Endogenous protein tagging in medaka using a simplified CRISPR/Cas9 knock-in approach

  1. Ali Seleit  Is a corresponding author
  2. Alexander Aulehla
  3. Alexandre Paix  Is a corresponding author
  1. Developmental Biology Unit, European Molecular Biology Laboratory, Germany

Abstract

The CRISPR/Cas9 system has been used to generate fluorescently labelled fusion proteins by homology-directed repair in a variety of species. Despite its revolutionary success, there remains an urgent need for increased simplicity and efficiency of genome editing in research organisms. Here, we establish a simplified, highly efficient, and precise strategy for CRISPR/Cas9-mediated endogenous protein tagging in medaka (Oryzias latipes). We use a cloning-free approach that relies on PCR-amplified donor fragments containing the fluorescent reporter sequences flanked by short homology arms (30–40 bp), a synthetic single-guide RNA and Cas9 mRNA. We generate eight novel knock-in lines with high efficiency of F0 targeting and germline transmission. Whole genome sequencing results reveal single-copy integration events only at the targeted loci. We provide an initial characterization of these fusion protein lines, significantly expanding the repertoire of genetic tools available in medaka. In particular, we show that the mScarlet-pcna line has the potential to serve as an organismal-wide label for proliferative zones and an endogenous cell cycle reporter.

Introduction

The advent of gene editing tools (Wang et al., 2016; Jinek et al., 2012; Cong et al., 2013) in conjunction with the expansion of sequenced genomes and engineered fluorescent proteins (Chudakov et al., 2010; Shaner et al., 2013; Bindels et al., 2017; Campbell et al., 2020) has revolutionized the ability to generate endogenous fusion protein knock-in (KI) lines in a growing number of organisms (Paix et al., 2015; Paix et al., 2017a, Gratz et al., 2014; Kanca et al., 2019; Wierson et al., 2020; Gutierrez-Triana et al., 2018; Auer and Del Bene, 2014; Yoshimi et al., 2016; Yao et al., 2017; Cong et al., 2013; Dickinson et al., 2015; Leonetti et al., 2016; Wierson et al., 2019). These molecular markers expressed at physiological levels are central to our understanding of cellular- and tissue-level dynamics during embryonic development (Gibson et al., 2013). To this end, researchers have utilized the Streptococcus pyogenes CRISPR-associated protein 9 (Cas9) and a programmed associated single-guide RNA (sgRNA) to introduce a double strand break (DSB) at a pre-defined genomic location (Jinek et al., 2012). Cell DNA repair mechanisms are triggered by the DSB and it has been shown that providing DNA repair donors with homology arms that match those of the targeted locus can lead to integration of the donor constructs containing fluorescent reporter sequences in the genome by the process of homology-directed repair (HDR) (Danner et al., 2017; Jasin and Haber, 2016; Ceccaldi et al., 2016; Lisby and Rothstein, 2004). Despite its success, HDR mediated precise single-copy KI efficiencies in vertebrate models can still be low and the process of generating KI lines remains cumbersome and time consuming. Recent reports have improved the methodology by the usage of 5′ biotinylated long homology arms that prevent concatemerization of the injected dsDNA (Gutierrez-Triana et al., 2018) or by linking the repair donor to the Cas9 protein (Gu et al., 2018; Savic et al., 2018; Carlson-Stevermer et al., 2017; Aird et al., 2018). In addition, repair donors with shorter homology arms in combination with in vivo linearization of the donor plasmid have been shown to mediate efficient knock-ins in zebrafish and in mammalian cells (Wierson et al., 2020; Hisano et al., 2015; Cristea et al., 2013; Yao et al., 2017).

In this work, we establish a simplified, highly efficient, and precise strategy for CRISPR/Cas9-mediated endogenous protein tagging in medaka (Oryzias latipes). Our approach relies on the use of biotinylated PCR-amplified donor fragments that contain the fluorescent reporter sequences flanked by short homology arms (30–40 bp), by-passing the need for cloning or in vivo linearization. We use this approach to generate and characterize a series of novel KI lines in medaka fish (Supplementary files 3a and 4). By utilizing whole genome sequencing (WGS) with high coverage in conjunction with Sanger sequencing of edited loci, we provide strong evidence for precise single-copy integration events only at the desired loci. In addition to generating an endogenous ubiquitous nuclear label and novel tissue-specific reporters, the KI lines allow us to record cellular processes, such as intra-cellular trafficking and stress granule formation in 4D during embryonic development, significantly expanding the genetic toolkit available in medaka. Finally, we provide proof-of-principle evidence that the endogenous mScarlet-pcna KI we generate serves as a bona fide proliferative cell label and an endogenous cell cycle reporter, with broad application potential in a vertebrate model system.

Results

A simplified, highly efficient strategy for CRISPR/Cas9-mediated fluorescent protein knock-ins in medaka

To simplify the process of generating fluorescent protein knock-ins in medaka we utilized PCR-amplified dsDNA repair donors with short homology arms (30–40 bp). In addition, biotinylated 5′ ends were used to prevent in vivo concatemerization of DNA (Gutierrez-Triana et al., 2018). We used a streptavidin-tagged Cas9 (Cas9-mSA), with the goal of enhancing its binding to the biotinylated repair donor constructs (Gu et al., 2018). This approach by-passes the need for cloning, as the short homology arms are added during PCR amplification. Also, given a linear PCR repair donor is used, there is no need for a second gRNA for in vivo plasmid linearization (Hoshijima et al., 2016; Shin et al., 2014; Zu et al., 2013; Yao et al., 2017; Cristea et al., 2013; Auer et al., 2014; Hisano et al., 2015; Li et al., 2019; Wierson et al., 2020; Kimura et al., 2014). The three-component mix: biotinylated PCR-amplified dsDNA donors, synthetic sgRNA, and Cas9-mSA mRNA (Supplementary file 3b-e) was injected into one-cell-stage medaka embryos (Figure 1 and Figure 1—figure supplement 1), for a detailed protocol see Supplementary files 1 and 2. We targeted a list of eight genes with a variety of fluorescent proteins (Figure 1 and Figure 1—figure supplement 1, Supplementary files 3a-d and 4), both N and C terminus tags were attempted (a list of all genomic loci targeted can be found in Supplementary file 3a). Targeting efficiency in F0 ranged from 11% to 59% of embryos showing mosaic expression (Supplementary files 3a and 4). Control injections with the actb sgRNA, Cas9-mSA mRNA, and the donor eGFP construct without homology arms showed no evidence of eGFP-positive cell clones in F0 (Supplementary file 3a), while the same construct with homology arms resulted in 39% of surviving injected embryos showing mosaic expression of eGFP (Supplementary files 3a and 4). The germline transmission efficiency of fluorescent F0 fish ranged from 25% to 100% for the different targeted loci (Supplementary files 3a and 4). For F0 adults with germline transmission, the percentage of positive F1 embryos ranged between 6.6% (2/30) and 50% (25/50). Using this method, we were able to establish eight stable KI lines. Importantly, a single injection round was sufficient to generate a KI line for most targeted loci (7/8; Supplementary files 3a-f and 4). As previously reported, the actb-eGFP tag was embryonic lethal (Gutierrez-Triana et al., 2018) and we could not obtain a KI line for that locus. We also performed an initial comparison (using fluorescent screening in F0) between different Cas9 designs, that is, with and without mSA. Our results indicate comparable efficiencies of KI insertions in F0s, irrespective of whether a streptavidin tag was included (Supplementary file 3g). Combined, our results provide evidence that highly efficient targeting of endogenous loci with large inserts (~800 bp) is obtained in medaka using the simplified KI approach presented here (Figure 1 and Figure 1—figure supplement 1). In addition to being highly efficient, this protocol is rapid and simple-to-implement, as it relies on a PCR-amplified repair construct and hence alleviates the need for any additional cloning or in vivo plasmid linearization (Figure 1—figure supplement 1).

Figure 1 with 3 supplements see all
Cloning-free single-copy CRISPR/Cas9-mediated knock-in (KI) lines in medaka.

(A) Schematic diagram of cloning-free CRISPR knock-in strategy. The injection mix consists of three components, a single-guide RNA (sgRNA) targeting the gene of interest, Cas9-mSA mRNA, and the PCR-amplified donor fragment containing short homology arms on both ends (30–40 bp) and the fluorescent protein of interest with no ATG and no stop codon. Note that the 5′ ends of the PCR donor fragment are biotinylated (Btn). The mix is injected in one-cell staged medaka embryos and the injected fishes are screened for potential in-frame integrations mediated by homology-directed repair (HDR). (B) eGFP-cbx1b F1 CRISPR KI line stage 40 medaka embryos. eGFP-Cbx1b labels all nuclei and is thus an ubiquitous nuclear marker. n > 10 embryos. Scale bar = 100 µm (C) mScarlet-pcna F1 CRISPR KI line stage 40 medaka embryos. mScarlet-Pcna labels exclusively cycling cells. n > 10 embryos. Scale bar = 100 µm. (D) mNG-myosinhc F1 CRISPR KI line stage 40 medaka embryos. mNG-Myosinhc labels exclusively muscle cells located in the myotome tissue of medaka embryos. n > 10 embryos. Scale bar = 100 µm.

Figure 2 with 4 supplements see all
Tissue- and organelle-specific expression of seven CRISPR/Cas9 knock-in (KI) lines in medaka.

(A) eGFP-cbx1b F1 stage 40 medaka embryo. eGFP-Cbx1b labels all nuclei. Nuclei in the spinal cord of medaka are highlighted (yellow arrowhead). n > 10 embryos. SP = spinal cord. Scale bar = 30 µm. (B) mScarlet-pcna F1 stage 40 medaka embryo. mScarlet-Pcna is localized in the nuclei of cycling cells. mScarlet-Pcna is visible in skin epithelial cell nuclei located in the mid-trunk region of a medaka embryo. The localization of Pcna as speckles within the nucleus indicates cells in S phase of the cell cycle (yellow arrowheads). n = 10 embryos. EC = epithelial cells. Scale bar = 20 µm. (C) eGFP-rab11a F1 stage 40 medaka embryo. Expression of the membrane trafficking marker eGFP-Rab11a is evident in the caudal fin region. eGFP-Rab11a is strongly expressed in the spinal cord (yellow arrowhead) and lateral line neuromasts (magenta arrowhead). n = 6 embryos. CF = caudal fin. Scale bar = 30 µm. (D) mNG-myosinhc F1 stage 40 medaka embryo. mNG-Myosinhc is expressed solely in muscle cells. Myofibrils containing chains of individual sarcomere can be seen (mNG-Myosinhc labels the Myosin A band inside each sarcomere, yellow arrowheads). n > 10 embryos. MT = myotome. Scale bar = 30 µm. (E) cdh2-eGFP F1 stage 40 medaka embryo. Cdh2-eGFP is localized at cell membranes in several tissues, including the spinal cord (magenta arrowhead) and the notochord (yellow arrowhead). n = 5 embryos. NT = notochord. Scale bar = 50 µm. (F–F’) g3bp1-eGFP F1 stage 34–35 medaka embryo. Time-lapse imaging of G3bp1-eGFP dynamics under normal and stress conditions. (F) G3bp1-eGFP localizes to the cytoplasm under physiological conditions. (F’) Under stress conditions (temperature shock), G3bp1-eGFP localizes to stress granules (yellow arrowheads). Time in hours. n = 8 embryos. SG = stress granules. Scale bar = 50 µm. (G) mapre1b-mScarlet F1 stage 35 medaka embryo. Mapre1-mScarlet is expressed in a number of tissues and cell types including epithelial cells, muscle cells, the notochord, neuromasts and is highly expressed in the spinal cord. n = 5 embryos. SP = spinal cord. Scale bar = 50 µm.

Precise, single-copy knock-ins of fluorescent protein reporters

We next assessed the specificity and precision of the approach. It is possible that either concatemerization of inserts or off-target integrations could occur after foreign DNA delivery and CRISPR/Cas9-mediated DSBs (Gutierrez-Triana et al., 2018; Doench et al., 2016; Fu et al., 2013; Paix et al., 2017a, Won and Dawid, 2017; Yan et al., 2013; Hackett et al., 2007; Wierson et al., 2020; Auer et al., 2014; Shin et al., 2014; Winkler et al., 1991; Hoshijima et al., 2016; Kimura et al., 2014). To identify off-target insertions genome wide and verify single-copy integration we performed next generation WGS with high coverage (for details of WGS, see Materials and methods) on three KI lines (Figure 1B–D): eGFP-cbx1b, mScarlet-pcna, and mNeonGreen-myosinhc. For the eGFP-cbx1b KI line, we could only identify paired-end eGFP reads anchored to the endogenous cbx1b locus and nowhere else in the genome (Figure 1—figure supplement 2A). Likewise, in the mScarlet-pcna line, mScarlet reads only mapped to the endogenous pcna locus (Figure 1—figure supplement 2B). For the mNeonGreen-myosinhc line, mNeonGreen sequences mapped to the myosinhc locus (Figure 1—figure supplement 2C), but paired-end analysis yielded a second, weakly supported partial insertion of mNeonGreen at an intronic region in the edf1 gene. We were not able to confirm the latter insertion by subsequent PCR and hence it remains unclear whether this a false-positive prediction or a rare insertion of very low frequency. Combined, the WGS results therefore provide strong evidence that the method we report results in single-copy insertions only at the targeted locus. In addition to WGS, genotyping F1 adults followed by Sanger sequencing confirmed the generation of single-copy in-frame fusion proteins in the eGFP-cbx1b, mGreenLantern-cbx1b, mScarlet-pcna, mNeonGreen-myosinhc, cdh2-eGFP, mapre1b-mScarlet, and eGFP-rab11a lines (Figure 1—figure supplement 3, Supplementary file 3f, and Materials and methods). Only 2/14 Sanger sequenced junctions showed evidence of imprecise repair following HDR, one of which was a partial duplication within the 5′ homology arm, 22 basepairs upstream of the start codon while the other showed a partial duplication within the 3′ homology arm four basepairs after the stop codon (for details see Materials and methods), in both cases the coding sequence of the targeted genes is unaffected. Overall, the method we present here shows high precision and specificity enabling the rapid generation of endogenously tagged alleles in a vertebrate model.

Visualization of endogenous protein dynamics enables in vivo recording of cellular processes in medaka

As a proof of principle, we employed the simplified CRISPR/Cas9 strategy to generate a series of endogenous fusion protein KI medaka lines (Supplementary files 3a-f and 4; Figures 1 and 2, and Figure 2—figure supplements 24). Here, we provide an initial characterization of eight of these novel KI lines that are made available to the community, to label cell compartments (nucleus), cell processes (cell cycle, intra-cellular trafficking, stress granule formation), cell adhesion (adherens junctions), microtubules (plus-ends), and specific cell types (muscle cells).

Ubiquitous nuclear marker

To generate a ubiquitously expressed nuclear label reporter line, we targeted the cbx1b (Chromobox protein homolog) locus with eGFP and mGreenLantern (mGL). Cbx1b is a member of the chromobox DNA-binding protein family and is a known component of heterochromatin that is expressed ubiquitously (Lomberk et al., 2006; Nielsen et al., 2001). Chromobox proteins are involved in several important functions within the nucleus, such as transcription, nuclear architecture, and DNA damage response (Luijsterburg et al., 2009; Gilmore et al., 2016). We generated two KI lines eGFP-cbx1b and mGL-cbx1b by targeting either eGFP or mGL to the N-terminus of the cbx1b coding sequence in medaka. The resulting lines express the fluorescent reporter in all nuclei of every tissue examined, and serve as endogenous ubiquitous nuclear labels in teleosts (Figure 1B and Figure 2A, Figure 2—figure supplements 1 and 2, and Video 1, n > 10 embryos).

Video 1
Z-stack through the caudal fin region of a stage 39–40 cbx1-eGFP medaka embryo.

eGFP-Cbx1b is expressed in all nuclei of the different cell types in the caudal fin region. n > 10 embryos. Scale bar = 30 µm.

Proliferative cell marker

With the goal of generating an endogenous cell cycle reporter, we targeted the pcna (proliferating cell nuclear antigen) locus to generate a mScarlet-pcna fusion protein. Pcna is an essential protein regulator of DNA replication and integrity in eukaryotic cells (Moldovan et al., 2007; Maga and Hubscher, 2003; Mailand et al., 2013). It has been previously shown that cells that exit the cell cycle, for example post-mitotic differentiated cell types, express very low levels of Pcna (Zerjatke et al., 2017; Thacker et al., 2003; Yamaguchi et al., 1995; Buttitta et al., 2010; Alunni et al., 2010). This has led researchers to utilize Pcna as a highly conserved marker for proliferating cells (Zerjatke et al., 2017; Barr et al., 2016; Leonhardt et al., 2000; Leung et al., 2011; Piwko et al., 2010; Alunni et al., 2010; Thacker et al., 2003; Santos et al., 2015; Held et al., 2010). In addition to being a specific label for cycling cells, the appearance of nuclear speckles of Pcna within the nucleus is a hallmark of cells in late S phase of the cell cycle (Zerjatke et al., 2017; Barr et al., 2016; Leonhardt et al., 2000; Leung et al., 2011; Piwko et al., 2010; Santos et al., 2015; Held et al., 2010). More recently, endogenously tagged Pcna has been used in mammalian cell lines to dynamically score all the different cell cycle phases (Zerjatke et al., 2017). We targeted the first exon of pcna with mScarlet with high efficiency (28% mosaic expression in F0s, and 50% germline transmission) and generated the mScarlet-pcna KI line (Figure 1C and Figure 2B). Using stage 40 medaka embryos, we detected mScarlet-Pcna-positive cells within the epidermis, specifically in supra-basal epidermal cells (Figure 2B, n = 10 embryos). A subset of these cells showed nuclear speckles of mScarlet-Pcna that likely represent replication foci and are a characteristic marker for late S phase (Figure 2B, yellow arrowheads). We validate the use of this line both as an organismal-wide label for proliferative zones, and an endogenous cell cycle reporter in later sections.

Intra-cellular trafficking

To generate a reporter line allowing monitoring subcellular trafficking of endosomes and exosomes, we targeted Rab11a (Ras-Related Protein), a small GTPase and known marker of intra-cellular trafficking organelles in vertebrates (Welz et al., 2014; Cullen and Steinberg, 2018; Stenmark, 2009). We generated an N-terminus tagged eGFP-rab11a fusion protein that shows punctate intra-cellular signal most likely corresponding to trafficking organelles (Figure 2C, Figure 2—figure supplement 3, and Videos 24, n = 4 embryos). As a proof of principle, we detected high levels of eGFP-rab11a in cells of the spinal cord (Figure 2C, yellow arrowhead) and in neuromasts of the lateral line (Figure 2C, magenta arrowhead, Figure 2—figure supplement 3, and Video 4, n = 6 embryos). Using the eGFP-rab11a KI line, we were also able to observe dynamics of what appear to be intra-cellular organelle trafficking in vivo both in individual skin epithelial cells in the mid-trunk region and in the caudal fin region of developing medaka embryos (Videos 2 and 3, n = 4 embryos) providing initial evidence of the utility of this line as a possible subcellular trafficking marker in medaka.

Video 2
Live imaging in the caudal fin region of a stage 39–40 eGFP-rab11a medaka embryo.

eGFP-Rab11a is an intra-cellular trafficking marker and localizes to intra-cellular vesicles. Notice the dynamics of vesicle trafficking in epithelial cells, neuromasts (magenta arrowhead), and peripheral lateral line nerve (yellow arrowhead). Time in minutes. n = 4 embryos. Scale bar = 10 µm.

Video 3
Live imaging of skin epithelial cells in the mid-trunk region of a stage 39–40 eGFP-rab11a medaka embryo.

On the left panel is a merged view of epithelial cells in bright-field and eGFP-Rab11a in green. On the right panel, eGFP-Rab11a in grey scale. eGFP-Rab11a vesicles appear as granules within the cytoplasm of epithelial cells. Time in minutes. n = 4 embryos. Scale bar = 10 µm.

Video 4
Z-stack through the caudal fin region of a stage 39–40 eGFP-rab11a medaka embryo.

eGFP-Rab11a is strongly expressed in the caudal neuromast and peripheral lateral line nerve. And is also expressed in epithelial cells, the notochord, and the spinal cord. n = 4 embryos. Scale bar = 30 µm.

Stress granule marker

We were able to generate a g3b1-eGFP KI line by targeting eGFP to the 11th exon of the medaka g3bp1 gene. G3bp1 (GTPase activating protein SH3-domain-binding protein) is a DNA/RNA-binding protein and an initiating factor involved in stress granule formation (Irvine et al., 2004; Yang et al., 2020). Stress granules are non-membrane bound cell compartments, which form under cellular stress and accumulate non-translating mRNA and protein complexes, and play an important role in cellular protection by regulating mRNA translation and stability (Decker and Parker, 2012; Protter and Parker, 2016). Under normal conditions G3bp1-eGFP is expressed in the cytoplasm (Figure 2F, Video 5, n = 8 embryos) but upon stress (temperature shock), we observe that the protein changes its localization and accumulates in cytoplasmic foci corresponding to forming stress granules (Figure 2F’, yellow arrowheads, Video 5, n = 8 embryos). This is consistent with previous reports showing similar changes in the localization of G3bp1 in response to stress in a number of organisms (Guarino et al., 2019; Wheeler et al., 2016; Kuo et al., 2020). The initial characterization of the g3bp1-eGFP line shows its potential to serve as a real-time in vivo reporter for the dynamics of stress granules formation in a vertebrate model.

Video 5
Live imaging of stage 34–35 g3bp1-eGFP under normal conditions (temperature 21°C) reveals the cytoplasmic localization of G3bp1-eGFP in epithelial and muscle cells in the mid-trunk region of medaka embryos.

Upon stress conditions (temperature shift to 34°C, 60 min after the beginning of the time-lapse), G3bp1-eGFP localization begins to shift into localized clusters of stress granule puncta. Time in hours. n = 8 embryos. Scale bar = 50 µm.

Muscle cell marker

To label muscle cells, we targeted muscular myosin heavy chain with mNeonGreen (mNG). Myosins are a highly conserved class of motor proteins implicated in actin microfilament reorganization and movement (Sellers, 2000; Hartman and Spudich, 2012). We generated an N-terminus fusion of mNG-myosinhc KI that exclusively labels muscle cells (Figure 1D and Figure 2D and Figure 2—figure supplement 4, n > 10 embryos). In the medaka myotome, we were able to observe mNG-myosinhc chains of individual sarcomeres (A-bands separated by the I-bands), indicating that tagged Myosinhc is incorporated correctly in muscle fibers (Taylor et al., 2015; Loison et al., 2018). We use this line to record the endogenous dynamics of Myosinhc during muscle growth in vivo for the first time to the best of our knowledge, in a vertebrate model (Video 6, n = 9 embryos). The mNG-myosinhc line therefore enables the in vivo recording of endogenous Myosinhc dynamics during myogenesis in medaka.

Video 6
Live imaging of stage 34 mNG-myosinhc medaka embryo during muscle formation.

Muscle cell growth is driven by local buckling of individual muscle cells. Muscle growth and expression of mNG-Myosinhc does not seem to be polarized in an anterior–posterior or dorsal–ventral axis. Instead muscle cells have a heterogenous expression of mNG-Myosinhc that increases as muscle cells grow in length and mature. Time in hours. n = 9 embryos. Scale bar = 50 µm.

Cell adhesion marker

Cadherins are a highly conserved class of transmembrane proteins that are essential components of cell–cell adhesion and are thus expressed on cellular membranes (Leckband and de Rooij, 2014). A large number of cadherin genes exist in vertebrates where they exhibit tissue-specific expression patterns and are implicated in various developmental processes (Halbleib and Nelson, 2006). We decided to tag the C-terminus of medaka cadherin 2 (cdh2, n-cadherin) with eGFP. cdh2 is known to be expressed primarily in neuronal tissues in a number of vertebrates (Harrington et al., 2007; Suzuki and Takeichi, 2008). The cdh2-eGFP KI line shows cellular membrane expression in a variety of neuronal and non-neuronal tissues including the spinal cord, the eye, and the notochord (Figure 2E, n = 5 embryos) and neuromasts of the lateral line (Figure 2—figure supplement 3, n = 5 embryos, Video 7, n = 3 embryos), in addition to the developing heart (data not shown) (Chopra et al., 2011). The high expression of cdh2 in both differentiated notochord cell types (Figure 2E, Figure 2—figure supplement 3) has not been previously reported in medaka but is not unexpected as this tissue experiences a high level of mechanical stress and requires strong cell–cell adhesion (Lim et al., 2017; Adams et al., 1990; Garcia et al., 2017; Seleit et al., 2020). The cdh2-eGFP KI can be used to study dynamics of n-cadherin distribution in vivo during vertebrate embryogenesis (Video 8, n = 2 embryos).

Video 7
Z-stack through the posterior trunk region of a stage 39–40 cdh2-eGFP medaka embryo.

Cdh2-eGFP is expressed on the cellular membranes of neuronal tissue including neuromasts, the spinal cord, and the notochord. n = 3 embryos. Scale bar = 50 µm.

Video 8
Live imaging of the dorsal side of a cdh2-eGFP medaka embryo at the 12-somite stage reveals the endogenous dynamics of Cdh2-eGFP at high temporal resolution.

Time in minutes. n = 2. Scale bar = 30 µm.

Microtubule marker

Microtubule plus-end binding proteins are conserved regulators of microtubule dynamics, acting as a scaffold to recruit several additional proteins to ensure essential cell functions such as cell polarity, intra-cellular transport, and mitosis (Nehlig et al., 2017; Tirnauer and Bierer, 2000; Galjart, 2010). We successfully targeted the microtubule plus-end binding protein mapre1b (eb1), generating a C-terminal fusion protein with mScarlet. mapre1b-mScarlet is widely expressed in medaka embryos: epithelial cells, muscle cells, the notochord, and neuromasts all show mapre1b-mScarlet expression (Figure 2G, Video 9, n = 5 embryos), the highest level of expression occurs in the spinal cord (Figure 2G, yellow arrowhead). We were also able to record microtubule dynamics in the spinal cord of living embryos (Video 10, n = 5 embryos) highlighting the utility of this line for exploring the dynamics of microtubules in vivo during development.

Video 9
Z-stack through the trunk region of a stage 35 mapre1b-mScarlet medaka embryo.

Mapre1b-mScarlet is expressed in a number of tissues including epithelial cells, muscle cells, the notochord, neuromasts and is highly expressed in the spinal cord. n = 5 embryos. Scale bar = 100 µm.

Video 10
Live imaging of stage 35 mapre1b-mScarlet medaka embryo.

A close-up view on the developing spinal cord. Microtubule dynamics can be observed within the spinal cord (yellow arrowhead). Time in minutes. n = 5 embryos. Scale bar = 10 µm.

mScarlet-pcna: an organismal-wide marker for proliferative zones

We reasoned that the novel mScarlet-pcna line can act as an organismal-wide bona fide marker for the location of proliferative cells within any tissue or organ of interest. We therefore decided to generate double transgenic animals with eGFP-cbx1b as a ubiquitous nuclear marker and mScarlet-pcna as a label for cycling cells (Figure 3). As a proof of principle, we set out to investigate the location of proliferative zones in a number of organs and tissues in medaka. We began by assessing the position of proliferative cells in neuromast organs of the lateral line (Seleit et al., 2017b; Pinto-Teixeira et al., 2015; Romero-Carvajal et al., 2015). Neuromasts are small rosette shaped sensory organs located on the surface of teleost fish that sense the direction of water flow and relay the information back to the central nervous system (Seleit et al., 2017a; Romero-Carvajal et al., 2015; Jones and Corwin, 1993; Wada et al., 2013). They consist of four cell types: differentiated hair cells (HCs) in the very centre, underlying support cells (SCs), a ring of mantle cells (MCs), and neuromast border cells (nBCs) (Seleit et al., 2017b; Dufourcq et al., 2006). Previous work in medaka has established MCs to be the true life-long neural stem cells within mature neuromast organs (Seleit et al., 2017b). While the eGFP-cbx1b labels all neural cells within a mature neuromast organ (HCs, SCs, and MCs) (Figure 3A), mScarlet-pcna expression matches the previously reported location of proliferative MCs (Seleit et al., 2017b; Figure 3A’–A’’, white arrowhead). Neither the differentiated HCs nor the SCs directly surrounding them show evidence of Pcna expression in mature neuromast organs under homeostatic conditions in medaka (Figure 3A–A’’, n = 10 neuromast organs). Our results validate the utility of mScarlet-pcna as an in vivo marker of proliferative cells. Previous work has shown that nBCs are induced to form from epithelial cells that come into contact with neuromast precursors during organ formation and that these induced cells become the stem cell niche of mature neuromast organs (Seleit et al., 2017b). However, an open question is whether transformed nBCs are differentiated, post-mitotic cells or whether they remain cycling. Utilizing the mScarlet-pcna line we were able to observe nBCs (4/42) in late S phase of the cell cycle, as evident by the presence of nuclear speckles, in mature neuromast organs (Figure 3A’–A’’, yellow arrowheads). This provides direct evidence that nBCs retain the ability to divide and are thus not post-mitotic cells.

Figure 3 with 1 supplement see all
mScarlet-pcna line acts as an organismal-wide marker for proliferative zones.

(A–A’’) (eGFP-cbx1b) (mScarlet-pcna) double positive stage 40 medaka embryo. Maximum projection of a mature secondary neuromast (centre of image) within the lateral line system surrounded by epithelial cells, labelled by: endogenous eGFP-Cbx1b in (A) and endogenous mScarlet-Pcna in (A’). The merge is shown in (A’’). (A) eGFP-Cbx1b is a ubiquitous nuclear marker and labels all cell types within a mature neuromast. Those are: hair cells (HCs) in the centre of a neuromast, surrounded by support cells (SCs), and an outer ring of mantle cells (MCs) surrounded by the elongated neuromast border cells (nBCs). (A’) mScarlet-Pcna labels cycling cells, which are located at the very edge of the mature neuromast organ, a proportion of MCs express mScarlet-Pcna (white arrowheads). nBCs also express mScarlet-Pcna. Speckles can be seen in several mScarlet-Pcna-positive nBC nuclei (yellow arrowheads), indicating cells in late S phase of the cell cycle. (A’’) Merged image. n = 10 neuromast organs. NU = neuromast. Scale bar = 20 µm. (B–B’’) (eGFP-cbx1b) (mScarlet-pcna) stage 40 medaka embryo. Single Z-slice showing the medaka optic tectum. (B) eGFP-Cbx1b is a ubiquitous nuclear marker, (B’) whereas mScarlet-Pcna labels a subset of cells at the outer periphery of the optic tectum, indicating the position of proliferative cells in this tissue. A graded expression of mScarlet-Pcna is observed, with more central cells in the optic tectum losing the expression of mScarlet-Pcna. (B’’) Merged image. n = 4 embryos. OT = optic tectum. Scale bar = 30 µm. (C–C’’) Maximum projection of the pectoral fin of stage 40 medaka embryos. (C) eGFP-Cbx1b is a ubiquitous nuclear marker, (C’) while a subset of cells is labelled by mScarlet-Pcna indicating the position of proliferative cells. Note the proximal to distal gradient of mScarlet-Pcna expression, with proliferative cells at the base of the fin (left) and differentiated cells at the outer edges of the fin (right) (C’’) Merged image. n = 4 embryos. PC = pectoral fin. Scale bar = 50 µm.

Next, we turned our attention to the optic tectum, which is essential for integrating visuomotor cues in all vertebrates (Lavker and Sun, 2003; Alunni et al., 2010; Nguyen et al., 1999). We show that proliferative cells in the optic tectum of medaka are located at the lateral, caudal, and medial edge of the tectum in a crescent-like topology (Figure 3B–B’’, n = 4 embryos). Moreover, mScarlet-pcna expression is graded, with the more central cells gradually losing expression of Pcna (Figure 3B’). This is in line with previous histological findings using BrdU/IdU stainings in similarly staged medaka embryos (Nguyen et al., 1999; Alunni et al., 2010). We next analyzed the expression of mScarlet-pcna in the developing pectoral fin (Figure 3C–C’’, n = 4 embryos). We found that cells located proximally expressed the highest levels of mScarlet-pcna, with mScarlet-pcna expression decreasing gradually along the proximo-distal axis (Figure 3C’). To the best of our knowledge this proliferation pattern has not been previously reported and our data provide evidence that the differentiation axis of the pectoral fin is spatially organized from proximal to distal in medaka. Lastly, we reveal that proliferative cells are present in the spinal cord of stage 40 medaka embryos, a finding that has not been previously reported, and we show that these mScarlet-Pcna-positive cells occur in clusters preferentially located on the dorsal side of the spine (Figure 3—figure supplement 1, n = 4 embryos). The newly developed mScarlet-pcna line therefore acts as a stable label of proliferative cells and as such can be used to uncover the location of proliferation zones in vivo within organs or tissues of interest in medaka.

mScarlet-pcna: an endogenous cell cycle reporter

In addition to its use as a marker for cells in S phase, it has been shown that endogenously tagged Pcna can be used to determine all other cell cycle phases. This is based on the fact that both the levels and dynamic distribution of Pcna show reproducible characteristics in each phase of the cell cycle (Held et al., 2010; Piwko et al., 2010; Santos et al., 2015; Zerjatke et al., 2017; Leonhardt et al., 2000; Leung et al., 2011). To assess whether the endogenous mScarlet-pcna line recapitulates these known characteristic expression features during the cell cycle, we aimed to quantitatively analyze endogenous mScarlet-Pcna levels in individual cells during their cell cycle progression. To this end, we imaged skin epithelial cells located in the mid-trunk region of medaka embryos (Figure 4, Figure 4—figure supplements 1 and 2). Cells in the G1 phase of the cell cycle have been shown to decrease the levels of Pcna within the nucleus over time (Figure 4—figure supplement 1, Video 11, n = 9 epithelial cells) (Zerjatke et al., 2017). On the other hand, cells progressing through to S phase have been shown to increase the levels of Pcna expression within the nucleus over time (Leonhardt et al., 2000; Piwko et al., 2010; Leung et al., 2011; Santos et al., 2015; Barr et al., 2016; Zerjatke et al., 2017; Held et al., 2010). Indeed, we found that all tracked epithelial cells that eventually underwent a cellular division showed an increase in nuclear intensity of mScarlet-Pcna prior to the appearance of nuclear speckles (Figure 4A–B’’, Figure 4—figure supplement 1, Video 12, n = 9 epithelial cells). Nuclear speckles of Pcna mark the presence of replication foci in late S phase of the cell cycle (Leonhardt et al., 2000; Piwko et al., 2010; Leung et al., 2011; Barr et al., 2016; Zerjatke et al., 2017; Held et al., 2010). Previous work has also shown that the S/G2 transition can be identified as the point of peak pixel intensity distribution of endogenous Pcna within the nucleus (Zerjatke et al., 2017). We were are able to determine the peak pixel intensity distribution within nuclei by a combination of 3D surface plots and histograms of pixel intensity distributions over time (Figure 4B–D, Figure 4—figure supplements 2 and 3, and Videos 1214, n = 9 epithelial cells). Finally, onset of M phase is marked by a sharp decrease in nuclear levels of Pcna (Zerjatke et al., 2017; Leung et al., 2011; Piwko et al., 2010; Held et al., 2010), which we could consistently detect in the endogenous mScarlet-Pcna intensity tracks of epithelial cells undergoing division (Figure 4A–C, Figure 4—figure supplements 1 and 2, and Videos 1214, n = 9 epithelial cells). We therefore provide initial evidence that the mScarlet-pcna line recapitulates known dynamics of Pcna within the nucleus (Held et al., 2010; Piwko et al., 2010; Santos et al., 2015; Zerjatke et al., 2017; Barr et al., 2016; Leonhardt et al., 2000; Leung et al., 2011) and that it can therefore be utilized as an endogenous ‘all-in one’ cell cycle reporter in vertebrates.

Figure 4 with 3 supplements see all
Quantitative live cell tracking of endogenous mScarlet-Pcna levels enables cell cycle phase classification.

(A–A’’’) Selected frames from a time-lapse imaging of a mScarlet-Pcna-positive skin epidermal cell nucleus (yellow circle) undergoing cell division. The different phases of the cell cycle are deduced from mScarlet-Pcna expression as highlighted within the panels. Late S phase can be distinguished by the presence of nuclear speckles that correspond to replication foci. n = 9 skin epithelial cells. Time in hours. (B–B’’’) 3D surface plots of cell from (A). The S/G2 transition is marked as the point of peak pixel intensity distribution within the nucleus, which is reached at 15:40 hr (B’) and is equivalent to the largest width of mScarlet-Pcna pixel intensity distribution shown in panel D (red arrow). n = 9 skin epidermal cells. (C) Normalized mScarlet-Pcna intensity level within the nucleus from cell in (A–A’’’) over the course of one cell division. Vertical dashed red lines demarcate the different cell cycle phases based on the intensity and distribution of mScarlet-Pcna within the nucleus. Initially, an increase of endogenous mScarlet-Pcna expression over time indicates cells in S phase of the cell cycle. M phase is characterized by a sharp drop in nuclear mScarlet-Pcna levels beginning at 1200 min. (D) Width of pixel intensity distribution over time on cell from (A–A’’’). The S/G2 transition is marked as the point of peak pixel intensity distribution within the nucleus, which is reached at 940 min and is equivalent to the largest width of mScarlet-Pcna pixel intensity distribution (red arrow). n = 9 skin epidermal cells.

Video 11
Live imaging of stage 39–40 mScarlet-Pcna-positive non-dividing epithelial cell nucleus (shown in Figure 4—figure supplement 1).

Intensity profiles were extracted from within the nucleus (yellow circle). The cell does not divide over the course of the time-lapse. A decrease of mScarlet-Pcna levels over time is indicative of cells in the G1 phase of the cell cycle. Time in hours. n = 9 mScarlet-Pcna-positive non-dividing epithelial cells. Scale bar = 5 µm.

Video 12
Live imaging of stage 39–40 mScarlet-Pcna-positive epithelial cell nucleus undergoing cell division (shown in Figure 4).

Intensity profiles were extracted from within the nucleus (yellow circle). An increase of mScarlet-Pcna levels over time is indicative of cells in the S phase of the cell cycle. The appearance of nuclear speckles is indicative of cells in late S phase. The S/G2 transition is marked as the point of peak pixel intensity distribution within the nucleus (yellow circle). The sharp drop of endogenous mScarlet-Pcna levels is indicative of cells in M phase. Time in hours. n = 9 mScarlet-Pcna-positive and dividing epithelial cells. Scale bar = 15 µm.

Video 13
Left panel: mScarlet-Pcna-positive epithelial cell nucleus undergoing cell division (shown in Figure 4).

Middle panel: histogram of pixel intensity distribution of mScarlet-Pcna within the nucleus of the tracked cell. Right panel: frequency distribution width obtained from the histogram of pixel intensity distribution. Peak pixel intensity distribution is reached at 15:40 hr and is used to mark the S/G2 transition. Time in hours. n = 9 dividing epithelial cells. Scale bar = 15 µm.

Video 14
Left panel: 3D surface plot of mScarlet-Pcna-positive epithelial cell nucleus undergoing cell division (from Figure 4).

Middle panel: normalized endogenous mScarlet-Pcna levels of epithelial cell (from Figure 4). Right panel: histogram of pixel intensity distribution. Time in hours. n = 9 epithelial cells that undergo cell division.

Discussion

Despite the CRISPR/Cas9 system being repurposed as a broad utility genome editing tool almost a decade ago (Jinek et al., 2012; Cong et al., 2013; Wang et al., 2016) and despite its revolutionary impact as a method to generate knock-ins by HDR (Danner et al., 2017; Jasin and Haber, 2016; Ceccaldi et al., 2016; Lisby and Rothstein, 2004), there is still a paucity of precise, single-copy fusion protein lines in vertebrates, in general, and in teleost fish in particular. In fact, in medaka there are a total of three validated single-copy fusion protein lines by CRISPR/Cas9 reported prior to this work (Gutierrez-Triana et al., 2018), and in zebrafish, a handful of lines have been reported so far (Wierson et al., 2020; Hisano et al., 2015; Auer et al., 2014; Kimura et al., 2014; Li et al., 2019). This underscores the complexity of generating and validating precise single-copy fusion protein KI lines in teleost models. Previous techniques to generate large KIs (such as fluorescent reporters) required the usage of plasmid vectors commonly containing long homology arms (>200 bp) (Zu et al., 2013; Shin et al., 2014; Hoshijima et al., 2016; Kimura et al., 2014; Li et al., 2019). Problems arising during and after injection include DNA concatemerization of the donor construct (Gutierrez-Triana et al., 2018; Auer et al., 2014; Winkler et al., 1991; Hoshijima et al., 2016; Shin et al., 2014), in addition to possible imprecise and off-target integration of either the fluorescent protein sequence or the plasmid backbone (Auer et al., 2014; Gutierrez-Triana et al., 2018; Won and Dawid, 2017; Wierson et al., 2020; Shin et al., 2014; Hoshijima et al., 2016; Kimura et al., 2014; Li et al., 2019; Yan et al., 2013; Hackett et al., 2007). The vast majority of reported HDR-mediated knock-ins in teleosts rely on in vivo linearization of the plasmid donors. This strategy is utilized due to the observation that, although linear dsDNA donors can drive HDR, they might be prone to degradation, concatemerization, and are generally thought to be more toxic than plasmid donors (Auer et al., 2014; Cristea et al., 2013; Auer and Del Bene, 2014; Hisano et al., 2015; Hoshijima et al., 2016; Wierson et al., 2020; Yao et al., 2017; Winkler et al., 1991). Plasmid donors therefore contain an additional guide RNA sequence to drive in vivo linearization in order to synchronize the availability of the linear DNA donor with Cas9 activity (Auer et al., 2014; Cristea et al., 2013; Hisano et al., 2015; Hoshijima et al., 2016; Kimura et al., 2014; Li et al., 2019; Wierson et al., 2020; Yao et al., 2017). We reasoned that directly injecting PCR-amplified linear DNA with short homology arms (~35 bp) could be highly effective since these donors are relatively small (~780 bp) compared to plasmids (several kbs), and therefore a small quantity of donors (~10 ng/µl) will provide a large number of molecules (~20 nM) available to engage the HDR machinery following the Cas9-induced DSB.

Building on recent improvements in CRISPR/Cas9 KI strategies, we used 5′ biotinylated primers in order to limit in vivo concatemerization of the donor construct (Gutierrez-Triana et al., 2018), and synthetic sgRNAs were used to increase the efficiency of DSBs by Cas9 (Paix et al., 2015; Kroll et al., 2021; Hoshijima et al., 2019). In addition, we utilized a monomeric streptavidin-tagged Cas9 that has a high affinity to the biotinylated donor fragments to increase targeting efficiency (Gu et al., 2018). While 7/8 KI lines were generated and validated using Cas9 mSA and 5′ biotinylated donor fragments, the benefit of using the mSA/Biotin system to increase targeting efficiency in F0 needs to be further evaluated: in our initial comparison, using (2xNLS) Cas9 with and without mSA, and repair donors with and without biotinylation, we have found a comparable lethality and mosaic KI efficiency rates in F0s (Supplementary file 3g). We also tested another Cas9 containing only one NLS and no mSA (Gagnon et al., 2014) and found a lower lethality rate and a lower mosaic KI efficiency in F0s compared to Cas9 with two NLSs (Supplementary file 3g). Irrespective of these considerations, the approach reported here is a highly efficient, precise, and scalable strategy for generating single-copy fusion proteins (Supplementary files 3g and 4). The fact that the repair donors are synthesized by PCR amplification eliminates the need for both cloning and a second guide RNA for in vivo linearization. Therefore, the strategy we utilize significantly simplifies the process of endogenous protein tagging in a vertebrate model.

Very recently a similar approach to the one we present here showed the potential to generate CRISPR/Cas9-mediated KI lines in zebrafish by targeting non-coding genomic regions with PCR-amplified donor constructs (Levic et al., 2021). Together with our present work in Medaka, this supports the notion that donors with short homology arms are sufficient to drive HDR when they are in the form of linear dsDNA. One possible explanation is that following the Cas9-induced DSB, the donor is integrated by synthesis-dependent strand annealing (SDSA). During SDSA, the 3′ ends of chromosomal DNA strands at the DSB can anneal with the repair donor and drive its replication and insertion at the site of the DSB (Lisby and Rothstein, 2004; Danner et al., 2017; Jasin and Haber, 2016; Ceccaldi et al., 2016). Short homologies (30–40 bp) appear to be sufficient to anneal with chromosomal DNA and engage the SDSA machinery (Paix et al., 2017a, Paix et al., 2016; Grzesiuk and Carroll, 1987).

An important aspect of any KI strategy to generate fusion proteins is its precision. The validation process of single-copy insertions is complicated in approaches that use long homology arms (>200 bps) to generate knock-ins as concatemerization (Winkler et al., 1991; Gutierrez-Triana et al., 2018) and the formation of episomes (Aljohani et al., 2020; Wade-Martins et al., 1999; Udvadia and Linney, 2003; Winkler et al., 1991; Wierson et al., 2020) cannot be easily ruled out. Locus genotyping by PCR and Sanger sequencing is difficult when using primers external to the repair donor due to the large size of the expected fragment and competition for amplification with the wildtype allele. Internal primers within the donor (junction PCR) have been used to avoid this limitation, but this can lead to PCR artefacts and crucially, it does not rule out concatemerization of the injected dsDNA (Won and Dawid, 2017; Gutierrez-Triana et al., 2018; Auer et al., 2014; Hoshijima et al., 2016; Shin et al., 2014). Southern blotting is considered the gold standard to assess single-copy integration (Wierson et al., 2020; Gutierrez-Triana et al., 2018; Won and Dawid, 2017; Auer et al., 2014; Shin et al., 2014; Zu et al., 2013). While it has its advantages, Southern blotting depends on experimental design (genomic DNA preparation and digestion strategy) and probe design/sensitivity, and therefore cannot exclude that part of the donor construct or part of the vector backbone integrates elsewhere in the genome. Indeed, it has been reported that plasmid donors can lead to additional unwanted insertions in the genome (Won and Dawid, 2017; Auer et al., 2014; Hoshijima et al., 2016; Kimura et al., 2014; Li et al., 2019; Shin et al., 2014). We address those issues by performing WGS with high coverage on KI lines and provide evidence that our approach yields single-copy integration only at the desired locus. In addition, utilizing repair donors with short homology arms on both ends (30–40 bp) simplifies the validation of the insertion by using primers that sit outside the targeting donor fragment. These external primers can then be used for genotyping of the full insertion by simple PCR followed by Sanger sequencing to know the precise nature of the edit. We show that the usage of donor fragments with short homology arms, in combination with high coverage WGS, to be important aspects in validating the precision of single-copy CRISPR/Cas9-mediated KI lines in vertebrate models.

We were able to generate eight novel endogenous protein fusion lines that significantly expand the repertoire of genetic tools to track cellular dynamics in medaka. The eGFP-cbx1b and mGL-cbx1b KI lines serve as endogenous ubiquitous nuclear markers (Nielsen et al., 2001; Lomberk et al., 2006). The generation of truly ubiquitous lines by transgene overexpression in teleost fish (Centanin et al., 2014; Burket et al., 2008) is a difficult endeavour and requires constant monitoring for variegation and silencing (Goll et al., 2009; Akitake et al., 2011; Burket et al., 2008; Stuart et al., 1990). Yet these ubiquitous fluorescent reporter lines are invaluable tools for researchers. Ubiquitous fusion proteins expressed from the endogenous locus avoid potential issues with transgene overexpression and variegation. The highly conserved cbx1b locus could therefore provide an alternative strategy to generate faithful ubiquitous nuclear markers in other teleosts and non-model organisms. In addition, this locus could serve as a landing site for ubiquitous expression of genetic constructs (e.g. utilizing a 2A self-cleaving peptide) in medaka (Li et al., 2019; Kim et al., 2011). Next, we validate the use of g3bp1-eGFP KI as a stress granule formation marker, and utilizing 4D live imaging show the formation of stress granules in response to temperature shock in real time, as previously shown in other models using a variety of stress conditions (Guarino et al., 2019; Kuo et al., 2020; Wheeler et al., 2016; Decker and Parker, 2012; Protter and Parker, 2016). This line can therefore be used both as an in vivo marker of stress conditions and to study the process of stress granule formation. The eGFP-rab11a line serves as an intra-cellular trafficking (Welz et al., 2014; Cullen and Steinberg, 2018; Stenmark, 2009) marker that allows us to dynamically follow exosomes and endosomes in vivo. We report that both neuromasts and the spinal cord show substantially higher expression of rab11a than other tissues, the basis of this remains unclear but could indicate that these tissues exhibit higher levels of protein turnover. Despite being a highly conserved protein involved in myogenesis (Sellers, 2000; Hartman and Spudich, 2012), no endogenous KI of any myosin family member has been reported in teleosts. The mNeonGreen-myosinhc KI enables the detection and recording of endogenous myosin dynamics in vivo during muscle growth in a vertebrate model. We also generate cdh2-eGFP KI line (Leckband and de Rooij, 2014; Halbleib and Nelson, 2006) and show that it is expressed in a tissue-specific manner primarily in the spinal cord, neuromasts and the notochord. Since N-cadherin has been shown to be involved in epithelial–mesenchymal transition (EMT) (Harrington et al., 2007; Suzuki and Takeichi, 2008; Desclozeaux et al., 2008), this line can be used to study dynamical changes in N-cadherin distribution in vivo facilitating our understanding of EMT and other fundamental cell adhesion processes in vertebrates. Lastly, we generate and characterize the mScarlet-pcna KI line and discuss its usage and implications across teleosts below.

An overarching goal of developmental and stem cell biology is to discover the location of stem and progenitor cells in different organs and tissues, followed by a molecular characterization of their properties (Rhee et al., 2006; Nowak et al., 2008; Snippert et al., 2010; Buczacki et al., 2013; Lu et al., 2012; Lavker and Sun, 2003). Major advances have relied on finding resident stem cell markers that differentiates stem cells from other cell types within the same tissue, followed by BrdU/IdU staining to confirm their proliferative abilities (Nguyen et al., 1999; Rhee et al., 2006; Nowak et al., 2008; Nowak and Fuchs, 2009; Alunni et al., 2010; Snippert et al., 2010; Lu et al., 2012; Buczacki et al., 2013; Stolper et al., 2019; Tsingos et al., 2019). However, BrdU/IdU staining requires the sacrifice of the animal precluding the ability to perform 4D live imaging to analyze stem cell behaviour in vivo over time. The medaka KI line with endogenously labelled Pcna that we present here helps to circumvent this limitation. In addition, since Pcna is expressed exclusively in cycling cells (Yamaguchi et al., 1995; Thacker et al., 2003; Buttitta et al., 2010; Zerjatke et al., 2017; Alunni et al., 2010), it has the potential to be used to discover the location of proliferative zones in vivo within any organ or tissue of interest. We provide proof-of-principle evidence that the mScarlet-pcna KI line acts as a bona fide marker for proliferative zones in a variety of tissues in medaka fish. This line therefore represents an important new tool for stem cell research in medaka. A similar strategy could be adopted to generate endogenously tagged Pcna both in the teleost field and in other organisms.

In addition to its use as a bona fide marker for proliferative zones, we provide evidence that the mScarlet-pcna line can be used as an endogenous cell cycle reporter in medaka. It has previously been shown that both the levels and dynamic distribution of Pcna are indicative of the different cell cycle phases (Held et al., 2010; Piwko et al., 2010; Santos et al., 2015; Zerjatke et al., 2017; Leonhardt et al., 2000; Barr et al., 2016; Leung et al., 2011). This led researchers to successfully utilize it as an ‘all-in-one’ cell cycle reporter in mammalian cells (Held et al., 2010; Piwko et al., 2010; Santos et al., 2015; Zerjatke et al., 2017; Barr et al., 2016; Leonhardt et al., 2000). By quantitatively tracking endogenous Pcna levels during one cell cycle in epidermal cells of medaka fish, we were able to confirm the dynamic nature of mScarlet-pcna expression, which correlated with the previously described behavior of the Pcna protein within the nucleus of other vertebrates (Held et al., 2010; Piwko et al., 2010; Santos et al., 2015; Zerjatke et al., 2017; Barr et al., 2016; Leonhardt et al., 2000; Leung et al., 2011). As such, we provide proof-of-principle evidence that the mScarlet-pcna line can be successfully used as an endogenous cell cycle reporter in a teleost model. Using the visualization of endogenous Pcna for cell cycle phase classification offers an attractive alternative to cell cycle reporters that rely on the insertion of two-colour transgenes, such as the FUCCI system (Sugiyama et al., 2009; Dolfi et al., 2019; Araujo et al., 2016; Bajar et al., 2016; Oki et al., 2014; Sakaue-Sawano et al., 2008). First, by using endogenous fusion proteins there is no requirement for overexpression of cell cycle regulators. Second, the potential issue with transgene variegation and silencing is avoided (Akitake et al., 2011; Goll et al., 2009; Burket et al., 2008; Stuart et al., 1990). Finally, utilizing a single-colour cell cycle reporter allows its simultaneous use with other fluorescent reporters during live-imaging experiments. Due to the high conservation of Pcna in eukaryotes, developing Pcna reporters in other model organisms using a similar strategy is an attractive possibility to pursue.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Oryzias latipes) CabOtherMedaka Southern wild-type population
Strain, strain background (Oryzias latipes)eGFP-cbx1bThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)mGL-cbx1bThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)cdh2-eGFPThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)g3bp1-eGFPThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)mapre1b-linker-3XFlag-mScarletThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)mNG-HAtag-Linker-myosinhcThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)mScarlet-pcnaThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes)eGFP-rab11aThis paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
Strain, strain background (Oryzias latipes) (eGFP-cbx1b) × (mScarlet-pcna)This paperCRISPR KI line, Aulehla Lab EMBL Heidelberg
AntibodyPrimary rabbit anti-GFP (polyclonal)TorreyPines Biolabs#TP4011:500
AntibodySecondary goat anti-rabbitAbcam AlexaFluor 488#ab1500771:500
Recombinant DNA reagentCas9-mSA plasmidPMID:29889212
Recombinant DNA reagentCas9 no-mSA plasmidPMID:29889212
Recombinant DNA reagentCas9 plasmidPMID:24873830
Sequence-based reagentgRNA actbSigmaSequence (spacer)GGAUGAUGACAUUGCCGCAC
Sequence-based reagentgRNA cbx1bSigmaSequence (spacer)GGAAGAUGUGGCAGAAGAAG
Sequence-based reagentgRNA cdh2SigmaSequence (spacer)GGGAGCGAUGACUAAGACAA
Sequence-based reagentgRNA g3bp1SigmaSequence (spacer)CCCCAGCGAGAGCCGCUUCU
Sequence-based reagentgRNA mapre1bSigmaSequence (spacer)UCCAGAUGCUGAGGAACAGG
Sequence-based reagentgRNA myosinhcSigmaSequence (spacer)CAUCUCUGCGUCAGUGCUCA
Sequence-based reagentgRNA pcnaSigmaSequence (spacer)GACCAGGCGAGCCUCAAACA
Sequence-based reagentgRNA rab11aSigmaSequence (spacer)UCGGAUUAACGCGAGGACGA
Commercial assay or kitRNAeasyMiniKitQiagen#74104
Commercial assay or kitQIAquickGel Extraction KitQiagen#28115
Commercial assay or kitmMachineSP6 Transcription KitInvitrogen#AM1340
Software, algorithm CC-TopPMID:25909470
Software, algorithm EnsemblPublic
Chemical compound, drug Hoechst 33342Thermo Fischer#H35701:500 dilution of the 10 mg/ml stock solution
Chemical compound, drug TricaneSigma-Aldrich#A5040-25G
Chemical compound, drug LMP AgaroseBiozymePlaque Agarose #840,1010.6% in 1× H2O

Animal husbandry and ethics

Request a detailed protocol

Medaka (O. latipes, Cab strain) (Iwamatsu, 2004; Naruse et al., 2004; Kasahara et al., 2007) were maintained as closed stocks in a fish facility built according to the European Union animal welfare standards and all animal experiments were performed in accordance with European Union animal welfare guidelines. Animal experimentation was approved by The EMBL Institutional Animal Care and Use Committee (IACUC) project code: 20/001_HD_AA. Fishes were maintained in a constant recirculating system at 27–28°C with a 14 hr light/10 hr dark cycle.

Cloning-free CRISPR/Cas9 knock-ins

Request a detailed protocol

A detailed step-by-step protocol for the cloning-free approach is provided in Supplementary files 1 and 2. A detailed list of all repair donors, PCR primers, fluorescent protein sequences, and sgRNAs used is provided in Supplementary file 3a-e. Briefly, for the preparation of Cas9-mSA mRNA: the pCS2+ Cas9 mSA plasmid was a gift from Janet Rossant (Addgene #103882, Supplementary file 3d; Gu et al., 2018). 6–8 µg of Cas9-mSA plasmid was linearized by Not1-HF restriction enzyme (NEB #R3189S). The 8.8 kb linearized fragment was cut out from a 1.5% agarose gel and DNA was extracted using QIAquick Gel Extraction Kit (Qiagen #28115). In vitro transcription was performed using mMachine SP6 Transcription Kit (Invitrogen #AM1340) following the manufacturer’s guidelines. RNA cleanup was performed using RNAeasy Mini Kit (Qiagen #74104). Other Cas9 encoding plasmids used were pCS2+ Cas9 (Addgene #122948) (Gu et al., 2018) and pCS2-Cas9 (Addgene #47322) (Gagnon et al., 2014; Supplementary file 3d,g). sgRNAs were manually selected using previously published recommendations (Paix et al., 2017a; Paix et al., 2019; Doench et al., 2016; Gagnon et al., 2014; Paix et al., 2017b) and in silico validated using CCTop and CHOPCHOP (Labun et al., 2019; Stemmer et al., 2015; Supplementary file 3e). The genomic coordinates of all genes targeted can be found in Supplementary file 3a. Synthetic sgRNAs used in this study were ordered from Sigma-Aldrich (spyCas9 sgRNA, 3 nmol, HPLC purification, no modification). PCR repair donor fragments were designed and prepared as described previously (Paix et al., 2014; Paix et al., 2017b, Paix et al., 2015; Paix et al., 2016; Paix et al., 2017a) and a detailed protocol is provided in Supplementary file 1. Briefly the design includes approximately 30–40 bp of homology arms and a fluorescent protein sequence with no ATG or stop codon (Supplementary file 3b, d). PCR amplifications were performed using Phusion or Q5 high fidelity DNA polymerase (NEB Phusion Master Mix with HF buffer #M0531L or NEB Q5 Master Mix # M0492L). MinElute PCR Purification Kit (Qiagen #28004) was used for PCR purification. Primers were ordered from Sigma-Aldrich (25 nmol scale, desalted) and contained Biotin moiety on the 5′ ends for repair donor synthesis. A list of all primers and fluorescent protein sequences used in this study can be found in Supplementary file 3c, d. The injection mix in medaka contains the sgRNA (15–20 ng/µl) + Cas9 mSA mRNA (or Cas9 mRNA without mSA) (150 ng/µl) + repair donor template (8–10 ng/µl). For injections, male and female medakas are added to the same tank and fertilized eggs collected 20 min later. The mix is injected in one-cell staged medaka embryos (Iwamatsu, 2004), and embryos are raised at 28°C in 1XERM (Seleit et al., 2017a; Seleit et al., 2017b; Rembold et al., 2006). A list of KI lines generated and maintained in this study can be found in Supplementary file 3f.

Live-imaging sample preparation

Request a detailed protocol

Embryos were prepared for live imaging as previously described (Seleit et al., 2017a; Seleit et al., 2017b). 1× Tricaine (Sigma-Aldrich #A5040-25G) was used to anaesthetize dechorionated medaka embryos (20 mg/ml – 20× stock solution diluted in 1XERM). Anaesthetized embryos were then mounted in low melting agarose (0.6–1%) (Biozyme Plaque Agarose #840101). Imaging was done on glass-bottomed dishes (MatTek Corporation Ashland, MA, USA). For g3bp1-eGFP live imaging, temperature was changed from 21 to 34°C after 1 hr of imaging.

Immunofluorescence

Request a detailed protocol

Immunohistochemistry was performed as previously described (Centanin et al., 2014). Primary rabbit anti-GFP antibody (Torrey Pines Biolabs #TP401) was used at a 1:500 dilution from the stock solution. Secondary goat anti-rabbit antibody (Abcam AlexaFluor 488 #ab150077) was used at a 1:500 dilution from the stock. Hoechst 33342 (Thermo Fischer #H3570) was used with a dilution of 1:500 of the 10 mg/ml stock solution.

Microscopy and data analysis

Request a detailed protocol

For all embryo screening, a Nikon SMZ18 fluorescence stereoscope was used. All live-imaging, except for g3bp1-eGFP and cdh2-eGFP embryos, was done on a laser-scanning confocal Leica SP8 (CSU, White Laser) microscope, ×20 and ×40 objectives were used during image acquisition depending on the experimental sample. For the SP8 confocal equipped with a white laser, the laser emission was matched to the spectral properties of the fluorescent protein of interest. g3bp1-eGFP line live imaging was performed using a Zeiss LSM780 laser-scanning confocal with a temperature control box and an Argon laser at 488 nm, imaged through a ×20 plan apo objective (numerical aperture 0.8). For cdh2-eGFP, 4D live imaging was performed on a Luxendo TruLive SPIM system using a ×30 objective. Open-source standard ImageJ/Fiji software (Schindelin et al., 2012) was used for analysis and editing of all images post-image acquisition. Stitching was performed using standard 2D and 3D stitching plug-ins on ImageJ/Fiji. For quantitative values on endogenous mScarlet-Pcna dynamics, ROI manager in ImageJ/Fiji was used to define fluorescence intensity within the nucleus of tracked cells (yellow circle in Figure 4 and Videos 10 and 11), fluorescent intensity measurements were then extracted from the time series and the data was normalized by dividing on the initial intensity value in each time-lapse movies. Data was plotted using R software. Pixel intensity distribution within nuclei were analyzed using a custom python based script (Source code file 1). Individual live-cell tracks were plotted using PlotTwist (Goedhart, 2020).

Fin-clips, genotyping, and Sanger sequencing

Request a detailed protocol

Individual adult F1 fishes were fin clipped for genotyping PCRs. Briefly, fish were anaesthetized in 1× Tricaine solution. A small part of the caudal fin was cut by sharp scissors and placed in a 2 ml Eppendorf tube containing 50 µl of fin-clip buffer. The fishes were recovered in small beakers and were transferred back to their tanks. Eppendorf tubes were then incubated overnight at 65°C. 100 µl of H2O was then added to each tube and then the tubes were incubated for 10–15 min at 90°C. Tubes were then centrifuged for 30 min at 10,000 rpm in a standard micro-centrifuge. Supernatant was used for subsequent PCRs. Fin-clip buffer is composed of 0.4 M Tris–HCl pH 8.0, 5 mM EDTA pH 8.0, 0.15 M NaCl, 0.1% SDS in H2O. 50 µl of proteinase K (20 mg/ml) was added to 1 ml fin-clip buffer before use. 2 µl of genomic DNA from fin-clips was used for genotyping PCRs. A list of all genotyping primers used in this study can be found in Supplementary file 3c. After PCRs the edited and wild-type amplicons were sent to Sanger sequencing (Eurofins Genomics). Sequences were analyzed using Geneious software (Figure 1—figure supplement 3). In-frame integrations were confirmed by sequencing for eGFP-cbx1b, mScarlet-pcna, mNG-myosinhc, eGFP-rab11a, mapre1b-mScarlet, mGL-cbx1b, and cdh2-eGFP. We were able to detect an internal partial duplication of the 5′ homology arm in the mScarlet-pcna line that does not affect the protein coding sequence nor the 5′ extremity of the homology arm itself. Specifically, 22 basepairs upstream of the start codon of pcna (and within the 5′ homology arm); we detect a 21-bp partial duplication of the 5′ homology arm (CGCAACCCTCCACAGAATAAC) and a 7-bp insertion (GGTCGAC) indicative that the repair mechanism involved can lead to errors (Paix et al., 2017a, Wierson et al., 2020). The 5′ homology junction itself is unaltered and precise. We were also able to detect a partial duplication (26 basepairs) of the 3′ homology arm in the cdh2-eGFP line (TTTCCTCGGTGTGGACCTTCCTACTT) that does not affect the protein coding sequence and occurs four basepairs after the stop codon.

Whole genome sequencing

Request a detailed protocol

Five to ten positive F1 medaka embryos (originating from the same F0 founder) of the eGFP-cbx1b, mScarlet-pcna, and mNeonGreen-myosinhc lines were snap frozen in liquid nitrogen and kept at −80°C in 1.5 ml Eppendorf tubes. Genomic DNA was extracted using DNeasy Blood and Tissue Kit (Qiagen #69504) according to the manufacturer’s guidelines. The libraries were prepared on a liquid handling system (Beckman i7 series) using 200 ng of sheared gDNA and 10 PCR cycles using the NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB #E7645S). The DNA libraries were indexed with unique dual barcodes (8 bp long), pooled together and then sequenced using an Illumina NextSeq550 instrument with a 150 PE mid-mode in paired-end mode with a read length of 150 bp. Sequenced reads were aligned to the O. latipes reference genome (Ensembl! Assembly version ASM223467v1) using BWA mem version 0.7.17 with default settings (Li and Durbin, 2009). The reference genome was augmented with the known inserts for eGFP, mScarlet, and mNeonGreen to facilitate a direct integration discovery using standard inter-chromosomal structural variant (SV) predictions. The insert sequences are provided in Supplementary file 3b, d. After the genome alignment, reads were sorted and indexed using SAMtools (Li et al., 2009). Quality control and coverage analyses were performed using the Alfred qc subcommand (Rausch et al., 2019). For SV discovery, aligned reads were processed with DELLY v0.8.7 (Rausch et al., 2012) using paired-end mapping and split-read analysis. SVs were filtered for inter-chromosomal SVs with one breakpoint in one of the additional insert sequences (eGFP, mScarlet, and mNeonGreen). Plots shown in Figure 1—figure supplement 2 were made using Integrative Genomics Viewer (IGV) (Thorvaldsdóttir et al., 2013). The estimated genomic coordinates for integration are: eGFP-cbx1b (chr19:19,074,552), mScarlet-pcna (chr9:6,554,003), and mNeonGreen-myosinhc (chr8:8,975,799). Coverage of eGFP-cbx1b_gDNA1 is 20.4× and eGFP-cbx1b_gDNA2 is 23.6×. Coverage of mScarlet-pcna is 14.4×. Coverage of mNeonGreen-myosinhc is 14.5×. Raw sequencing data were deposited in European Nucleotide Archive (ENA) under study number ERP127162. Accession numbers are: eGFP-cbx1b(1) ERS5796960 (SAMEA8109891), eGFP-cbx1b(2) ERS5796961 (SAMEA8109892), mScarlet-pcna ERS5796962 (SAMEA8109893), and mNeonGreen-myosinhc ERS5796963 (SAMEA8109894).

Data availability

Sequencing data have been deposited in European Nucleotide Archive (ENA) under study number ERP127162. Accession numbers are: eGFP-cbx1b(1) ERS5796960 (SAMEA8109891), eGFP-cbx1b(2) ERS5796961 (SAMEA8109892), mScarlet-pcna ERS5796962 (SAMEA8109893) and mNeonGreen-myosinhc ERS5796963 (SAMEA8109894).

The following data sets were generated

References

Decision letter

  1. Didier YR Stainier
    Senior and Reviewing Editor; Max Planck Institute for Heart and Lung Research, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

[Editors' note: this paper was reviewed by Review Commons.]

Decision letter after peer review:

Thank you for submitting your article "Endogenous protein tagging in medaka using a simplified CRISPR/Cas9 knock-in approach" for consideration by eLife. Your article has been reviewed by 3 peer reviewers at Review Commons, and the evaluation at eLife has been overseen by Didier Stainier as the Senior and Reviewing Editor.

Based on the previous reviews and the revisions, the manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1) Importantly, the control experiments at F0 including non-biotinylated donor templates and non-SA tagged Cas9 are now provided in the revised version (Table S7), and inconclusive results were obtained in these experiments. Comparable efficiency was obtained using both Cas9 with and without mSA, and repair donors with and without biotinylation. Moreover, likely due to the limited size of the samples, statistics analyses were not applied to these data. While these observations have only limited impact as the efficiency of the protocol is maintained independently of the biotin-streptavidin system, the manuscript needs to be clarified in terms of the mSA/Biotin dependence. Thus, this point should be openly discussed in the text to avoid misinterpretations on how the protocol works. Although this is currently mentioned in the discussion (page 10: lines 464-474), it should also be included in the Results section, and the mention of streptavidin tagged Cas9 should be removed from the abstract, and the recommendation to use streptavidin tagged Cas9 should be removed from the methods section. Moreover, the central scheme of the protocol in Figure 1A seems misleading in light of the new controls, and it should be modified to be in agreement with the fact that "the utility of the mSA/Biotin system to increase targeting efficiency in F0 needs to be further evaluated".

https://doi.org/10.7554/eLife.75050.sa1

Author response

Based on the previous reviews and the revisions, the manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

1) Importantly, the control experiments at F0 including non-biotinylated donor templates and non-SA tagged Cas9 are now provided in the revised version (Table S7), and inconclusive results were obtained in these experiments. Comparable efficiency was obtained using both Cas9 with and without mSA, and repair donors with and without biotinylation. Moreover, likely due to the limited size of the samples, statistics analyses were not applied to these data. While these observations have only limited impact as the efficiency of the protocol is maintained independently of the biotin-streptavidin system, the manuscript needs to be clarified in terms of the mSA/Biotin dependence. Thus, this point should be openly discussed in the text to avoid misinterpretations on how the protocol works. Although this is currently mentioned in the discussion (page 10: lines 464-474), it should also be included in the Results section, and the mention of streptavidin tagged Cas9 should be removed from the abstract, and the recommendation to use streptavidin tagged Cas9 should be removed from the methods section. Moreover, the central scheme of the protocol in Figure 1A seems misleading in light of the new controls, and it should be modified to be in agreement with the fact that "the utility of the mSA/Biotin system to increase targeting efficiency in F0 needs to be further evaluated".

We thank the senior editor for his comments. We have followed all suggestions.

We have replaced Cas9-mSA in the abstract with Cas9 mRNA (line 26).

We have integrated the results from Supp Table 7 into the Results section as suggested (line 128-131).

We have updated the scheme in Figure 1 and removed the reference to Cas9-mSA, instead replacing it with Cas9 mRNA.

We have updated the methods section to reflect our usage of other Cas9s (without mSA). We have also removed the recommendation of using the Cas9-mSA from the methods section in Supplementary file 1 (the protocol) under the section “Cas9 mRNA in vitro transcription”.

[Editors' note: we include below the reviews that the authors received from Review Commons, along with the authors’ responses.]

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

This is an excellent follow up study that uses new approaches to deliver cargos for homology-directed repair in Medaka. The goal of the study was to improve the rates of targeted knock-in of DNA to create fluorescently tagged alleles to follow protein localization and gene expression. The authors nicely characterized a scarlet-pcna line that can be used to follow cell division. This should be widely requested and cited. They also created eGFP-cbx1b to follow nuclei and mNeongreen-myosin-hc line to visualize muscle, an eGFP-rab11a line to visualize vesicle transport, a g3b1-eGFP to follow stress granules, and a cdh2-eGFP line to follow cellular junctions. The real contribution here might be the method. For ease of production, the authors use PCR templates that have biotin at the ends to block concatemer formation with only short regions of homology, around 30-40 bases. They describe precise junctions following targeting, minimal off-target effects, and high rates of integration (11-59% F0 showing mosaic expression, 25-100% germline transmission, 6-50% positive F1). While the use of short homology, biotin modification of DNA ends and the use of a Cas9 fused to monomeric streptavidin (Cas9-mSA) have all been previously described, their use together to obtain high rates of targeted integration is impressive. The ability to simply add your homology arms by PCR simplifies and streamlines the production of templates. I look forward to trying this in my own laboratory.

We thank the Reviewer for his/her comment and agree that the ease of the strategy we provide here in addition to its high efficiency will make it the method of choice for CRISPR/Cas9 mediated fluorescent protein KIs in organisms.

As the paper was under review we were able to generate two additional KI lines. mScarletmapre1b, Mapre1b (EB1) is a protein that is known to bind to the positive ends of microtubules and we also generated a second version of the cbx1b KI using the newly engineered mGreenLantern as a fluorescent reporter (mGL-cbx1b). We have integrated this new data in the manuscript text, Figures and Supplementary files accordingly.

The data presented is very convincing and has some beautiful images, however the following points could be clarified and/or supported with data:

Diagram in S1 represents the homology arms as very long compared to the cargo. Please fix the diagram for clarity.

We thank the reviewer for his/her comment and have modified Figure S1 as suggested.

The need for Cas9-mSA is not demonstrated. For example, an experiment demonstrating the difference in activity between Cas9 alone and Cas9 fused to mSA using short homology. This could be done simply in F0 fish.

We thank the reviewer for his/her comment and have followed this suggestion.

We have done new rounds of F0 injections to address the points raised by the reviewer(s) using Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) without mSA and a second Cas9 (Ct SV40NLS) without mSA (from the Schier lab, Gagnon et al., 2014). This new data is presented in Table S7. The results show that Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) without mSA show comparable lethality rates and mosaic KI efficiency in F0s, with Cas9-mSA having slightly higher rates in both. In addition, a second Cas9, i.e. Cas9 (Ct SV40NLS) without mSA (from the Schier lab) shows a lower lethality rate and a lower mosaic KI efficiency in F0s compared to the experiments in which we used Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) or Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS).

While the paper was under review we were able to generate a new KI line mGL-cbx1b utilizing Cas9 (Ct SV40NLS) without mSA, albeit with slightly lower efficiency of F0 targeting and germline transmission (updated Table 1 and Table S7). Combined, these results indicate that the strategy we employ also works, in principle, using a Cas9 without mSA. As we have obtained higher efficiencies and importantly, done the validation, confirming single-copy integration specifically in the locus of interest, using Cas9-mSA, we recommend using Cas9mSA.

All this new information is added to the updated manuscript.

Data presented in S2 (A, B, C) is very difficult to see and interpret. Please consider revising the figure to enable the font to be read and the data to be interpreted. More description of the conclusions from these experiments would also help the readers.

We thank the reviewer for his/her comment. We have split Figure S2 into Figure S2 and S3 and have increased the font and size of the panels as suggested. In addition, we have augmented the figure legend of S2 to more clearly state the conclusions obtained from this figure as the reviewer suggests.

The manuscript is nicely written and does not overstate the results. The authors were very clear about previous work that led them to the current methodology. This was nicely referenced. The paper also nicely outlines homology arm design.

Reviewer #1 (Significance (Required)):

As alluded to above the individual aspects of the methodology are well documented in other studies, however, the real significance of the paper is the combination of these methodologies to achieve high rates of precise targeted integration and ease of cargo template production. I predict that this will become the method of choice for targeted integration in model organisms and will be highly referenced. Our laboratory has published similar rates of targeted integration using plasmid templates, but given the ease of production, I look forward to trying and possibly switching to this technology.

We thank the reviewer for his/her comments and are looking forward to his/her lab members trying this approach.

Referee Cross-commenting

I agree, but selection is part of the protocol. Their F0 numbers are very high, making selection maybe not as important? I do agree that their numbers are not higher than previous work, but the ease of what they are doing streamlines design. Other advantages include not having a vector and single copy integration.

We agree with Reviewer 1. In addition, we would like to point out that we present 3 measures of efficiency: 1 – F0 mosaic targeting 2 – F0 germline transmission rate 3 – germline transmission efficiency (% of positive F1 embryos per single transmitting F0). All three measures are comparable to highly efficient ones previously reported for medaka (the model organism we are using) (Gutierrez-Triana et al. 2018) and all three are on par (Wierson et al., 2020) or better (Levic et al., 2021) than ones published in zebrafish. As the reviewer acknowledges, a key additional benefit of this approach is its simplicity of implementation, combined with ease of targeting/validation and single copy integration.

I think all three reviewers are in agreement that the utility of Cas9-mSA needs to be addressed. The authors previous paper looked at capping templates with biotin, but this would be an easy control to examine in the current study too, +/- biotin. Wierson et al. showed there does not seem to be a difference between long and short homology arms, so I am not sure how important this is to address here.

We thank the reviewer for his/her comment. We want to point out that the primary reason for using biotin is preventing donor construct concatemerization as previously shown in medaka (Gutierrez-Triana et al. 2018). That said, we have performed new F0 injections with and without biotin (Table S7) and results indicate that there is no clear difference in mosaic KI efficiency in F0s between them.

Regarding the utility of Cas9-mSA, please see our response to reviewer 1 earlier, who raised this point, too.

The efficiency we report using short homology arms is very comparable to a highly efficient KI method previously shown in medaka using long homology arms (Gutierrez-Triana et al., 2018) and is on par (Wierson et al., 2020) or better than recently reported KIs (Levic et al., 2021) in zebrafish. Given the ease of generation of KI constructs (single PCR), the ease of validating single insertions (by sanger sequencing spanning the entire locus) and the added benefit of cost reduction of our approach we believe it will be widely preferred and utilized.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Seleit et al. utilize CRISPR/Cas9-generated Double Strand Breaks (DSBs) to stimulate Homology Directed Repair (HDR) that results in the insertion of sequences encoding reporter proteins into the reading frames of selected genes. They generate many original lines in zebrafish that will be useful to others in the research community. The work is well designed and well performed, the data is clear, the expression patterns are often beautiful, and the lines they create are likely to be useful.

We thank the reviewer for his/her comment. We would like to point out that we are working with Oryzias latipes (medaka), not zebrafish, which is important when evaluating and comparing our results with previous published approaches, see below.

The authors claim they are introducing an improved method for accomplishing HDR. This claim is not well substantiated, and the advantage of their method over other published approaches is not demonstrated.

We would like to answer this general criticism by citing reviewer 1” As alluded to above the individual aspects of the methodology are well documented in other studies, however, the real significance of the paper is the combination of these methodologies to achieve high rates of precise targeted integration and ease of cargo template production (no cloning required to make the repair donor). I predict that this will become the method of choice for targeted integration in model organisms and will be highly referenced.” We share this view of reviewer 1.

The authors utilize an approach in zebrafish that has been shown to have some promise in mice and in tissue culture cells.

We report about our experiments in medaka, not zebrafish, see above.

The idea of the approach is to tether donor template DNA to a CRISPR/Cas9 Ribonucleoprotein (RNP) complex so that repair template is present very close to the targeted locus at the time a chromosomal DSB is generated.

The authors accomplish this by injecting fertilized zebrafish eggs with a cocktail containing mRNA encoding Cas9 linked to monomeric streptavidin (mSA), a synthetic single guide RNA molecule, and linear dsDNA template that is biotinylated at its 5' ends. The linear donor molecules are created by PCR. The ends of these molecules carry about 40bp of homology to the left and right borders of the DSB break in the host locus. HDR results in "pasting in" protein-coding sequences into the endogenous coding locus so that the introduced sequences are in-frame and expressed from the targeted locus.The method of using short homology arms to introduce donor sequences at a CRISPR/Cas9-targeted site has been explored quite well by other researchers in the field, notably Wierson et al., 2020 and Hisano et al. 2015. The gene fusion of Cas9 linked to monomeric streptavidin was developed by the Rossant lab and the method for generating the donor template by PCR was developed in the Wittbrodt lab. So the questions are: what is new here and is the method an advance? The new element is to introduce a combination of Cas9-mSA and biotinylated donor template to drive HDR in the zebrafish. Does the method in fact lead to interactions between Cas9-mSA and donor molecules? Does the method increase efficiency of HDR events? These questions are not answered.

We would like to point out that our method uses PCR donors with short homology arms, no plasmid repair donors, no in vivo linearization, and no long homology arms, as is reported in previous methods. This clearly simplifies the process of repair donor synthesis and KI generation. We also provided strong evidence for the specificity of such donors using Whole Genome Sequencing. In addition, and as we discuss in the manuscript, the KI lines generated represent important new tools for the community.

There are two critical pieces of information that are missing:

– First and foremost: the authors fail to perform parallel, necessary control experiments with non-biotinylated donor templates. Thus it is impossible to know whether biotinylation might enhance complex formation with the RNPs and might enhance the frequency of HDR events.

As pointed out by all reviewers, this is indeed an important point that we have now addressed in the revised manuscript. We performed additional injection experiments, summarized in new Table S7. In these experiments, we compared F0 targeting efficiency based on visual screening of fluorescence and using Cas9 either with or without mSA and in addition, testing the effect of using biotinylated vs non-biotinylated primers. From these results, we conclude that Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) without mSA show comparable lethality rates and mosaic KI efficiency in F0s, with Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) having slightly higher rates in both.

Results also indicate that the targeting efficiency in F0 using Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) with and without biotinylated primers is comparable. As the purpose of biotinylated primers has been shown previously (Gutierrez-Triana et al., 2018) to be the prevention of concatemerization of exogenous donor DNA, our advocated approach in this manuscript promotes usage of biotinylated primers in order to increase single-copy integration events.

– Second: it is very difficult to decipher the true frequency of HDR events from their work. The authors inject the cocktails in fertilized eggs, and then select F0 embryos that express the reporter protein. Some portion of these (perhaps only those with extensive expression of the reporter?) are raised to adulthood and tested for the ability to transmit a correctly modified locus to the next generation. As reported in Table S1, very very few F0 animals are tested for transmission of a modified locus. Because of the selection of F0's to be tested, it is hard to know what portion of all the injected F0's have modified genomes that can be transmitted. This strong bias makes it difficult to assess the actual HDR efficiency in donor-injected F0 embryos and to compare the present results with previous published results.

This is an important point as indeed, the approach includes a selection step based on the visual inspection of fluorescence in F0 injected founders. We have adjusted the representation and text accordingly with the aim to further clarify how to estimate efficiencies. To this end, we calculate rates using three complementary readouts:

1. F0 mosaic targeting based on fluorescence screening;

2. Germline transmission rate, pooled: percentage of GLT across tested F0;

3. Germline transmission rate, individual: GLT rate based on positive F1 embryos, per single transmitting F0.

We report that on all three measures, the KI efficiencies we obtain are comparable to highly efficient ones previously reported for medaka (the model organism we are using) (Gutierrez-Triana et al., 2018). In addition, all three measures are on par (Wierson et al., 2020) or better (Levic et al., 2021) than ones published in zebrafish. The added advantage is the simplicity of our approach, its ease of use, ease of validation and precise single copy integration.

We believe that pre-selection is an important aspect of the time and resources saving strategy we report here. Pre-selection is also widely reported and used in previous literature on the subject when reporting on efficiency (for instance, Wierson et al., 2020, Levic et al., 2021, Gutierrez-Triana et al. 2018). But in a hypothetical in which one could not pre-select, the approach we report here would still be extremely useful as the KI efficiencies in F0 and the germline transmission rates are quite high. If a researcher were to inject 100 embryos and have an early lethality of around 40% this leaves him/her with 60 embryos. The researcher can grow all 60 embryos to adulthood and then perform fin-clips on all 60 to identify positive carriers of the insertion (with an average KI efficiency of around 20%, close to what we report, this would be 12 fish). The researcher can then outcross the 12 positive fishes to the WT strain and screen for founders. While the effort put in and resources used are substantially higher than with pre-selection, the method is guaranteed to work.

Minor Points

Main Text

Line 127:.…or a mosaic insertion of very low frequency.

This seems to be incorrect. Because the WGS is performed on F1 embryos, even if a mosaic insertion occurred with very low frequency in the F0 animals, it would be heterozygous in F1 progeny, but not mosaic.

We thank the reviewer for his/her comment. The reviewer is correct in stating that any insertion will be heterozygote in F1. However, since we pool F1 embryos from the same founder for the F1 deep sequencing it is possible that not all embryos show the same insertion. We have changed the phrasing to ‘rare’ instead of ‘mosaic’ to clarify this point.

Line 365: Should this be Figure 4B-B’ but not 4C-C’?

We thank the reviewer for his/her comment and have corrected this error.

Figure 1 legend

Line 574:.… the PCR amplified donor plasmid containing short homology arms.… It should be donor "fragment" not “plasmid".

We thank the reviewer for his/her comment and have corrected this error.

Figure 4 legend

Line 657: beginning at 1200 minutes rather than 20:00h.

We have modified the text accordingly.

Supplemental Figure 1 legend (A)? (B)? or one panel?

We have modified the text accordingly.

Supplemental Figure 2 legend

The explanation is not enough to understand the supplemental figure 2.

The WGS analysis revealed targeted integration of fluorescent markers, but the reason for a single copy insertion is not clearly explained.

We have split Figure S2 into Figure S2 and S3 and have increased the font and size of the panels as suggested. In addition, we have augmented the figure legend of S2 to outline the evidence for single copy insertion more clearly. Please note Figure S2 is just a visualization using IGV (Interactive Genome Viewer), the details of the bioinformatic tools and steps used to arrive at our conclusions are included in the Materials and methods section: if the donor construct integrated elsewhere in the genome we expect to see mapped reads of the fluorescent reporter sequence overlapping with other endogenous loci. If concatemerization occurs at one locus, we expect to see reads of the fluorescent reporter repeated in tandem with short homology arms in between. We do not see evidence for either. In addition, Sanger sequencing confirms single copy integration.

Please explain the turquoise bars in A and in B and the dark grey bars in C that are mapped away from predicted integration sites.

As we state stated in the figure legend the turquoise bars represent inter-chromosomal reads that span both the eGFP sequence and the endogenous cbx1b locus. Regarding the fact that some are mapped slightly away from the predicted integration site: it all depends on where the sequencing read starts. If a fluorescent reporter sequence read happens to be at the very end of the fluorescent reporter sequence we expect that it maps slightly away from the predicted insertion site (which is the beginning of the FP sequence).

What are orange bars?

Orange/red bars are anomalous mappings that either deviate from the expected insert size or paired-end mapping orientation. We have added this in the figure legend.

What are dark grey bars in A and B?

There are no dark grey bars in B.

As per information from IGV: dark grey bars in A represent “unpaired mate, mate not mapped or otherwise unknown status.” The mapping quality of these two reads is 0 and therefore they most likely represent false mapping. We have added this in the figure legend.

What are purple bars in B?

There is one purple bar in B. IGV has flagged this one read as possibly an inter-chromosomal read between chromosome 9 and chromosome 17. The mapping quality of this read is 0 which again means it is in all likelihood a false mapping. We added this in the figure legend.

In D, E, F, each chromatograph shows single peaks, but schematic bars above the chromatographs indicate heterozygotes – is this a conflict?

We are showing sequencing results only for the edited allele in the new Figure S3. The WT allele was also sent to sequencing for all three genes but is not shown here. The schematic bars directly above the chromatographs are simply four colors representing ATGC. They do not indicate heterozygosity. Any difference between the sequencing results and the in-silico in-frame fusion protein is indicated as a black bar within the consensus coverage on the top most panel. Since there are no differences between the in-silico and the sequenced reads this means there are no mismatches. We have modified the figure legend to indicate the location of the consensus sequence more clearly.

Supplemental Figure 5 legend

Yellow arrowhead in (B): it looks like "otic" vesicle sensory organs but not "optic" vesicle sensory organ.

We thank the reviewer for spotting the mistake. We have corrected it to otic vesicle.

Yellow arrowhead in (G): eGFP expression cannot be seen in the panel.

We thank the reviewer for his/her comment. We have adjusted the location of the Yellow arrowheads in (G) to more clearly represent the vacuolated cells of the notochord. Please note that vacuolated cells have a highly peculiar morphology where the vast majority of the cell body is filled with one large (unlabeled) vacuole. The eGFP signal is therefore weak and located at the thin plasma membrane. We have enhanced the signal in the corresponding panel and have adjusted the text to reflect the weak labeling.

Materials and methods

Line 955 – 996: Authors should mention whether or not each of the biotinylated primers contains phosphorothioate bonds as described in Gutierrez-Triana et al.

We thank the reviewer for his/her comment. We updated the manuscript material and methods, and file S1, to clearly state that the biotinylated primers do not contain any additional modifications.

Line 1004 – 1006: should describe concentration of each component in fin-clip buffer, rather than volume of each stock solution.

We thank the reviewer for his/her comment. We have adjusted the text accordingly.

Reviewer #2 (Significance (Required)):

In sum, the zebrafish lines and reagents provided here will be highly useful to the community.

In contrast the assertion that the authors have developed a new and improved method for achieving Homology-Directed Repair in zebrafish is not well-supported by the experiments.

Our work was done in medaka, and we agree, these lines and the ones that will be generated with this approach will prove invaluable new tools for the medaka community.

As above, we would agree with reviewer 1 ” As alluded to above the individual aspects of the methodology are well documented in other studies, however, the real significance of the paper is the combination of these methodologies to achieve high rates of precise targeted integration and ease of cargo template production. I predict that this will become the method of choice for targeted integration in model organisms and will be highly referenced.”

Referee Cross-commenting

I still think there is insufficient indication that the efficiency they achieve is any different from what anyone else has achieved in the past. all data boils down to Table S1 which tests very few animals, each of which were pre-selected for expression. what would happen if an investigator could not pre-select? I don't think they establish efficiency at all.

We believe that pre-selection is an important aspect of the time and resources saving strategy we report here. Pre-selection is also widely reported and used in previous literature on the subject when reporting on efficiency (for instance, Wierson et al., 2020, Levic et al., 2021, Gutierrez-Triana et al. 2018). But in a hypothetical in which one could not pre-select, the approach we report here would still be extremely useful as the KI efficiencies in F0 and the germline transmission rates are quite high. If a researcher were to inject 100 embryos and have an early lethality of around 40% this leaves him/her with 60 embryos. The researcher can grow all 60 embryos to adulthood and then perform fin-clips on all 60 to identify positive carriers of the insertion (with an average KI efficiency of around 20%, very close to what we report, this would be 12 fish). The researcher can then outcross the 12 positive fishes to the WT strain and screen for founders. While the effort put in and resources used are substantially higher than with pre-selection, the method will still work.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

In the article "Endogenous protein tagging in medaka using a simplified CRISPR/Cas9 knock-in approach" Seleit et al. describe a knock-in method for efficient tagging of endogenous proteins by CRSPR-mediated homologous recombination. The protocol takes advantage of previous advances by combining fluorescent reporters flanked by biotinylated short homology arms together with streptavidin-Cas9. The authors have used this approach to tag six different loci in medaka, generating a useful collection of reporter lines (particularly mScarlet-Pcna), initially characterized in this work. The main claims of the article are on the advantages and efficiency of the developed knock-in method and on the added value provided by the lines generated.

Overall, the authors provided enough evidence to support the main conclusions of the article. The work is technically sound and I would only request a few additional controls indicated below. The manuscript is well written, the topic is well introduced, the relevant literature referenced, and the data in general terms nicely presented (see minor comment for a few exceptions). Particularly useful are the descriptions of the lines generated (and yet see again minor comments for a few exceptions) and, in particular, the accurate report of the knock-in protocol, which will facilitate other groups to replicate the method.

Major Comments:

1. Beyond previous knock-in strategies in medaka focused in blocking the 5' ends of the donor molecules (Gutierrez-Triana et al., 2018), the authors are now combining in medaka methods adapted from Gu et al. 2018 (i.e. linking donor to the Cas9 via a streptavidin-biotin system) together with short homology arms (Wierson et al. 2020, Levic et al. 2021). Although the strategy seems extremely effective, the current description does not allow to deduce which of the modifications of the protocol had a bigger impact on its efficiency. Understanding this aspect may be important for further methodological refinements. Particularly, it is difficult to know if the 5' biotin block of the template is by itself sufficient to achieve the high % of integration or if it requires the interaction with the Streptavidin-tagged Cas9. Although it is likely that this interaction is relevant to increase targeting efficiency, this has not been formally demonstrated in the medaka context. A few control injections using regular Cas9, could resolve this issue in F0 embryos.

We thank the reviewer for his/her comment. We have performed the control experiments suggested by the reviewer using Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) without mSA. In addition, we have tested another Cas9 (Ct SV40NLS), from the Schier lab that is widely used in the teleost community. This new data is presented in Table S7.

The results show that Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) without mSA show comparable lethality rates and mosaic KI efficiency in F0s, with Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) having slightly higher rates in both. In addition, a second Cas9 without mSA, i.e. Cas9 (Ct SV40NLS) from the Schier lab shows a lower lethality rate and a lower mosaic KI efficiency in F0s compared the experiments in which we used Cas9 (Nt SV40NLS + Ct NucleoplasminNLS) or Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS).

As we have obtained highest efficiencies with Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) and importantly, have done the validation confirming single-copy integration specifically in the locus of interest using Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS), it is the recommended approach put forward in our manuscript. We have updated the manuscript with this new data.

2. Related to the previous point: Although the authors already included a control for the presence of the homology arms, additional controls comparing biotinylated vs unbiotinylated templates will also be informative.

We have performed the control experiments suggested by the reviewer and they are presented in the new Table S7. Results indicate that the targeting efficiency in F0 using Cas9-mSA (Nt SV40NLS + Ct NucleoplasminNLS) with and without biotinylated primers is comparable. As the function of biotinylated primers has been shown previously (Gutierrez-Triana et al., 2018) to be the prevention of concatemerization of exogenous donor DNA, our advocated approach in this manuscript promotes usage of biotinylated primers in order to increase single-copy integration events.

3. In general terms the observations are sufficiently replicated, and the number of embryos observed are properly indicated in the tables and figure legends. Statistic analyses are not required for this work in principle. However, the authors should clarify whether the representation of the mScarlet-PCNA distribution vs time (Figure 4 and Supplementary Figure 8) corresponds to the average plot of 9 epithelial nuclei recorded independently or is a representative individual track.

All data presented in Figure 4 represent measurements from a single cell. In the newly numbered Figure S9 we present individual measurements of mScarlet-Pcna intensity from 9 epithelial cells, in addition to the average (trend). In the newly numbered Figure S10 we present data from the single cell shown in Figure 4A. A similar analysis was done on all 8 other cells and we show the results in the new Figure S11.

Given the importance of this line as a tool to follow in vivo the cell cycle progression, it would be necessary to show how variable these measurements are, either by plotting the average distribution + SD, or by showing several individual recordings.

We plot the individual measurements of intensity over time in Figure S9 and we also plot their average combined. Cells progressing into S phase show an increase of endogenous Pcna levels, peak pixel intensity distribution marks the transition from S/G2 (seen as bright speckles in the time-series) and a sharp drop in endogenous nuclear Pcna levels marks the beginning of M phase (Figure S9). In addition, we have now included individual measurements of pixel intensity distribution over time from all 8 other cells marking the transition from S/G2 in new Figure S11 as the reviewer suggests.

Minor points:

1. Although in general the knock-in lines are sufficiently described in this work; a few co-staining experiments will help to explore the universality of the tools generated. For example, a simple co-staining with DAPI may help to deduce if the line eGFP-cbx1b is a general nuclei marker.

We have performed a double staining as requested. Results show the overlap between Hoechst and the anti-eGFP antibody in the eGFP-cbx1b line. The results are presented in new Figure S5.

Similarly, a co-labelling with an endosomal marker would be a nice conformation of the intracellular distribution of the eGFP-Rab11a fusion.

We have confirmed that the fusion protein generated is in-frame. Future work will focus on a detailed characterization of the line.

2. Figures and videos of medaka tissues (figures 1-3, Supplementary figures 3-6, and Supplementary videos 1-7) are in general poorly labelled. In order to facilitate the interpretation of the data through self-explanatory figures, I would recommend adding a few abbreviations indicating key anatomical landmarks in each panel. It would also help including a legend stating which structure is included in each panel, as it has been done in Figure 3.

We thank the reviewer for his/her comment. We have added the suggested annotations in Figure 1, 2 and 3 and have modified the figure legends as requested. And we have also added the suggested annotations in Supplementary Figures 4-8 and modified the legends as requested.

3. The manuscript includes some unnecessary claims of novelty that do not add much to the interpretation of the text and are in some cases questionable [e.g. page 10; "first reported fusion-protein for a cadherin"; or "first reported endogenous ubiquitous nuclear marker in teleosts"; page 8 "provide first evidence that the mScarlet-pcna line recapitulates known dynamics of Pcna within the nucleus]. I would suggest to remove them.

We thank the reviewer for his/her comment. We have removed references to ‘first’ in the two instances outlined. We noted that our statement "provide first evidence that the mScarlet-pcna line recapitulates known dynamics of Pcna within the nucleus” was misleading and have corrected it to read "provide initial evidence that the mScarlet-pcna line recapitulates known dynamics of Pcna within the nucleus”.

4. Text accompanying Supplementary figure S2, as well as the figure itself, are difficult to follow. Increasing the fonts size and the resolution of the images in panels A-C will improve their interpretation. In the text, it will also help referring to the individual panels instead of the whole figure. e.g. "identify paired-end eGFP reads anchored to the endogenous cbx1b locus.… (Figure S2)"…should be Figure S2A.

We thank the reviewer for his/her comment and have followed all the suggestions.

5. Figure legend of S5: the otic capsule (yellow arrowhead in B) is wrongly indicated as "optic vesicle”.

We thank the reviewer for spotting the mistake. We have adjusted the text accordingly.

Reviewer #3 (Significance (Required)):

While the work is technically very solid and the described method seems valuable, I still have mixed feelings on its potential impact in the field. The fact that the technology developed is a refinement of published methods (Particularly Gu et al. 2018, Gutierrez-Triana et al., 2018, Wierson et al. 2020) reduces the novelty of the study. That being said, I think the protocol guarantees a precise and high-efficient integration of the reporters with limited experimental effort, and thus the advance is significant.

We thank the reviewer for these supporting comments.

The method will be of interest to groups working in teleost genetics (such as this referee), mainly to those working with medaka, which may take advantage of the helpful knock-in lines generated. In that regard, the impact of the work could be greater if a few comparative experiments, testing the protocol applicability in other teleost models, were included. This is particularly the case for zebrafish, given the size of the research community working with this model.

We agree with the reviewer that future work should address the potential of this approach in other organisms including zebrafish, but this is beyond the scope of this paper. Our initial results from pilot injections in a number of other organisms suggest that the approach is both functional and efficient, as such we are expecting its adoption both in the teleost field and most likely beyond.

Referee Cross-commenting

I do agree with the last comment by reviewer #1. Selection is not as important. Being practical, the current protocol guarantees an efficiency level that is already high enough. Improving this aspect is likely not a key methodological demand.

https://doi.org/10.7554/eLife.75050.sa2

Article and author information

Author details

  1. Ali Seleit

    Developmental Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    For correspondence
    ali.seleit@embl.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8144-2286
  2. Alexander Aulehla

    Developmental Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Conceptualization, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3487-9239
  3. Alexandre Paix

    Developmental Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany
    Contribution
    Conceptualization, Data curation, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    For correspondence
    alexandre.paix@embl.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8080-7546

Funding

H2020 European Research Council (866537)

  • Alexander Aulehla
  • Ali Seleit

EMBL interdisciplinary Postdoc (847543)

  • Ali Seleit

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank all members of the Aulehla lab for the fruitful discussions on the work presented here. We would like to thank Aissam Ikmi for input on the manuscript. We would like to thank Takehito Tomita for help with Python scripts. The European Molecular Biology Laboratory (EMBL-Heidelberg). Genecore is acknowledged for support in WGS data acquisition and analysis. We would like to thank Vladimir Benes and members of his team at Genecore EMBL Heidelberg for continuous help and support, Tobias Rausch for computational work on the WGS data and Mireia Osuna Lopez for help in library preparation of WGS DNA. In addition, we would like to thank all animal-care takers at EMBL Heidelberg and in particular Sabine Goergens for excellent support. We would also like to thank Addgene for access to plasmids. This work was supported by the European Molecular Biology Laboratory (EMBL-Heidelberg) and the EMBL interdisciplinary Postdoc (EIPOD4) under Marie Sklodowska-Curie Actions Cofund (grant agreement number 847543) fellowship for funding to Ali Seleit. This work also received support from the European Research Council under an ERC consolidator grant agreement no. 866537 to AA.

Ethics

Medaka (Oryzias latipes, Cab strain) (Iwamatsu, 2004; Naruse et al., 2004; Kasahara et al., 2007) were maintained as closed stocks in a fish facility built according to the European Union animal welfare standards and all animal experiments were performed in accordance with European Union animal welfare guidelines. Animal experimentation was approved by The EMBL Institutional Animal Care and Use Committee (IACUC) project code: 20/001_HD_AA. Fishes were maintained in a constant recirculating system at 27–28°C with a 14 hr light /10 hr dark cycle.

Senior and Reviewing Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Publication history

  1. Received: October 27, 2021
  2. Accepted: December 5, 2021
  3. Accepted Manuscript published: December 6, 2021 (version 1)
  4. Accepted Manuscript updated: December 8, 2021 (version 2)
  5. Version of Record published: December 21, 2021 (version 3)

Copyright

© 2021, Seleit et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,185
    Page views
  • 379
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Ali Seleit
  2. Alexander Aulehla
  3. Alexandre Paix
(2021)
Endogenous protein tagging in medaka using a simplified CRISPR/Cas9 knock-in approach
eLife 10:e75050.
https://doi.org/10.7554/eLife.75050

Further reading

    1. Developmental Biology
    2. Evolutionary Biology
    Alexandre P Thiery et al.
    Research Article Updated

    Development of tooth shape is regulated by the enamel knot signalling centre, at least in mammals. Fgf signalling regulates differential proliferation between the enamel knot and adjacent dental epithelia during tooth development, leading to formation of the dental cusp. The presence of an enamel knot in non-mammalian vertebrates is debated given differences in signalling. Here, we show the conservation and restriction of fgf3, fgf10, and shh to the sites of future dental cusps in the shark (Scyliorhinus canicula), whilst also highlighting striking differences between the shark and mouse. We reveal shifts in tooth size, shape, and cusp number following small molecule perturbations of canonical Wnt signalling. Resulting tooth phenotypes mirror observed effects in mammals, where canonical Wnt has been implicated as an upstream regulator of enamel knot signalling. In silico modelling of shark dental morphogenesis demonstrates how subtle changes in activatory and inhibitory signals can alter tooth shape, resembling developmental phenotypes and cusp shapes observed following experimental Wnt perturbation. Our results support the functional conservation of an enamel knot-like signalling centre throughout vertebrates and suggest that varied tooth types from sharks to mammals follow a similar developmental bauplan. Lineage-specific differences in signalling are not sufficient in refuting homology of this signalling centre, which is likely older than teeth themselves.

    1. Developmental Biology
    2. Evolutionary Biology
    Sophie Pantalacci
    Insight

    The tooth shape of sharks and mice are regulated by a similar signaling center despite their teeth having very different geometries.