higher percentage of individuals showing no obvious reduction in pigmentation, i.e. wild-type-looking phenotype in tyr sgRNA injected larvae, we also genotyped a few of those to obtain a better picture of tyr sgRNA efficiency. In tyr sgRNA injected larvae we used tail clips obtained from freshly euthanized larvae, while in shh mutants we used the whole embryos after we photographed and analyzed their phenotypes or gene expression patterns. We digested the tissue with Proteinase K (80 IU/ml; Sigma-Aldrich) in 1X PCR buffer at 55°C for 12 h. The obtained genomic DNA extracts served as templates in PCR amplification reactions using GoTaq polymerase (Promega) with amplification primers listed in Table 1. The PCR program followed the manufacturer’s recommendations with the primer-specific annealing temperatures of 57.5° for both loci. We subcloned the resulting amplicons into pJet1.2 plasmid (Thermo Fisher) and fragments obtained from purified colony PCR reactions were sequenced using M13 forward and reverse primers. The alignments of mutant and wild-type sequences were prepared manually. We considered an individual to bear a mutant genotype if at least two sequences represented mutant alleles (see Table 4).

Whole mount in situ hybridization (ISH) was carried out as described in detail by Minarik et al. (2017). We retrieved the putative sterlet homolog of ripply3 from our pharyngula transcriptomes. The 317-bp long DNA template sequence was PCR amplified from sterlet cDNA (amplification primers in Table 1) and subcloned into pGEM T-Easy vector (Promega) by standard procedures. Acquisition of foxd3 and twist1 sterlet sequences was described by Stundl et al. (2020). ISHs using injected and wild-type embryos were performed in separate tubes though in parallel and under the precisely same conditions to avoid variations in probe penetration and signal development.

All photographs of sterlet embryos and larvae in PBS were taken with Olympus SZX12 stereoscopic microscope using z-stacking Deep Focus technology of QuickPhoto software (Promicra).

We designed and synthesized two sgRNAs targeting two different exons of the sterlet homolog of human Tyr1 - sgRNA three and four against exons 2 and 3, respectively, sequences of which allowed us to design sgRNAs according to our criteria (see Methods and Square et al., 2015; Square et al., 2020). First, we co-injected 50 one-cell sterlet embryos with a mix containing both guides at a total amount of ∼2.67 ng of sgRNA and 5.3 ng of Cas9 protein (in 1:2 wight ratio) per single embryo. The total amount was calculated to match the amount of guide RNA and Cas9 protein relative to the egg size successfully used in Xenopus laevis CRISPR/Cas9 mutagenesis (Square et al., 2015; Square et al., 2020). In total, 32 injected embryos were LRD positive (i.e., with the lysinated Rhodamine-dextran of the injection mix glowing under the fluorescent light) at the neurula stage, however, no larva showed reduction in pigmentation or other developmental malformation. We suspected that this could have resulted from concentration of each individual sgRNA being too low in this combined sgRNA mix and therefore we used the same total amount of the injection mix but with each sgRNA separately in the next experiment. Both these experiments produced albinos with the variable extent of pigment reduction and normal morphology lacking any visible morphological abnormalities. We visually assessed the extent of pigment reduction and scored the phenotypes to four classes at increments of 25% pigment reduction (Table 2). Injection with sgRNA three resulted in >70% partial or total albinism occurrence, while treatment with sgRNA four produced >50% partially or completely albinotic larvae. Mutagenesis with sgRNA three was more efficient in producing phenotypes with stronger pigment reduction, although the numbers of complete or near-complete albinos were similar between the guides. We recorded 25–33% mortality, which was similar to the uninjected dechorionated wild-type larvae raised to the same stage (Table 2) and to the mortality of embryos injected with non-functional sgRNA (14/50 = 28%).

We verified the disruption of the target tyr loci by genotyping selected individuals from each phenotype class (Figure 2, Supplementary Figure S1). We found mutant alleles in 7/10 and 4/7 genotyped larvae generated with sgRNA 3 and 4, respectively, and confirmed mutants in all phenotype classes. Approximately half of all obtained sequences were mutant. We observed 27 unique mutant alleles (out of 36 sequences), 22 of which had only deletions, one had only insertion and four had both deletions and insertions (Supplementary Figure S1). While we did not observe a correlation between the relative occurrence of WT vs. mutant alleles and severity of the observed phenotypes, it appears that the stronger phenotypes have more alleles with indels that cause frameshifts (Figure 2, Supplementary Figure S1).

Similar to our initial experiments with tyr mutagenesis, we first injected 50 single-cell sterlet embryos with a mix of both guides targeting shh locus with the total amount of ∼2.67 ng of sgRNA and 5.3 ng of Cas9 protein per embryo. We recorded relatively high mortality (36/50 = 72%), but the surviving LRD positive embryos (13; one was LRD negative) showed no discernable phenotypic effect. Next, we injected each sgRNA separately at the same total amount and sgRNA/Cas9 ratio. Due to high mortality in the initial experiment, we injected higher numbers with each guide in a separate mix - 191 with sgRNA 1 and 117 with sg RNA 2. We observed mortality of 34 and 38%, respectively (64/191 individuals in sgRNA one and 44/117 in sgRNA two injections) and obtained 93 and 57 LRD positive embryos that reached developmental st. 28. The embryos were severely affected with strong morphological deformations in almost all embryonic structures present at that stage (Table 3). The highest frequency phenotypes included missing pharyngeal pouches and pre-oral gut (up to 81%), followed by missing, reduced or malformed optic placode, not discernible heart, pronephros reduction and brain malformations. Such deformations were consistent with patterns of shh gene expression observed during the sterlet embryonic stages preceding the analyzed stages (Figures 3A–C). Only 7–10% of injected LRD positive individuals appeared to have no visible phenotype (9/93 and 4/57 for sgRNA 1 and 2, respectively). We did not observe any of the described phenotypes in the control embryos and their mortality was similar to the mortality in tyr mutagenesis experiment.

Next, we analyzed effects of shh disruption on expression patterns in injected sterlet embryos by ISH. We examined three genes known to be involved in development of some of the affected structures at early embryonic stages: foxd3 marking NCCs at early migration stage, twist1 NCCs at later stages of migration, and ripply3 in forming pharyngeal pouches (Figure 4, Supplementary Figure S3). We found that all these markers were reduced in the injected sterlets at st. 28. foxd3 showed a reduction in the size of all three streams of migrating NCC (trigeminal, hyoid, and branchial), twist1 expression was reduced in all later NCC populations (maxillary, mandibular and branchial) and even entirely missing in some embryos. Only three out of 14 hybridized embryos appeared to have a wild-type expression pattern of the NCC markers. ripply3 ISH confirmed our observation of the strong effect of shh mutation on pharyngeal pouch formation in putative shh mutants; all seven analyzed embryos had either missing or reduced expression domains (Table 3).

To verify mutagenesis in shh locus in analyzed sterlet embryos, we genotyped selected individuals injected by both sgRNAs, including those used for ISHs. All 33 obtained sequences had indels and consequently all nine genotyped individuals were confirmed mutants. We identified 21 unique alleles in total, five of which only showed insertions, 12 of them contained deletions and four had both insertions and deletions (Table 4). The majority of the observed mutations caused frameshifts; these were observed in at least half of all obtained sequences of each mutant individual. Only 8/26 sequences of mutations introduced by sgRNA one were in-frame (Figure 4, Figure 3, Supplementary Figure S2, Table 4).

Here we report the application of a CRISPR/Cas9 mutagenesis strategy previously optimized in the sea lamprey (Square et al., 2015) and African clawed frog (Square et al., 2020), to the sterlet. Previous reports using TALENs and CRISPR/Cas9 suggested sterlet zygotes will tolerate injection with proteins and nucleic acids and were amenable to gene targeting (Chen et al., 2018; Baloch et al., 2019). We found that our optimized protocol introduced biallelic mutations in the targeted loci at relatively high efficiency and yielded consistent and reproducible phenotypes. As expected, targeting of the gene tyr, resulted in larval albinism with no other discernible developmental defects. These mutants have similar mortality as their uninjected wild type siblings. The visual scoring of the larval phenotypes is straightforward, suggesting that tyr targeting could serve as a robust positive control in future studies. In contrast to tyr disruption, mutating the pleiotropic developmental regulator shh caused defects in a wide range of embryonic and larval tissues and structures. Interestingly, both sgRNAs used to target shh resulted in all injected individuals displaying defects consistent with the known roles of shh in vertebrate development. This demonstrates that determining the function of developmental regulators, even highly pleiotropic ones, is readily achievable in sterlet. Arguably, in CRISPR/Cas9 sterlet F0 mutants it could be more challenging to detect and score very subtle and less conspicuous phenotypes than in models with stable inbred lines.

Though sterlet is not well suited as a laboratory-propagated model organism, several features of its natural history make it a strong candidate for routine genetic analyses of “F0” mosaic mutants. Along with other sturgeon species, it is a frequently farmed fish, whose eggs can be seasonally obtained in large quantities. The eggs are relatively large and can be easily manipulated, dechorionated, and injected with just forceps and also simple hand-held and mouth operated injector. The embryos and larvae thrive in simple aquaria with no special care requirements beyond clean oxygenated water and a constant temperature of 15–17°C. Several common laboratory methods have also already been optimized in the species including histological sectioning, microCT scanning, RNA in situ hybridization, lineage tracing following injections, and immunohistochemistry (Minarik et al., 2017; Stundl et al., 2020).

Other features of sterlet, and acipenseriform fishes in general, were likely important for the successful application of our CRISPR/Cas9 mutagenesis strategy. Sterlet eggs are rich in yolk that is distributed throughout the egg (Dettlaff et al., 1993). The eggs of lamprey and frog, which are highly amenable to CRISPR/Cas9 mutagenesis (Square et al., 2015; Square et al., 2020), are similarly rich in yolk. Since as in frog, the sterlet egg yolk is slightly more concentrated on the vegetal pole, the animal pole with more yolk-free cytoplasm and nucleus are better accessible for the injected CRISPR/Cas9 mix. Therefore, we tried to target the animal pole, usually conveniently situated on the top of the embryo. However, we suspect that because acipenseriforms undergo holoblastic cleavage (Dettlaff et al., 1993; Ostaszewska and Dabrowski, 2009) injecting any part of the zygote’s cytoplasm would likely work well. Similar to frog, but unlike lamprey, sturgeon eggs have relatively thick envelope. While this does not prevent injection of the embryo with a capillary needle, we preferred to remove the thick outer layer with forceps to better see the injected droplet size and target the animal pole (frog eggs jelly membranes are removed by cysteine treatment; Sive et al., 2000). Due to the similarities between the zygotes of sterlet and those of lamprey with frog, we decided to use approximately the same concentration of injected RNA and protein, and approximately the same relative volume of injection solution. In practical terms, this means we injected a droplet of CRISPR/Cas9 injection mixture with a diameter approximately 1/7 that of the zygote diameter. Because this strategy works well in sterlet, lamprey, and African clawed frog, we suspect a 1/7 size droplet would yield good results in any vertebrate with isolecithal, mesolecithal, or polylecithal eggs displaying holoblastic cleavage. In contrast, zygotes of embryos with meroblastic cleavage (e.g., Takeuchi et al., 2009), such as zebrafish and other teleost fish, appear to tolerate lower amounts of RNA/protein (Hwang et al., 2013; Jao et al., 2013), and lower volumes of injection mix. For such embryos, the injected volume should likely be adjusted based on the size of the blastodisc rather than the entire embryo.

Recent publication of high-quality sterlet genome (Du et al., 2020) should allow the effective design of highly specific sgRNAs. We initially only searched our embryonic transcriptomes for potential off-target sequences. However, additional searches of the new sterlet genome, including non-coding sequences, did not identify any other off-target sequences. When designing sgRNAs, we found that approximately one third of the initially selected guide sequences conformed with off-target sequences according to the criteria described in Methods. When performing CRISPR/Cas9 mutagenesis in sea lamprey and African clawed frog we found that using at least two different sgRNAs per gene, and injecting them separately, is the best strategy for producing verifiable and consistent phenotypes and control for any defects caused by off-target lesions. We observed highly consistent phenotypes and no indication of off-target effects in sea lamprey after mutating more than 20 different genes (Square et al., 2020). The results presented here suggest a similar level of specificity and consistently can be expected in sterlet.

We only observed a moderate disparity in the efficiency of individual sgRNAs. While the mutation rates varied between tyr and shh loci, they were very similar between the guides targeting the same gene (Table 4), with subtle differences recorded among mutant categories. This contrasts with sea lamprey and other species, in which some degree of variation was observed among different guides targeting the same genes (Hsu et al., 2013; Square et al., 2015). We expect that injecting more eggs and guides would result in higher variation in sgRNA efficiency in sterlet as well. On the other hand, as we observed in sea lamprey, mutations causing frameshifts in coding DNA sequences correlated with more severe phenotypes in sterlet as well.

To confirm successful mutagenesis, we also genotyped representative individuals from each phenotypic class. While PCR fragment length is often used to tell whether an individual harbors mutant alleles, we find that this method often produces ambiguous results. We thus chose to clone and sequence genomic fragments including the target sequence and approximately ∼200 bp of flanking sequence on each side. Sequencing of phenotypic mutants and controls, confirmed highly efficient mutagenesis with all tested sgRNAs, with the frequency and type of alleles correlating with phenotype. While even two to three of sequences per individual are sufficient to confirm a baseline level of mutagenesis, we find sequencing ten or more clones gives a view of allelic composition that usually correlates with phenotypes, and also suggests the efficiency of the sgRNA. We also found that, as with lamprey embryos (York et al., 2017), fixed in situ hybridized sterlet embryos can be successfully genotyped if they are not post-fixed with formaldehyde for extended periods of time (see also Square et al., 2020).

The efficiency, precision, and ease of use of CRISPR/Cas9 mutagenesis in sterlet, lamprey, and African clawed frog strongly suggest this method can be successfully applied to other non-teleost fish. Besides sterlet, several other sturgeon species are currently farmed. While sterlet is probably the most suitable model for developmental studies due to its smaller size and monoploid genome, several other species with different levels of polyploidy (Havelka et al., 2016) could be attractive for studies of genome evolution. While polyploidy necessarily complicates targeted gene mutagenesis by requiring simultaneous disruption of homeologs to yield a valid loss-of-function, the efficiency of CRISPR/Cas9 can be exploited to overcome this issue. We and others have found that targeting conserved homeolog sequences, and/or using multiple sgRNAs can yield strong phenotypes in the allotetraploid African clawed frog (Wang et al., 2015; Square et al., 2020). Gene manipulation in larger species of sturgeons can potentially be useful in sturgeon food production industry when targeting genes involved in growth, immunity, or egg production.

American paddlefish (Polyodon spathula) is morphologically unusual acipenseriform relative of sturgeons. Like sturgeon, it is also frequently farmed, and its embryos have been used in several evolutionary developmental studies (e.g. Davis et al., 2007; Modrell et al., 2011). There are currently no published reports of CRISPR/Cas9 mutagenesis in paddlefish. However, high similarity of paddlefish and sterlet embryos, and successful previous experiments involving dye injections at later embryonic stages suggest that CRISPR/Cas9 mutagenesis could be highly effective in paddlefish. Unfortunately, collecting of early paddlefish embryos appears to be more challenging than sterlet in part because of the shorter paddlefish spawning season (Modrell et al., 2017).

Besides acipenseriforms, three non-teleost fish lineages have extant members that have been utilized in comparative embryological studies and have published genomes - bichirs (Polypteriformes), gars (Lepisosteiformes), and bowfins (Amiiformes) (Askary et al., 2016; Braasch et al., 2016; Minarik et al., 2017; Stundl et al., 2019; Funk et al., 2020; Stundl et al., 2020; Thompson et al., 2021; Bi et al., 2021). While bichirs and bowfins are evolutionarily very attractive and informative species with unique morphologies, the main challenge in targeted mutagenesis in these organisms is obtaining zygotes for injection. Bichirs reproduce in aquaculture settings and can occasionally spawn in home aquaria. However, their mating is secretive and usually occurs at night. Thus, collection of zygotes requires constant observation, which could disturb the animals and prevent spawning. Once collected, though, bichir embryos are relatively sturdy, and can be manipulated and injected with vital dyes much like African clawed frog (Stundl et al., 2019). Bowfin eggs are typically only collected in the wild, and their zygotes are thus less accessible than those of bichir (Funk et al., 2020). Fortunately, decapsulating of later bowfin embryonic stages is relatively easy, so it is likely that early embryos could be injected if obtained (Brent Hawkins, personal communication). Cleavage of bowfin embryos is intermediate between the holoblastic mode of most non-teleost fishes and the meroblastic cleavage of teleosts, with reduced cleavage on the vegetal hemisphere (Cooper and Virta, 2007). This would suggest that if bowfin zygotes were collected, a smaller injection volume might be needed for embryos to survive. Besides sturgeons, perhaps the most promising non-teleost fish candidates for CRISPR/Cas9 mutagenesis are gars. Several species are kept and propagated at fish farms, and like in sturgeons, large quantities of eggs can be seasonally obtained in a controlled fashion in vitro fertilization (Braasch et al., 2014 in spotted gar; Minarik et al., 2017 in tropical gar). Subsequent manipulation with zygotes and raising the offspring is not complicated and the similarity of gar and sturgeon eggs should make injecting relatively straightforward. One potentially confounding factor worth mentioning is that gar embryos, like those of lamprey, lack maternal pigmentation, making it difficult to discern when cleavage begins (Comabella et al., 2014).

While other non-teleost ray-finned fish species are likely well suited to CRISPR/Cas9 F0 mutagenesis, two key groups of early-diverging vertebrates, the chondrichthyans, and hagfish, are unlikely to be amenable to this method, at least as described here. Representatives of both groups have been used in developmental studies in the last 2 decades and have provided some fundamental insights into the developmental bases of vertebrate evolution (Ota et al., 2008; Gillis et al., 2009; Oisi et al., 2013; Gillis and Tidswell, 2017). Embryos of both chondrichthyan clades; elasmobranchs (sharks, rays, and skates) and holocephalans (chimaeras) can be collected seasonally in the wild or in public aquaria, sometimes in the dozens (Gillis, 2011). However, zygotes that could potentially be injected are extremely difficult to obtain and, so far, have not been collected in the numbers needed for reproducible gene disruption. The logistics of hagfish embryology are even more challenging. The few classical reports of hagfish embryology have been based on the scarce embryonic material collected in late 19th century. Only recently have substantial numbers of live hagfish embryos been collected and analyzed (Ota and Kuratani, 2006; Kuratani and Ota, 2008; Ota and Kuratani, 2008). To our knowledge, none of these embryos have ever been successfully injected as zygotes.