Generation of knock-in lampreys by CRISPR-Cas9-mediated genome engineering

Strategy for generating knock-in lampreys with the LcHsp70A promoter

For CRISPR-Cas9-mediated knock-in via non-homologous end joining (NHEJ) in the lamprey, we used an experimental strategy previously established in zebrafish¹³ and medaka¹⁴ with minor modification: co-injection of sgRNA1 (for genome digestion), sgRNA2 (for plasmid digestion), donor plasmids, Cas9 mRNA, and fast green (for visualization of the injection cocktail) (Fig. 1A).

CRISPR-Cas9-mediated knock-in strategy using LcHsp70A promoter. (A) For the generation of knock-in lamprey, sgRNA1 (for genome digestion), sgRNA2 (for plasmid digestion), the donor plasmids having a bait sequence, and Cas9 mRNA in distilled water containing Fast Green to aid visualization of the spread of the injection are co-injected into lamprey zygotes. (B) After injection, the CRISPR-Cas9-mediated concurrent cleavage occurs in the genome at the site upstream (approximately, 200–600 bp) of the target gene and in the donor plasmid at the bait sequence. This leads to a homology independent DNA repair, resulting in the integration of the donor plasmid into the targeted locus. Here, both forward and reverse integrations can occur. Cis-regulatory DNA sequences for the target gene expression act on the LcHsp70A promoter (enhancer-trapping), resulting in the expression of the reporter gene in cells that express the target gene.

Species-specific heat shock protein 70 (Hsp70) promoters work effectively as a minimal promoter in zebrafish and medaka^13,14. Thus, we first aimed to isolate a lamprey Hsp70-like promoter. We found two homologs of Hsp70-like genes in the P. marinus and L. camtschaticum genomes, one of which we named LcHsp70A. Phylogenetic analysis suggests that those two homologs are the result of lineage-specific duplication in the lamprey (Fig. S1A). The LcHsp70A sequence was also found in the transcriptomes of L. camtschaticum embryos obtained previously¹⁵. In this study, we extracted a ~ 0.2 kb sequence of the LcHsp70A promoter for plasmid construction (Fig. S1B).

The donor plasmid contains a bait sequence (Tbait, a 23 bp sequence derived from the mouse Tet1 gene¹³, upstream of the insertion cassette for sgRNA2-guided DNA cleavage (Fig. 1B). This bait sequence was selected because the corresponding sgRNA (sgT) appears to have no potential off-target binding sites except for those with three or more mismatches in the L. camtschaticum genome (Table S1). Tbait is followed by the LcHsp70A promoter, which we expect to work as a minimal promoter (Fig. 1B). Finally, the LcHsp70A promoter is followed by reporter genes (in this study, EGFP or Dendra2).

We set the target site for genome digestion approximately 200–600 bp upstream of the predicted transcriptional-start site of the target genes (Fig. 1B; the sequences used are shown in Table S2). Concurrent digestion of the genome and plasmid DNA (guided by sgRNA1 and sgRNA2, respectively) with Cas9 would result in the integration of the donor plasmid into the genome via NHEJ (Fig. 1B). Then, cis-regulatory sequences for tissue-specific expression of the target gene would act on the LcHsp70A promoter, resulting in expression of a transgene in cells that express the target gene (enhancer-trapping).

Generation of Bra:EGFP and MA2:EGFP knock-in lampreys

To examine the effect of the CRISPR-Cas9-mediated knock-in system, we first targeted brachyury (Bra) and muscle actin 2 (MA2) genes for an EGFP reporter assay (hereafter, we call the knock-in animals Bra:EGFP and MA2:EGFP knock-in lampreys, respectively). In the lamprey, a T-box family transcription factor Bra is expressed in the dorsal protostome (axial mesoderm progenitor), the notochord, and the tailbud^16,17, while MA2 is a muscle marker^18,19,20. We tested two sgRNAs for Bra and one for MA2 (Table S2). Each sgRNA was co-injected with sgRNA for Tbait (sgT), the donor plasmids, and Cas9 mRNA into zygotes. We incubated the injected embryos and investigated their EGFP expressions during development (Figs. 2 and 3). In this investigation procedure, we screened and counted embryos or prolarvae with specific EGFP expression at stage 16 for Bra:EGFP and stage 26 for MA2:EGFP knock-in lampreys (Table S3).

Bra:EGFP knock-in lampreys generated by microinjection of Bra-sg1. (A) At stage 16, in dorsal view (A) and posterior view (A’). EGFP is expressed in the axial mesodermal cells (am). A, D, P, V indicate anterior, dorsal, posterior, and ventral, respectively. Scale bar: 500 μm. (B) At stage 21, in lateral view (B) and dorsal view (B’). EGFP is persistently expressed in the axial mesodermal cells (am). L, P, R, V indicate left, posterior, right, and ventral, respectively. Scale bar: 500 μm. (C) At stage 25, in lateral view. The EGFP expression was restricted mostly in the tailbud and anal regions. In the york, some non-specific signals (by unintegrated plasmids) were observed (arrows). A, D, P, V indicate anterior, dorsal, posterior, and ventral, respectively. Scale bar: 500 μm.

MA2:EGFP knock-in lampreys. (A) A MA2:EGFP knock-in lamprey showing EGFP expression in the head region at stage 30, in lateral view. The head region is magnified in (A’). The asterisk (*) indicates the eyeball. EGFP is expressed both somatic and pharyngeal muscles (m.). In the york, some non-specific signals (by unintegrated plasmids) were observed (arrows). A, D, P, V indicate anterior, dorsal, posterior, and ventral, respectively. Scale bar: 500 μm. (B) A MA2:EGFP knock-in lamprey showing EGFP expression in the trunk region at stage 30, in lateral view. A part of the trunk region is magnified in (B’). Both EGFP-positive and EGFP-negative somatic muscle cells are observed in the same somite. In the york, some non-specific signals (by unintegrated plasmids) were observed (arrows). A, D, P, V indicate anterior, dorsal, posterior, and ventral, respectively. Scale bar: 500 μm.

With Bra-sg1, we observed axial mesoderm progenitor-specific EGFP expression (see Fig. 2A as an example) in 34% (37 of 108) of the surviving animals (Table S3). The remaining 66% showed either sparse, widespread non-specific EGFP expression or no expression. The efficiency was not improved by injecting Cas9 protein instead of mRNA (Table S4). To determine the injection damage and toxicity, we also performed sgRNA1-removed cocktail injection and water injection, comparing the result with the no-injection control. The water injection control showed slightly higher survival rate (28.5%) than that of the sgRNA1-removed cocktail injection control (18.5%), which was within the range of the experimental group (21.0%) (Table S4). As the survival rate of the no injection control was far higher (90.5%) than water injection control, both injection damage and RNA toxicity affect to the survival rate.

At stage 16 (late gastrula), EGFP expression was observed in axial mesoderm-progenitor cells (Fig. 2A) and it was persistently observed in those cells at stage 21 (late neurula) (Fig. 2B). The expression was confirmed histologically (Fig. S2). At stage 25 (late pharyngula), the EGFP expression was restricted mostly to the tailbud and anal regions with some non-specific signals (by unintegrated plasmids) observed in the yolk (Fig. 2C). Likewise, in the case of Bra-sg2, we found specific EGFP expression in 23% (25 of 110) of the surviving animals (Table S3). The expression patterns were identical to those observed in Bra-sg1-injected animals (Fig. S3). These expression patterns were consistent with reports based on in situ hybridization analysis^16,17. However, the specific EGFP expression was observed only on one side of the animal body in all examined embryos/prolarvae, suggesting that the integration of the donor plasmid occurred only in some of the blastomeres, causing mosaicism.

In the case of MA2-sg1, we observed muscle-specific EGFP expression in 21% (25 of 119) of the surviving animals (Table S3). The EGFP signals were persistently detected in muscle cells of stage 30 ammocoetes larvae (~ 40 days after fertilization; Fig. 3). In all prolarvae examined, the EGFP expression was again restricted to some cells only on the left or right side of the animal body, as observed in the Bra:EGFP knock-in lampreys described above. In the animals that expressed EGFP in somatic muscle cells, mosaicism was found even in the same somite (Fig. 3B). Both EGFP-positive and -negative muscle cells were present in these somites, suggesting heterogeneity of mesodermal progenitor cells in somitogenesis. In the animals that expressed EGFP in the head region, the EGFP signals were detected in pharyngeal derivatives, including the upper lip muscle, ventral labial elevator, buccal constrictor, superficial branchial constrictors, interbranchial muscles, and branchial sphincters (Fig. 3A’; nomenclature based on previous description^21,22).

These results strongly suggest that CRISPR-Cas9-mediated knock-in occurred in some of the blastomeres in the injected animals. To confirm that the integration of the donor plasmid indeed occurred in the targeted loci of the genome, we performed insertion mapping of the MA2:EGFP knock-in lampreys (Figs. S4A and S4B). This indicated that the targeted integrations occurred both in the forward (#1 and #2 in Fig. S4B) and reverse (#3 in Fig. S4B) directions. Further evidence of the targeted integration was obtained by the sequencing the transgene-integrated animals (Fig. S4C).

Generation and photoconversion of SoxE3:Dendra2 knock-in lampreys

To explore the potential versatility of our knock-in strategy, we next generated SoxE3-targeted knock-in lampreys using donor plasmids that contained the coding sequence for the photoconvertible fluorescent protein Dendra2²³. In the lamprey, a Sox family transcription factor SoxE3 is expressed in neural crest cells (NCCs) and plays key roles in pharyngeal development^24,25. The migration patterns of NCCs have been investigated from the viewpoint of jaw evolution in cell-labeling experiments using lipophilic tracers, such as DiI^20,26,27. However, these dyes cannot selectively label NCCs, causing inevitable artifact signals. To overcome this problem, we planned to generate SoxE3:Dendra2 knock-in lampreys and to perform cell type-specific lineage tracing by selectively highlighting SoxE3-positive NCCs.

For this purpose, sgRNA SoxE3-sg1 (Table S2) was co-injected with sgRNA for Tbait (sgT), the donor plasmids containing the Dendra2 sequence, and Cas9 mRNA into zygotes. We incubated the injected embryos and investigated them for their native (green) Dendra2 expression at stage 21. We observed NCC-specific expression in 31% (18 of 59) of the surviving animals (Table S3). We then selectively highlighted some of Dendra2-positive cells by photoconversion with region-specific application with ultraviolet light at this stage (Fig. 4A). We further incubated the photoconversion-treated embryos and re-examined them at stage 24. In these animals, some photoconverted cells (originally situated at the neural tube level on the dorsoventral axis) were observed in the pharyngeal region, suggesting that those cells migrated ventrally (Fig. 4B). These results indicate the usefulness of the CRISPR-Cas9-mediated knock-in system in this kind of lineage tracing analysis.

Cell lineage analysis of photoconverted SoxE3:Dendra2 knock-in lampreys. (A) Photoconversion experiments were performed in the head region of SoxE3:Dendra2 knock-in lampreys at stage 21. Native green fluorescence, photoconverted red fluorescence (shown in magenta), and the merged image are shown in (A), (A’), and (A’’), respectively. The midbrain–hindbrain boundary (MHB) is indicated with dashed lines. Scale bar: 500 μm. (B) The same animal shown in (A) is raised to the stage 24 and reexamined. Some photoconverted cells are observed in the mandibular arch (ma), suggesting ventral migration of these NCC cells (arrows). Native green fluorescence, photoconverted red fluorescence (shown in magenta), and the merged image are shown in (B), (B’), and (B’’), respectively. The midbrain–hindbrain boundary (MHB) is indicated with dashed lines. Scale bar: 200 μm.