The development of CRISPR for a mollusc establishes the formin Lsdia1 as the long-sought gene for snail dextral/sinistral coiling

Left-right body asymmetry is established by tightly regulated genetic processes. Interest has centred on the diverse mechanisms proposed for different phyla, the possibility of a unified mechanism (Blum et al., 2014; Coutelis et al., 2014; Juan et al., 2018; Nakamura and Hamada, 2012; Tingler et al., 2018; Vandenberg and Levin, 2013) and the involvement of cellular chirality (Danilchik et al., 2006; González-Morales et al., 2015; Meshcheryakov and Beloussov, 1975; Schonegg et al., 2014; Taniguchi et al., 2011; Tee et al., 2016; Wan et al., 2011); however, many important aspects are unresolved. To study the molecular and cellular mechanisms of chiromorphogenesis, we have used the snail Lymnaea (L.) stagnalis as an organism with unique advantages. Notably, it has both the sinistral (recessive) and the dextral (dominant) strains within a species; it is a hermaphrodite that both self- and cross-fertilizes; and the chirality of the related L. peregra (Boycott and Diver, 1923; Freeman and Lundelius, 1982; Sturtevant, 1923) and of L. stagnalis (Asami et al., 2008; Hosoiri et al., 2003; Kuroda et al., 2009; Shibazaki et al., 2004) species has been suggested but not proven to be determined by a single locus that functions maternally. We have recently identified the actin-related diaphanous gene Lsdia1 as a candidate for the handedness-determining gene by positional cloning (Kuroda, 2014; Kuroda et al., 2016). Lsdia1 (3261 bp and 1087 amino acids) is expressed maternally and the protein is present from ovipositioning to the gastrulation stage in the dextral strains. By contrast, in the sinistral strains, both of the Lsdia1 alleles carry a frameshift mutation early in the coding region (c.184delC) that leads to protein truncation (p.Leu62Serfs*24) (Kuroda et al., 2016) (Fig. 1A). The loss of LsDia1 in the sinistral strains is compensated for by the presence of the tandemly duplicated Lsdia2 gene, the protein product of which has 89.4% amino acid similarity to LsDia1 (Kuroda et al., 2016). Lsdia2 (3297 bp, 1099 amino acids) is also expressed maternally at the one-cell stage in both dextral and sinistral snails (Kuroda et al., 2016).

Another group (Davison et al., 2016) has also reported that Dia is associated with left-right asymmetry; however, their key results do not agree with ours. Notably, the whole-mount in situ hybridization results are opposite for the two studies. We observed no localization of Lsdia1 mRNA (Kuroda et al., 2016), in contrast to Davison's strong asymmetrically localized expression of dia as well as fry mRNAs confined to one of the blastomeres or to part of a blastomere in the two- and four-cell-stage embryos. Second, we could selectively target Lsdia1 mRNA based on our sequencing of Lsdia1 and Lsdia2 genes; however, they could not discriminate the two mRNAs in their whole-mount in situ hybridization experiments because of DNA sequence differences from ours in some parts. Finally, regarding the formin inhibition experiments used for corroboration, owing to lethality even at 5.0 µM inhibitor concentration, we concluded that these experiments do not provide any meaningful insights (Kuroda et al., 2016). Davison et al., on the other hand, drew a conclusion based on their results with 100 µM inhibitor concentration. Regardless of these key differences, the 2016 papers left open the likely possibility that the key mutation was genetically linked to the mutation we found, rather than being the mutation we found itself. Thus, although the 2016 papers attempted to identify the key gene, decisive identification has awaited the construction and testing of independent mutants with the same effects. Accordingly, we aimed to identify the gene by directly knocking out Lsdia1 using CRISPR/Cas9.

To establish that Lsdia1 is the handedness-determining gene and to elucidate what role Lsdia1 has in the cellular programme that directs snail chirality, we developed Lsdia1-targeted CRISPR/Cas9 techniques (Jinek et al., 2012) to create gene-knockout snails (Fig. 1A,B). Out of four independent injection experiments, we obtained 10 CRISPR/Cas9-injected adult snails (numbered 1-10) and two control adult snails (11-12) (Fig. S1B,C). By using DNA extracted from the clipped foot while keeping the snails alive, heteroduplex mobility assays (HMA) (Zhu et al., 2014) were carried out on the snails (Fig. 1B), which showed mosaic mutational patterns at the F0 stage (Fig. 1C). Germline transmission to F1 embryos occurred in snails 2, 3, 5, 8 and 10, of which snails 2, 3 and 10 produced only dextral offspring, whereas snail 8 had only sinistral offspring (Fig. 1D; Fig. S2; Table S1). Remarkably, snail 5 produced both dextral and sinistral F1 (denoted as 5-F1-dex and 5-F1-sin) within an egg mass (Fig. S2B; Table S1). This must be a consequence of mosaicity of frameshift and non-frameshift mutations introduced into the tissues of the gonad at F0, which are originated from the 4d mesentoblast (Morrill, 1998). When and where in the gonad the Lsdia1 mRNA is transcribed and provided to the oocyte is an interesting unresolved issue.

To prove the shell coiling direction is really the maternal effect of the edited Lsdia1 gene, we studied the F2 generation where the genotype is homogeneous throughout the body (Fig. 1E). PCR genotyping of F1 adult snails showed only a limited number of mutation patterns in each F0 (Fig. 1F; Table S2). By analysis of many thousands of offspring from each F1 lineage, it is clear that F1 snails with a homozygous knockout mutation (e.g. del5/del5, del13/del17) produced sinistral snails. By contrast, snails with biallelic non-frameshift mutations (e.g. del57/del57) or with heterozygous knockouts where at least one of the alleles possesses no mutation (wt) or a non-frameshift mutation (e.g. wt/del5, del3/del10) produced only the dextral F2 generation (Fig. 1G). This outcome was also observed for 5-F1-dex and 5-F1-sin, regardless of the body handedness of the mother (Fig. 1G; details in Table S2). Clearly, the genetic chirality control introduced by the CRISPR/Cas9 at F0 is stably inherited and continues generation after generation (verified up to F4) (F3 in Fig. 1H; Fig. S3; Tables S1, S2). This is in sharp contrast to the mechanically induced mirror-image snails whose effect lasts one generation only (Kuroda et al., 2009). Sinistral-specific species such as Physa acuta and Indoplanorbis exustus carry only the dia type 2 gene (Kuroda et al., 2016). Thus, our gene-editing results are interesting as they may eventually lead to the generation of a new species (Hoso et al., 2010). We have confirmed that there was no gene editing in Lsdia2. Moreover, off-target effects are unlikely as clear and consistent results were obtained from independent CRISPR/Cas9 experiments.

Cytoskeletal dynamics at the crucial 3rd cleavage (four- to eight-cell stage) is also controlled by the Lsdia1 genotype. This stage displays the first definitive chirality, i.e. clockwise (CW) and anti-clockwise (ACW) micromere rotation for the dextral and the sinistral embryos, respectively, and the CW micromere rotation of dextral embryos is accompanied by the spiral deformation of macromeres (SD) and the consequent spindle inclination (SI). In the sinistral embryos, spindles are radially positioned (no SI) and micromeres are formed just above the macromeres without SD, which are then rotated levotropically (Shibazaki et al., 2004; Kuroda, 2014, 2015; Kuroda et al., 2009, 2016). We have checked the 3rd-cleavage pattern of progeny of F1 and classified it in the same way as we previously classified the wild-type embryos (Kuroda, 2014). Progeny of F1 with biallelic non-frameshift mutants and heterozygous knockouts exhibited CW rotation with SD and SI, just like the wild-type dextral strains, even with up to a 57-base pair deletion in both alleles (Fig. 2A-C; Fig. S4). By contrast, offspring of F1 snails with homozygous knockouts exhibited varied cleavage patterns similar to the wild-type (−/−) sinistral embryos, i.e. mostly ACW, but sometimes partial ACW, neutral and very rarely partial CW and CW, with no SD/SI in all these cases (Fig. 2A-C; Fig. S4; Movies 1-3). SD ensures correct direction and sufficient degree of micromere rotation at the 3rd cleavage. Hence, homozygous knockout or wild-type (−/−) sinistral embryos without this control sometimes result in micromere rotations other than ACW, leading to abnormal development (Kuroda, 2014) (Fig. 1G; Tables S1, S2). The marginally higher incidence of the unusual development in the offspring of homozygous knockout mothers may be due to the fact that CRISPR/Cas9-mediated homozygous knockout allows the translation of the protein up to the FH1 domain (Fig. 1A) and thus, the partial proteins may interfere with the activities of LsDia2 protein. The interpretation that the very rare dextrality that emerged is not associated with genetic changes is supported by the finding that sinistral F3 offspring were born from the dextral snails according to the F2 genotype (Fig. S5; Table S2).

Evidence for Lsdia1 as the handedness-determining gene is further strengthened by western blot analyses carried out for each F1 genotype. A perfect correlation was found between the presence and absence of the protein in the one-cell embryos immediately after ovipositioning and the maternal genotype (Fig. 2D). It appears that LsDia1 protein present in oviposited embryos induces dextral spiral cleavage with SD and SI. Involvement of the actin cytoskeleton during the 3rd cleavage is in agreement with our previous finding that agents inhibiting actin but not microtubule polymerization destroyed SD (Shibazaki et al., 2004).

Body handedness of L. stagnalis is uniquely established at the 3rd cleavage (Kuroda et al., 2009). Is this the first observable chirality? When is chirality first exhibited in development? To address this important issue, we used trypsin treatment of the one-cell-stage embryos to decrease the vitelline membrane tension, a method first described by Meshcheryakov and Beloussov (1975). We revealed, for the first time, substantial mirror-image chirality for the sinistral and dextral embryos within a species at the 1st cleavage. Remarkably, the chirality depended on the maternal Lsdia1 genotype regardless of whether it is for the wild-type or the gene-edited embryos (Fig. 3A,B; Movie 4). The observations indicate that the single fertilized egg carries intrinsic left or right chirality information depending on the absence or presence of functional LsDia1 protein. This is the earliest observed symmetry-breaking event linked directly to body handedness in the phyla of the animal kingdom (Blum et al., 2014; Coutelis et al., 2014; Schonegg et al., 2014; Vandenberg and Levin, 2013).

The mirror-relationship of cytoskeletal dynamics between the dextral and sinistral embryos within a species is kept until the end of the 2nd cleavage but broken at the crucial 3rd cleavage where only the dominant dextral embryos exhibit SD/SI (Shibazaki et al., 2004; Kuroda et al., 2009). It seems that chirality is realized initially using the information contained within each blastomere. This may solve the enigma whereby micromeres of sinistral embryos rotate only ACW at the 3rd cleavage after the non-chiral emergence from the macromeres, rather than equally both ACW and CW. During and after the 3rd cleavage, cell-cell interactions become dominant, which can overwrite the initial intracellular chirality if appropriately forced by artificial external power, and establish definitive chirality.

Chiromorphogenesis at the organismal level is achieved by asymmetric expression of nodal and Pitx genes in both invertebrates and vertebrates (Abe et al., 2014; Blum et al., 2014; Coutelis et al., 2014; Grande and Patel, 2009; Kuroda, 2014; Kuroda et al., 2009; Vandenberg and Levin, 2013). We have previously shown that nodal and Pitx gene expression in L. stagnalis starts at the early cell stages of 33-49 and 49-64, respectively, and continues up to veliger stage (Kuroda, 2015), following the chirality of the eight-cell blastomere arrangement even when that is reversed by mechanical manipulation (Kuroda, 2014). We could show (Fig. 4) that homozygous knockout embryos produced by the gene-edited F1 snails exhibited nodal and Pitx gene expression patterns that were complete mirror images of the original dextral embryos, whereas expression in heterozygous knockout snails was just like the wild-type dextral embryos (Kuroda, 2015; Kuroda et al., 2009). Thus, nodal and Pitx expression patterns are also determined by the presence of LsDia1 protein.

How is the chirality that is established at the 3rd cleavage transmitted to produce asymmetric nodal and Pitx expression? We identified the cell that first expresses the nodal gene as cell 2c¹² at the 49-cell stage and that of Pitx as 2d²¹² at the 64-cell stage. Thus, the 24-cell stage at which the cell fates are determined (Martindale, 1986) must be a crucial period in the transfer of chiral information.

In this study, using the first successful gene knockouts in Mollusca, we revealed novel mechanistic features of left-right asymmetry establishment in L. stagnalis and its dependence on the actin-related LsDia1 protein. First, we present decisive genetic evidence that expression of Lsdia1 determines the outcome of left-right body asymmetry. Second, LsDia1 controls the intercellular chirality at the 3rd cleavage, which leads to the asymmetric nodal and Pitx expression at a specific cell at an early developmental stage. Third, we observe for the first time intracellular chirality at the fertilized one-cell stage dependent on LsDia1. The importance of actin-mediated cytoskeletal dynamics is increasingly recognized in the chiromorphogenesis of many animal phyla. Given that diaphanous orthologues are also present in human cells, studies on LsDia1 and LsDia2 may have implications for understanding situs inversus, a characteristic of 0.01% of the human population (Blum et al., 2014; Coutelis et al., 2014). Therefore, our work featuring formin-controlled early onset of chirality in L. stagnalis may provide new insight on unifying mechanisms in eukaryotes.

Both the sinistral (−/−) and the dextral (+/+) wild-type strains of L. stagnalis used in this study have been maintained for many years from the original strain kindly supplied by G. Smit (Vrijie Universiteit, Amsterdam, The Netherlands). Genome-edited snails were created from this dextral strain. They were maintained under a 16/8 h light/dark cycle at 20-22°C and given fresh lettuce leaves and artificial food for tropical fish. All snail genome-editing experiments in this study were approved by the Tokyo University of Science (approval no. 1750/1751, 1873/1874), Japan. The care and use of animals were in accordance with the guidelines for animal experiments of Tokyo University of Science.

To create the Lsdia1-targeting gRNA, a template plasmid vector was constructed by cloning the synthesized oligonucleotides for the target sequence (Fig. S1A) into BsmBI-digested pT7-gRNA vector (Addgene plasmid #46759) (Jao et al., 2013). The gRNA was generated by in vitro transcription from the BamHI-digested template plasmid using the MEGAscript T7 Kit (Ambion) and purified using the mirVana miRNA isolation kit (Ambion). The Cas9 nuclease mRNA was transcribed from NotI-digested pCS2-nCas9n vector (Addgene Plasmid #47929) (Jao et al., 2013) using the mMESSAGE mMACHINE SP6 kit (Ambion) and purified by phenol extraction and ethanol precipitation. The sequence comparison of Lsdia1 and Lsdia2 was performed using the ClustalX 1.83 program (Thompson et al., 1997).

A solution containing 100 ng/μl gRNA and 1500 ng/μl Cas9 mRNA diluted in nuclease-free water was injected into the dextral strain eggs at the one-cell stage according to our previously described protocol (Abe et al., 2009). Injected embryos and non-injected control embryos were cultured overnight in 5×HF solution (30 mM NaCl, 0.335 mM KCl, 0.34 mM CaCl₂, 0.1 mM MgSO₄, 0.12 mM NaHCO₃) based on Holtfreter's solution (Holtfreter, 1931). Some of these embryos were transferred to ex ovo culture in capillaries, as previously described (Kuroda et al., 2009), and were raised to adult snails (F0) individually. They produced the F1 generation by self-fertilization. The rest of the embryos were used for checking the genome-editing efficiency in the respective experiments by HMA (Fig. S1C). It has been reported that CRISPR/Cas9 can function in the early embryos of molluscs (Perry and Henry, 2015).

The heteroduplex mobility assay (HMA) was used to analyse genome editing efficiency, and screening for germline transmission (Zhu et al., 2014; Ota et al., 2013, 2014). A 226 bp fragment including the target site was amplified from the Lsdia1 sequence by using primers D1-MA4 detection Fw and D1-MA4 detection Rv (Fig. S1A). Each genomic DNA of adult snails was extracted from foot clips employing the Maxwell 16 Tissue DNA Purification Kit (Promega) and used as a PCR template. In the efficiency check for each injection experiment or screening for germline transmission, a small embryo was added directly to the PCR solution, and direct PCR amplification was performed by using KOD Fx Neo (Toyobo). The PCR conditions for HMA were as follows: 96°C for 10 min; 35 cycles of 98°C for 10 s and 66°C for 30 s; followed by denaturation at 96°C for 3 min; and slow cooling to 20°C over 30 min. For the screening of germline transmission, the PCR products were run on 2% agarose gels and also sometimes on 10% polyacrylamide gels.

All egg masses laid by genome-edited F0 snails and at least two egg masses by each genome-edited F1 and F2 snail were cultivated individually in a plastic petri dish at room temperature (20-22°C) until hatching. The control strains, dextral (+/+) and sinistral (−/−) were also studied similarly. The number of offspring in each egg mass was counted for the left- or the right-shell coiling. Although malformations were observed frequently in either shells or soft body, or both, they were not distinguished in this study. Those and juveniles with a swollen hydropic body were classified as abnormal embryos. The detailed analysis is summarized in the Figs S3A, S5A and Tables S1, S2. Photographs of the juveniles were taken using an Olympus SZX12 stereomicroscope equipped with an Olympus DP-70 CCD camera.

F1 snails were separated individually before sexual maturation, and genomic DNA was extracted from the foot clips. HMA was then performed on each snail as above. The PCR products were treated with Exo-SAP-IT (USB) and reacted directly with BigDye terminator v3.1 (Applied Biosystems) using D1-MA4 seq Rv primer (Fig. S1A). Genome-editing patterns for these products were analysed by using an ABI3130xl Genetic Analyzer (Applied Biosystems). The same genotyping analysis was performed on the following generations.

A fresh egg mass was collected from the individual genome-edited snail, dextral (+/+) or sinistral (−/−), and the one-cell-stage embryos were isolated from the capsule in 5×HF solution. Healthy looking four-cell embryos, near the third division time, were transferred to 5×HF solution in a small enclosure made of plastic tape on a glass slide and sealed with a cover slip. Time-lapse images of the third cleaving embryos were acquired on a Zeiss Axioskop 2 mot microscope equipped with a Zeiss 10× Plan-Neofluar objective lens and a Zeiss Axiocam MRc CCD camera, controlled by Zeiss Axiovision software. The data were obtained every 20 s or 30 s. The third cleavage pattern of each embryo was analysed visually, and the analysed data are summarized in Fig. 2A.

LsDia1 and β-tubulin protein detection by western blotting was performed as previously described (Kuroda et al., 2016). Fifty one-cell-stage embryos were obtained from one snail for each genotype and used as an SDS-PAGE sample for one lane. The following antibodies were used to probe the respective proteins: anti-LsDia1 (Kuroda et al., 2016; 1/10,000) and horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (GE Healthcare, NA934; 1/50,000), anti-β-Tubulin (Sigma, TUB2.1; 1/20,000) and HRP-conjugated anti-mouse IgG (GE Healthcare, NA931; 1/50,000). The chemiluminescence signals were imaged using a ChemiDoc MP Imaging System (Bio-Rad) or a Luminescent Image Analyzer LAS-1000plus (FUJIFILM), and the images were analysed using Image Lab Software (Bio-Rad) or Image Gauge (FUJIFILM).

Whole-mount in situ hybridization for the nodal and Pitx genes was performed according to our standard protocols described previously (Kuroda et al., 2009, 2016). Digoxigenin-labelled antisense RNA probes were generated for their full-length coding sequences. Timing of fixing of embryos at the 49- or 64-cell stage was judged by observing the DAPI staining of embryos, whereas that for late trochophore-stage embryos was judged from their appearance. Stained embryo images were acquired using a Zeiss Axio Imager M1 microscope equipped with a Zeiss Axiocam MRc CCD camera.

Four-cell or four- to eight-cell-stage embryos in the genome-edited snails, dextral (+/+) or sinistral (−/−) snails, were fixed in 4% paraformaldehyde in MTSTr [MTS buffer (pH 6.9), 0.1% Triton X-100] at 4°C overnight. Cytoskeleton staining was performed as described previously (Shibazaki et al., 2004; Kuroda et al., 2009, 2016). Confocal images were obtained using a Zeiss LSM 5 pascal laser scanning microscope equipped with a 20× Plan-Neofluar objective lens. The z-stack images were made from optical slices acquired every 10 μm along the animal-vegetal axis by using Zeiss LSM5 pascal software.

Trypsin treatment for removing the vitelline membrane of the fertilized eggs was performed as described by Meshcheryakov and Beloussov (1975). Crystalline trypsin (Wako) was dissolved in 5×HF before use to prepare a fresh solution at a concentration of 10 mg/ml. Embryos used were obtained from the population of dextral (+/+) or sinistral (−/−) and the individual genome-edited snails, 3-F4 (wt/del5) or (del5/del5), which were derived from one self-fertilized heterozygous snail, 3-F1-5-F2-1-F3-1 (wt/del5). The one-cell-stage embryos were isolated from the capsule and washed to remove adherent albumin in 5×HF. The embryos were transferred to the trypsin solution and incubated for 30 min at room temperature (21-23°C). Then the trypsinized embryos were carefully washed four times with fresh 5×HF and cultured until the first cleavage. Although the trypsinized embryos became fragile, the first cleavage initiated normally. Immediately after completion of the 1st cleavage, the trypsinized embryos were fixed in 4% paraformaldehyde in MTSTw [MTS buffer (pH 6.9), 0.1% Tween-20] for 30-120 min at room temperature and the chirality of each embryo was carefully observed. Depending on the degree and the rotational direction of the front blastomeres with respect to the rear ones when looking down from a direction perpendicular to the cleavage plane, they were classified as CW, weak CW, neutral, weak ACW and ACW.

Time-lapse images of the 1st cleavage of the trypsinized embryos were acquired using an Olympus SZX12 stereomicroscope equipped with an Olympus DP-70 CCD camera, controlled by Olympus DP Controller software. The data were obtained every 20 s.

Statistical calculations were performed using Pearson's χ² test comparing the homozygous mutants against the wild-type dextral strain for number of sinistral juveniles, 3rd cleavage pattern and one-cell-stage chirality.

We thank Mr. S. Watanabe and Mr. K. Takanashi, and Mr. F. Ichikawa and Mr. H. Takahashi (graduate school students of Tokyo University of Science) for their help in collecting embryos and for their assistance in rearing gene-edited snails, respectively.