The genetic code stored within DNA provides cells with the instructions they need to carry out their role in the body. Any changes to these genes, or the DNA sequence around them, has the potential to completely alter how a cell behaves.

Scientists have developed various tools that allow them to experimentally modify the genome of cells or even entire living organisms. This includes the popular Cas9 enzyme which cuts DNA at specific sites, and base editors which can precisely change bits of genetic code without cutting DNA. While there are lots of Cas9 enzymes and base editors currently available, these often differ greatly in their activity depending on which cell type or organism they are applied to.

Finding a tool that can effectively modify the genome of an organism at the right time during development also poses a challenge. All the cells in an organism arise from a single fertilized cell. If this cell is genetically edited, all its subsequent daughter cells (which make up the entire organism) will contain the genetic modification. However, most genome editing tools only work efficiently later in development, resulting in an undesirable mosaic organism composed of both edited and non-edited cells.

Here, Thumberger et al. have developed a new ‘high efficiency-tag’ (also known as hei-tag for short) that can enhance the activity of gene editing tools and overcome this barrier. The tag improves the efficiency of gene editing by immediately shuttling a Cas9 enzyme to the nucleus, the cellular compartment that stores DNA. In all cases, gene editing tools with hei-tag worked better than those without in fish embryos and mouse cells grown in the laboratory. When Cas9 enzymes connected to a hei-tag were injected into the first fertilized cell of a fish embryo, this resulted in an even distribution of edited genes spread throughout the whole organism.

To understand how a gene affects an organism, researchers need to be able to edit it as early in development as possible. Attaching the ‘hei-tag’ to already available tools could help boost their activity and make them more efficient. It could also allow advances in medical research aimed at replacing faulty genes with fully functioning ones.

In the last decade, the CRISPR/Cas9 system and its derivatives facilitated and revolutionized genome editing across all phyla (Nidhi et al., 2021). The efficiency of editing crucially depends on the on-site activity of the particular Cas9 enzymes used (usually Streptococcus pyogenes Cas9, SpCas9) in the nucleus. State-of-the-art Cas9 variants differ by peptide tags added to the N- and C-termini of the respective endonuclease resulting in reported different activities (Liu et al., 2021; Zhang et al., 2014). Employed tags usually comprise diverse nuclear localization signals (NLSs) and epitope tags (e.g. FLAG, Myc, HA) for potential protein purification or visualization. To achieve nuclear localization of the Cas9 enzyme, the monopartite NLS originating from the SV40 large T-antigen (Kalderon et al., 1984) or a bipartite NLS discovered in Xenopus nucleoplasmin is routinely employed (Dingwall et al., 1988). However, the nuclear localization activity of commonly used NLSs is tightly controlled during early development (Poon and Jans, 2005) and is first detectable during gastrulation. In fish embryos, an optimized artificial NLS (Inoue et al., 2016) (oNLS) facilitates prominent nuclear localization already immediately after fertilization, while the SV40 NLS acts most prominently much later and facilitates nuclear localization approximately at the 1000-cell stage. For high targeting efficiency with low mosaicism, a peak activity should be achieved in the zygote or at early cleavage stages. Here, we present the hei-tag, a short bipartite tag composed of a myc-tag and optimized NLSs at the N- and C-termini, that boosts Cas9 or cytosine-to-thymine (C-to-T) base editor-mediated targeted genome editing in organismo and cell culture.

Assessing the genome editing efficiency requires a reliable and quantitative readout based on an apparent phenotype. We established a quantitative assay for loss-of-eye pigmentation to address the activity of different Cas9 variants in two teleost model systems, medaka (Oryzias latipes) and zebrafish (Danio rerio) covering a wide evolutionary distance of 200 million years (Furutani-Seiki and Wittbrodt, 2004). Our assay on retinal pigmentation provides a highly reproducible quantitative readout for the loss of the conserved transporter protein oculocutaneous albinism type 2 (oca2), required for melanin biosynthesis (Figure 1a). Only its bi-allelic inactivation results in the loss of pigmentation of eyes and skin (Lischik et al., 2019). A prominent knock-out phenotype thus can either result from a single to few early events, or from many events at subsequent developmental stages. Although phenotypically indifferent, the allele variance (genetic mosaicism) reflects the time point of action.

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 1—source data 1

Raw data for quantifications shown in Figure 1d and g.

https://cdn.elifesciences.org/articles/70558/elife-70558-fig1-data1-v3.csv

Download elife-70558-fig1-data1-v3.csv

State-of-the-art protocols employ high concentrations of Cas9 and respective sgRNAs to ensure efficient on-site editing. To facilitate uniform Cas9 action, we followed our successful mRNA injection protocol (Gutierrez-Triana et al., 2018). One-cell stage medaka embryos were co-injected with sgRNAs targeting the oca2 gene (OlOca2 T1, T2) together with mRNA encoding a Cas9 endonuclease and mRNA encoding the green fluorescent protein (GFP) as injection tracer. Injected embryos were fixed at 4.5 days post fertilization (dpf) (Iwamatsu, 2004) well after the onset of pigmentation in control injections and subjected to image analysis (Figure 1b). In brief, the eyes were segmented, (residual) pigmentation was thresholded (Figure 1c–c’) and quantified according to mean gray values (0, i.e. fully pigmented, 255, i.e. completely unpigmented, Figure 1d).

We first established the base activity level for the assay at standard conditions with high molar excess (150 ng/µl concentration) and determined the activity of a Cas9 variant codon optimized for zebrafish, that is, a Cas9 carrying an SV40 NLS at the N- and C-terminus (nls-zCas9-nls, hereinafter: zCas9, Plasmid #47929 Addgene, Jao et al., 2013). The analysis of medaka oca2 knock-out embryos injected with zCas9 revealed bi-allelic inactivation events of the oca2 gene, yet with a strong overall variability as apparent by patchy unpigmented domains in the eyes (median of mean gray values = 134.5 compared to uninjected controls, median = 0.4; Figure 1d). This patchy distribution of small, unpigmented areas indicated that bi-allelic targeting occurred only in few cells at later stages of development. To address whether different peptide domains (NLSs, Myc-tag, amino acid linkers) flanking the Cas9 enzyme enhance the targeting efficiency, we performed a permutation screen with Cas9 variants carrying these domains at different positions, which resulted in the identification of the ‘hei-tag’ (Figure 1—figure supplement 1). The hei-tag comprises a myc-tag connected via a flexible linker to an oNLS at the N-terminus complemented by a second oNLS fused to the C-terminus of a mammalian codon-optimized Cas9 (see Supplementary file 1 for sequence) and in this conformation displayed highest editing activity. Any alteration of those domains in relative order or sequence negatively impacted on editing efficiency compared to the hei-tag (Supplementary file 2).

When assessing the activity of the resulting heiCas9 at high molar excess (standard conditions, 150 ng/µl), heiCas9 displayed a 70% increase in bi-allelic targeting efficiency vs. the reference zCas9 (median zCas9 = 134.5, heiCas9 = 225.3; Figure 1d) in medaka. Embryos co-injected with heiCas9 mRNA and sgRNAs against oca2 essentially lost pigmentation. The observed absence of pigmentation argues for an early time point of action due to high activity and efficient nuclear translocation of the tagged heiCas9 variant already at the earliest cleavage stages. In developing organisms, the time point of genome editing essentially impacts on the allele variance, that is, the number of alleles established by the targeting attempt. To immediately provide a functional editing machinery, preassembled ribonucleoproteins (RNPs) containing Cas9 protein and guide RNA are popular, employing high molar excess/high concentrations of Cas9 (Kroll et al., 2021; Wu et al., 2018). Strikingly, the editing efficiency of injected heiCas9 mRNA was fully comparable to such RNP approaches (Figure 1d, Figure 1—figure supplement 2).

To address whether the enhancement by hei-tag fusion to Cas9 is applicable to different models, we next compared the activities of the zCas9 and heiCas9 in a second, evolutionarily distant fish species D. rerio (zebrafish) targeting the orthologous oca2 gene (sgRNAs DrOca2 T1, T2; Hammouda et al., 2019). Injected and control embryos were fixed well after the onset of pigmentation at 2.5 dpf (Kimmel et al., 1995; Figure 1e–f) and subjected to the quantitative assay for eye pigmentation described above. Taking the activity of zCas9 as base level (median = 14.7), heiCas9 delivered an outstanding targeting efficiency (median = 254.6), reflecting a 17-fold increase (p = 2.1e-56) (Figure 1g, Figure 1—figure supplement 2). Similar to the results in medaka, yet even more pronounced, nearly unpigmented embryos were obtained with the heiCas9, arguing for highly efficient, early targeting. Taken together, addition of the hei-tag to a mammalian codon-optimized Cas9 resulted in the highly efficient heiCas9, which boosted the targeting efficiency 17-fold, even when used at saturating concentrations. It prominently inactivated both alleles of the targeted oca2 locus, with a putatively early onset of action upon injection of heiCas9 mRNA and the respective sgRNAs at the one-cell stage.

To address whether the high targeting efficiency of heiCas9 was conveyed by the high molar excess employed or was possibly restricted to the oca2 locus, we turned to a multiplexing regime at 10-fold reduced concentrations of the Cas9 variants employed. We targeted four different genomic loci with four different sgRNAs: exonic targeting of oca2 (OlOca2 T2), targeting of the start codon of the retina-specific transcription factor 2 (rx2; Stemmer et al., 2015), and the crystallin alpha a (cryaa; Stemmer et al., 2015) as well as intronic targeting of rx3 (Zilova et al., 2021). Medaka one-cell stage embryos were co-injected with a mix of 12.5 ng/µl per sgRNA, the 10-fold reduced (15 ng/µl) zCas9 or heiCas9 mRNA and 20 ng/µl mCherry mRNA as injection tracer.

For each multiplexing experiment, the genomic DNA of three pools each containing eight randomly picked crispants was extracted at 4 dpf and subjected to allele-specific genotyping via Illumina sequencing. In the multiplexing approaches, a total of 823,898 reads for the zCas9 and 824,817 reads for the heiCas9, compared to 711,739 control reads, were analyzed (Supplementary file 3, Figure 2—figure supplement 1). In all cases, heiCas9 performed dramatically better than the reference zCas9 (Figure 2a; mean percentage of modified alleles zCas9 [black dots] vs. heiCas9 [red dots]: OlOca2: 3.38% vs. 54.59%, p = 0.026; OlRx2: 20.82% vs. 95.85%, p = 3.2e-06; OlRx3: 16.61% vs. 49.36%, p = 0.0041; OlCryaa: 83.50% vs. 98.44%, p = 0.039). Strikingly, although the overall targeting efficiency was consistently higher as reflected by the high percentage of edited alleles (Figure 2a), at the same time the allele variance was reduced in all cases when using heiCas9 (Figure 2b; mean percentage of allele variance: zCas9 [black hollow dots] vs. heiCas9 [red hollow dots]: OlOca2: 20.71% vs. 12.87%, p = 0.025; OlRx2: 15.63% vs. 7.86%, p = 7.6e-06; OlRx3: 17.91% vs. 12.75%, p = 0.00021; OlCryaa: 10.17% vs. 8.74%, p = 0.22). This reduced allele variance for all multiplexed loci indicates an early editing by heiCas9. Given this and the overall higher targeting efficiency in all loci analyzed in the multiplexing approach, heiCas9 outperformed zCas9. It resulted in a massive performance boost, which was partially masked at saturating conditions, and now became fully apparent. The high efficiency of heiCas9 thus allows efficient editing at low concentrations with the potential to reduce off-target effects. Whether this putative reduction of off-targets is (over-)compensated by the efficient nuclear localization needs to be assessed by whole-genome sequencing approaches in the future.

Figure 2 with 1 supplement see all

Download asset Open asset

While the early onset of action is required for uniform editing in developing organisms, cell culture approaches demand efficient translocation of the sgRNA/Cas9 complex in a large number of cells. To validate the range of action on the one hand and to address the relevance of the hei-tag in a mammalian setting, we expanded the scope of the analysis to mammalian cell culture. We focused on mRNA-based assays and compared the activity of heiCas9 to state-of-the-art Cas9 variants, that is, the commercially available GeneArt CRISPR nuclease as well as a mammalian codon-optimized Cas9 (JDS246-Cas9, Addgene #43861) in mouse SW10 cells. We assessed the respective genome editing efficiencies by independent and complementary tools, the Tracking of Indels by Decomposition (TIDE) analysis (Brinkman et al., 2014) as well as by Inference of CRISPR Editing (ICE) (Hsiau et al., 2018). Both approaches decompose the mixed Sanger reads of PCR products spanning the CRISPR target site and compute an efficiency score as well as the distribution of expected indels. To target the murine Periaxin (Prx) locus, mouse SW10 cells were co-transfected with MmPrx crRNA/ATTO-550-linked tracrRNA and the mRNAs of either JDS246-Cas9, GeneArt CRISPR nuclease, or heiCas9. The Prx locus was PCR amplified and sequenced. Similar to targeting in organismo, heiCas9 also exhibited the highest genome editing efficiency when compared to JDS246-Cas9 (TIDE: 123.6%, ICE: 113%) and GeneArt CRISPR nuclease (TIDE: 123.1%, ICE: 111%) in mammalian cell culture (Figure 3, Figure 3—figure supplement 1, R² > 0.9 (TIDE) and >0.9 (ICE) for all mRNAs tested). Notably, the KO-score efficiencies (ICE) amounted to 173% compared to JDS246-Cas9 and to 167% compared to GeneArt CRISPR nuclease, indicating higher abundance of frameshifts (Hsiau et al., 2018) at this genomic locus.

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Raw data for scatter plot shown in Figure 3.

https://cdn.elifesciences.org/articles/70558/elife-70558-fig3-data1-v3.xlsx

Download elife-70558-fig3-data1-v3.xlsx

Remarkably, heiCas9-transfected cells showed a highly increased number of mutant alleles with an increased abundance of a 26 nt deletion when compared to GeneArt CRISPR nuclease and JDS246-Cas9 (Figure 3—figure supplement 1).

Given the observed boosting of Cas9 activity by the simple addition of the hei-tag, we next tested if the hei-tag also improves further Cas9-based techniques. Base editing is an increasingly applied method with a potential for therapeutics (Antoniou et al., 2021). Base editors are composed of a modified Cas9 that only nicks one DNA strand and does not introduce a double-strand break (Cas9 nickase or Cas9n) and a nucleotide deaminase for precisely targeted nucleotide editing (Anzalone et al., 2020). To increase the efficiency of base editors, several iterative rounds of optimization of the employed deaminases and linkers have been undertaken, yielding optimal performance with the newest variants (Carrington et al., 2020; Cornean et al., 2022; Rosello et al., 2021; Zhao et al., 2020). To investigate if the addition of the hei-tag provides an easy and straightforward alternative route for increasing the activity of a nuclear protein of interest, we selected a C-to-T base editor version with intermediate efficiency (BE4-Gam Komor et al., 2017) to introduce non-sense or severe miss-sense mutations into the pigmentation gene oca2. We employed our tool ACEofBASEs (Cornean et al., 2022) to design and evaluate sgRNA target sites that introduce non-synonymous codon mutations and/or pre-mature STOP codons upon editing of the respective open reading frame (ORF). We compared three different sgRNAs (OlOca2 T1, T3, and T4) employing the original BE4-Gam and the hei-tag fused variant (heiBE4-Gam). In the oca2 ORF, the transition of cytosines 766, 922, and 997 to thymine all convert the respective codon to a pre-mature STOP (OlOca2 T3: C766T, leading to Q256*; OlOca2 T4: C922T, leading to Q308*; OlOca2 T1: C995-997T, leading to T332I and Q333*). Again, the loss of pigmentation was used as proxy for bi-allelic targeting efficiency following medaka one-cell stage injections with either one of the three sgRNAs (OlOca2 T1, T3, or T4, 30ng/µl) as well as 150ng/µl mRNA of either BE4-Gam or heiBE4-Gam. Screening and analysis was performed at 4.5 dpf as described above. For each sgRNA employed, heiBE4-Gam resulted in more pronounced loss of pigmentation in comparison to BE4-Gam (Figure 4a; control median = 0.0; medians BE4-Gam vs. heiBE4-Gam: OlOca2 T1, 0.6 vs. 28.0, p = 1.737e-20; OlOca2 T3, 0.0 vs. 0.8, p = 0.0471; OlOca2 T4, 93.8 vs. 170.1, p = 5.215e-12). Quantification of Sanger sequencing reads confirmed an increase of all C-to-T transitions at the OlOca2 T1 target site when heiBE4-Gam was used (74.1% ± 8.9% for heiBE4-Gam vs. 44.2% ± 6.8% for BE4-Gam; Figure 4—figure supplement 1, three replicates containing five randomly picked embryos each). In particular, the C997T transition introducing a pre-mature STOP codon was increased 1.7-fold (i.e. 68% in heiBE4-Gam vs. 41% in BE4-Gam) in case of heiBE4-Gam (Figure 4b and c).

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

Raw data for quantifications shown in Figure 4a.

https://cdn.elifesciences.org/articles/70558/elife-70558-fig4-data1-v3.csv

Download elife-70558-fig4-data1-v3.csv

In conclusion, using the hei-tag to extend the ORFs of a mammalian codon-optimized SpCas9 or a C-to-T base editor (BE4-Gam) severely enhanced the respective genome targeting efficiency.

While the use of the optimized NLS in the hei-tag explains the earlier and better performance of the hei-tagged versions of Cas9 and base editors in developing organisms, the impact of the specific topology of domains contained in the hei-tag remains elusive. It is speculated that the addition of certain peptide tags influences the efficacy and specificity of the fused protein of interest, due to their different isoelectric points and charge distributions (Zhang et al., 2014). Interestingly, our permutation screen demonstrated that although comprising the exact same peptides (for instance, compare MFO-Cas9-O [heiCas9] vs. OMF-Cas9-O and MSF-Cas9-S vs. SMF-Cas9-S in Figure 1—figure supplement 1), position of the particular tags relative to each other conveyed different genome editing efficiencies.

The hei-tag renders the resulting heiCas9 into a highly efficient endonuclease with broad applicability overcoming the limitations of current SpCas9 variants by dramatically increasing the efficiency of targeted genome editing in organismo, as demonstrated in two evolutionarily distant fish models, as well as in mouse cell culture. In those systems, heiCas9 leads to a high abundance of identical mutant alleles, important for testing specific hypotheses or introducing site-specific modifications by homology-directed repair (Gutierrez-Triana et al., 2018). Conversely, Cas9 variants without the hei-tag are better suited for targeted screening approaches since they introduce a large number of different mutant alleles. heiCas9 markedly increased the (bi-allelic) targeting rate alongside a decrease in allele variance, indicating a high targeting efficiency already at the earliest stages of development. Precedentially such early targeting in developing organisms was most of all reported using RNPs (Kroll et al., 2021; Wu et al., 2018), yet mRNA injection of heiCas9 is fully comparable to these protein approaches. The benefits of using mRNA over protein are apparent: new Cas9 variants can easily be generated and produced cost-efficiently by highly reproducible in vitro transcription, a standard method in molecular biology labs.

In light of the ever-expanding CRISPR tool kit, the addition of the hei-tag provides the means to boost current specialized and future variants, as the simple addition of the hei-tag sequence also potentiated the activity of a cytosine base editor, with heiBE4-Gam resulting in an overall increase of about 30% of C-to-T transition rates (Figure 4 and Figure 4—figure supplement 1). Taken together, the boosting activity of the hei-tag is neither limited by the species nor the approach, making it a powerful tweak to swiftly upgrade any specifically adapted Cas-based genome editing approach (Anzalone et al., 2020).

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Antoniou P
   2. Miccio A
   3. Brusson M

   (2021) Base and Prime Editing Technologies for Blood Disorders

   Frontiers in Genome Editing 3:618406.

   https://doi.org/10.3389/fgeed.2021.618406

   • PubMed
   • Google Scholar
2. 1. Anzalone AV
   2. Koblan LW
   3. Liu DR

   (2020) Genome editing with CRISPR-Cas nucleases, base editors, transposases and prime editors

   Nature Biotechnology 38:824–844.

   https://doi.org/10.1038/s41587-020-0561-9

   • PubMed
   • Google Scholar
3. 1. Brinkman EK
   2. Chen T
   3. Amendola M
   4. van Steensel B

   (2014) Easy quantitative assessment of genome editing by sequence trace decomposition

   Nucleic Acids Research 42:e168.

   https://doi.org/10.1093/nar/gku936

   • PubMed
   • Google Scholar
4. 1. Carrington B
   2. Weinstein RN
   3. Sood R

   (2020) BE4max and AncBE4max Are Efficient in Germline Conversion of C:G to T:A Base Pairs in Zebrafish

   Cells 9:1690.

   https://doi.org/10.3390/cells9071690

   • PubMed
   • Google Scholar
5. 1. Clement K
   2. Rees H
   3. Canver MC
   4. Gehrke JM
   5. Farouni R
   6. Hsu JY
   7. Cole MA
   8. Liu DR
   9. Joung JK
   10. Bauer DE
   11. Pinello L

   (2019) CRISPResso2 provides accurate and rapid genome editing sequence analysis

   Nature Biotechnology 37:224–226.

   https://doi.org/10.1038/s41587-019-0032-3

   • PubMed
   • Google Scholar
6. 1. Cong L
   2. Ran FA
   3. Cox D
   4. Lin S
   5. Barretto R
   6. Habib N
   7. Hsu PD
   8. Wu X
   9. Jiang W
   10. Marraffini LA
   11. Zhang F

   (2013) Multiplex genome engineering using CRISPR/Cas systems

   Science (New York, N.Y.) 339:819–823.

   https://doi.org/10.1126/science.1231143

   • PubMed
   • Google Scholar
7. 1. Cornean A
   2. Gierten J
   3. Welz B
   4. Mateo JL
   5. Thumberger T
   6. Wittbrodt J

   (2022) Precise in vivo functional analysis of DNA variants with base editing using ACEofBASEs target prediction

   eLife 11:e72124.

   https://doi.org/10.7554/eLife.72124

   • PubMed
   • Google Scholar
8. 1. Dingwall C
   2. Robbins J
   3. Dilworth SM
   4. Roberts B
   5. Richardson WD

   (1988) The nucleoplasmin nuclear location sequence is larger and more complex than that of SV-40 large T antigen

   The Journal of Cell Biology 107:841–849.

   https://doi.org/10.1083/jcb.107.3.841

   • PubMed
   • Google Scholar
9. 1. Furutani-Seiki M
   2. Wittbrodt J

   (2004) Medaka and zebrafish, an evolutionary twin study

   Mechanisms of Development 121:629–637.

   https://doi.org/10.1016/j.mod.2004.05.010

   • PubMed
   • Google Scholar
10. 1. Gutierrez-Triana JA
    2. Tavhelidse T
    3. Thumberger T
    4. Thomas I
    5. Wittbrodt B
    6. Kellner T
    7. Anlas K
    8. Tsingos E
    9. Wittbrodt J

    (2018) Efficient single-copy HDR by 5’ modified long dsDNA donors

    eLife 7:e39468.

    https://doi.org/10.7554/eLife.39468

    • PubMed
    • Google Scholar
11. 1. Hammouda OT
    2. Böttger F
    3. Wittbrodt J
    4. Thumberger T

    (2019) Swift Large-scale Examination of Directed Genome Editing

    PLOS ONE 14:e0213317.

    https://doi.org/10.1371/journal.pone.0213317

    • PubMed
    • Google Scholar
12. Preprint

    1. Hsiau T
    2. Conant D
    3. Rossi N
    4. Maures T
    5. Waite K
    6. Yang J
    7. Joshi S
    8. Kelso R
    9. Holden K
    10. Enzmann BL
    11. Stoner R

    (2018) Inference of CRISPR Edits from Sanger Trace Data

    bioRxiv.

    https://doi.org/10.1101/251082

    • Google Scholar
13. 1. Hwang WY
    2. Fu Y
    3. Reyon D
    4. Maeder ML
    5. Tsai SQ
    6. Sander JD
    7. Peterson RT
    8. Yeh JRJ
    9. Joung JK

    (2013) Efficient In Vivo Genome Editing Using RNA-Guided Nucleases

    Nature Biotechnology 31:227–229.

    https://doi.org/10.1038/nbt.2501

    • Google Scholar
14. 1. Inoue T
    2. Iida A
    3. Maegawa S
    4. Sehara-Fujisawa A
    5. Kinoshita M

    (2016) Generation of a transgenic medaka (Oryzias latipes) strain for visualization of nuclear dynamics in early developmental stages

    Development, Growth & Differentiation 58:679–687.

    https://doi.org/10.1111/dgd.12324

    • PubMed
    • Google Scholar
15. 1. Iwamatsu T

    (2004) Stages of normal development in the medaka Oryzias latipes

    Mechanisms of Development 121:605–618.

    https://doi.org/10.1016/j.mod.2004.03.012

    • PubMed
    • Google Scholar
16. 1. Jao LE
    2. Wente SR
    3. Chen W

    (2013) Efficient multiplex biallelic zebrafish genome editing using a CRISPR nuclease system

    PNAS 110:13904–13909.

    https://doi.org/10.1073/pnas.1308335110

    • PubMed
    • Google Scholar
17. 1. Kalderon D
    2. Roberts BL
    3. Richardson WD
    4. Smith AE

    (1984) A short amino acid sequence able to specify nuclear location

    Cell 39:499–509.

    https://doi.org/10.1016/0092-8674(84)90457-4

    • PubMed
    • Google Scholar
18. 1. Kimmel CB
    2. Ballard WW
    3. Kimmel SR
    4. Ullmann B
    5. Schilling TF

    (1995) Stages of embryonic development of the zebrafish

    Developmental Dynamics 203:253–310.

    https://doi.org/10.1002/aja.1002030302

    • PubMed
    • Google Scholar
19. 1. Kirchmaier S
    2. Lust K
    3. Wittbrodt J

    (2013) Golden GATEway cloning--A combinatorial approach to generate fusion and recombination constructs

    PLOS ONE 8:e76117.

    https://doi.org/10.1371/journal.pone.0076117

    • PubMed
    • Google Scholar
20. 1. Kluesner MG
    2. Nedveck DA
    3. Lahr WS
    4. Garbe JR
    5. Abrahante JE
    6. Webber BR
    7. Moriarity BS

    (2018) EditR: A Method to Quantify Base Editing from Sanger Sequencing

    The CRISPR Journal 1:239–250.

    https://doi.org/10.1089/crispr.2018.0014

    • PubMed
    • Google Scholar
21. 1. Komor AC
    2. Zhao KT
    3. Packer MS
    4. Gaudelli NM
    5. Waterbury AL
    6. Koblan LW
    7. Kim YB
    8. Badran AH
    9. Liu DR

    (2017) Improved base excision repair inhibition and bacteriophage Mu Gam protein yields C:G-to-T:A base editors with higher efficiency and product purity

    Science Advances 3:eaao4774.

    https://doi.org/10.1126/sciadv.aao4774

    • PubMed
    • Google Scholar
22. 1. Koster R
    2. Stick R
    3. Loosli F
    4. Wittbrodt J

    (1997) Medaka spalt acts as a target gene of hedgehog signaling

    Development (Cambridge, England) 124:3147–3156.

    https://doi.org/10.1242/dev.124.16.3147

    • PubMed
    • Google Scholar
23. 1. Kroll F
    2. Powell GT
    3. Ghosh M
    4. Gestri G
    5. Antinucci P
    6. Hearn TJ
    7. Tunbak H
    8. Lim S
    9. Dennis HW
    10. Fernandez JM
    11. Whitmore D
    12. Dreosti E
    13. Wilson SW
    14. Hoffman EJ
    15. Rihel J

    (2021) A simple and effective F0 knockout method for rapid screening of behaviour and other complex phenotypes

    eLife 10:e59683.

    https://doi.org/10.7554/eLife.59683

    • PubMed
    • Google Scholar
24. 1. Lischik CQ
    2. Adelmann L
    3. Wittbrodt J

    (2019) Enhanced in vivo-imaging in medaka by optimized anaesthesia, fluorescent protein selection and removal of pigmentation

    PLOS ONE 14:e0212956.

    https://doi.org/10.1371/journal.pone.0212956

    • PubMed
    • Google Scholar
25. 1. Liu P
    2. Liang S-Q
    3. Zheng C
    4. Mintzer E
    5. Zhao YG
    6. Ponnienselvan K
    7. Mir A
    8. Sontheimer EJ
    9. Gao G
    10. Flotte TR
    11. Wolfe SA
    12. Xue W

    (2021) Improved prime editors enable pathogenic allele correction and cancer modelling in adult mice

    Nature Communications 12:2121.

    https://doi.org/10.1038/s41467-021-22295-w

    • PubMed
    • Google Scholar
26. 1. Nidhi S
    2. Anand U
    3. Oleksak P
    4. Tripathi P
    5. Lal JA
    6. Thomas G
    7. Kuca K
    8. Tripathi V

    (2021) Novel CRISPR–Cas Systems: An Updated Review of the Current Achievements Applications, and Future Research Perspectives

    International Journal of Molecular Sciences 22:3327.

    https://doi.org/10.3390/ijms22073327

    • Google Scholar
27. 1. Poon IKH
    2. Jans DA

    (2005) Regulation of nuclear transport: central role in development and transformation?

    Traffic (Copenhagen, Denmark) 6:173–186.

    https://doi.org/10.1111/j.1600-0854.2005.00268.x

    • PubMed
    • Google Scholar
28. 1. Rosello M
    2. Vougny J
    3. Czarny F
    4. Mione MC
    5. Concordet J-P
    6. Albadri S
    7. Del Bene F

    (2021) Precise base editing for the in vivo study of developmental signaling and human pathologies in zebrafish

    eLife 10:e65552.

    https://doi.org/10.7554/eLife.65552

    • PubMed
    • Google Scholar
29. 1. Rupp RA
    2. Snider L
    3. Weintraub H

    (1994) Xenopus embryos regulate the nuclear localization of XMyoD

    Genes & Development 8:1311–1323.

    https://doi.org/10.1101/gad.8.11.1311

    • PubMed
    • Google Scholar
30. 1. Schindelin J
    2. Arganda-Carreras I
    3. Frise E
    4. Kaynig V
    5. Longair M
    6. Pietzsch T
    7. Preibisch S
    8. Rueden C
    9. Saalfeld S
    10. Schmid B
    11. Tinevez J-Y
    12. White DJ
    13. Hartenstein V
    14. Eliceiri K
    15. Tomancak P
    16. Cardona A

    (2012) Fiji: an open-source platform for biological-image analysis

    Nature Methods 9:676–682.

    https://doi.org/10.1038/nmeth.2019

    • PubMed
    • Google Scholar
31. 1. Stemmer M
    2. Thumberger T
    3. Del Sol Keyer M
    4. Wittbrodt J
    5. Mateo JL

    (2015) CCTop: An Intuitive, Flexible and Reliable CRISPR/Cas9 Target Prediction Tool

    PLOS ONE 10:e0124633.

    https://doi.org/10.1371/journal.pone.0124633

    • PubMed
    • Google Scholar
32. Software

    1. Team R

    (2020)

    RStudio: Integrated Development Environment for R

    RStudio.
33. Book

    1. Westerfield M

    (2000)

    The Zebrafish Book. A Guide for the Laboratory Use of Zebrafish (Danio Rerio

    University of Oregon Press.

    • Google Scholar
34. Book

    1. Wickham H

    (2016) Ggplot2

    springer.

    https://doi.org/10.1007/978-3-319-24277-4

    • Google Scholar
35. 1. Wu RS
    2. Lam II
    3. Clay H
    4. Duong DN
    5. Deo RC
    6. Coughlin SR

    (2018) A Rapid Method for Directed Gene Knockout for Screening in G0 Zebrafish

    Developmental Cell 46:112–125.

    https://doi.org/10.1016/j.devcel.2018.06.003

    • PubMed
    • Google Scholar
36. 1. Zhang JH
    2. Pandey M
    3. Kahler JF
    4. Loshakov A
    5. Harris B
    6. Dagur PK
    7. Mo YY
    8. Simonds WF

    (2014) Improving the specificity and efficacy of CRISPR/CAS9 and gRNA through target specific DNA reporter

    Journal of Biotechnology 189:1–8.

    https://doi.org/10.1016/j.jbiotec.2014.08.033

    • PubMed
    • Google Scholar
37. 1. Zhao Y
    2. Shang D
    3. Ying R
    4. Cheng H
    5. Zhou R

    (2020) An optimized base editor with efficient C-to-T base editing in zebrafish

    BMC Biology 18:190.

    https://doi.org/10.1186/s12915-020-00923-z

    • PubMed
    • Google Scholar
38. 1. Zilova L
    2. Weinhardt V
    3. Tavhelidse T
    4. Schlagheck C
    5. Thumberger T
    6. Wittbrodt J

    (2021) Fish primary embryonic pluripotent cells assemble into retinal tissue mirroring in vivo early eye development

    eLife 10:e66998.

    https://doi.org/10.7554/eLife.66998

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "hei-tag: a highly efficient tag to boost targeted genome editing" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Zacharias Kontarakis as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Didier Stainier as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Benchmark hei-tag activity against the state-of-art genome engineering tool in each experiment (see reviewer comments – also comparison to RNP use in fish is highly desirable).

2) Provide heiCas9 editing data across multiple targets/gRNAs, ideally showing diverse characteristics (e.g. genomic location, sequence, activity).

3) Edit manuscript to clearly describe the strengths and weaknesses of the "hei-tagging" method.

Reviewer #1 (Recommendations for the authors):

I have a few concerns regarding the way the authors structure and present their claims:

1) One of the main take home messages (based on the Abstract) is that the hei-tag boosts the activity of a "wide variety" of genome editing tools. Even though the BE data are promising, the manuscript could use more examples (e.g. prime editors, that have been used in zebrafish – Petri et al., 2021). Can the hei-tag make editing possible in cases where it is undetectable with the state of the art tools?

2) Throughout the manuscript, the authors use only a handful of guides. Adding more examples, especially using guides that show a range of efficiencies with the non-hei-tag Cas9 would provide a better description of the strengths and limitations of the hei-tag.

3) What is the rationale of using the Cas9 from Hwang et al., and not the zebrafish codon optimised Cas9 from Jao et al?

4) Why do the authors use the GeneArt nuclease as a "state of the art" for the cell culture? What other tags are included in this nuclease? We can see that the worse score in Fig1d comes from JDS246. Not having the hei-tag is only one difference between JDS246 and heiCas9. One other is that heiCas9 lacks the FLAG (Myc-Cas performs already pretty well, however zebrafish data are not shown). How does GeneArt Cas9 compare in that respect? Such information should be presented to readers.

5) In Figure 1, what do the single guide results look like? Boosting deletion outcomes using two guides is not the same as boosting activity of single guide editing. Is the activity of both guides improved? Are they just being brought at the same level? Did the authors use sequencing to analyse what happened at the DNA level?

6) The hei-tag could be affecting editing outcomes, rather than ON target activity. In Figure S1, it is clear that the frequency of the in frame -15 allele is reduced, while overall efficiency in not dramatically affected. Changing the editing profile by using different nucleases (variants) is not the same as boosting efficiency in KO. The in-frame indels could be favored in other sites, thus reducing KO-score (if out of frame are regarded as desired outcome).

7) Generally, it is advised to use NGS data for strong claims about editing efficiency.

8) If ON-target activity is boosted, increased OFF-target activity is of concern, especially in cell culture when editing accuracy and precision are key. Since the main focus of the authors is not mammalian cell culture or therapeutic applications, they should at least be clear about this limitation.

Generally, the manuscript would be strengthened by including more examples of editing tools (desirable, but not absolutely required), include analyses of more guides (required), add sequencing data wherever possible (highly advised), adequately discussing strengths and limitations of the hei-tag (required), presenting some NGS data (recommended), toning down some "generalised" claims (unless more supporting data are provided).

Reviewer #2 (Recommendations for the authors):

Several reports in zebrafish have shown that Cas9/gRNA complexes injections can reach gene editing up to 100% in F0 embryos including when targeting multiple loci at the same time (See for instance Wu et al., 2018 and Kroll et al., 2021). In light of these impressive results the utility of the present method for the fish community is overstated at best without mentioning these points anywhere in the discussion.

The authors should attempt to compare the injection of heiCAS9 protein as this is the current preferred and most efficient method in the fish community.

In the present manuscript is not clear what is the purpose of the Myc tag in the construct. This is not used at all to visualize protein localization or purification. Do the authors believe that the addition of Myc increases Cas9 activity? If not giving that this is a method paper they should demonstrate the utility in this specific contest.

The zebrafish experiment lacks the comparison of the heicas9 with the most efficient myc-Cas9 leading to an overestimation to the improved efficiency of the construct.

Finally methodological papers based on a single locus are difficult to appreciate as they may be influenced by "the gambler's luck" (i.e. the chosen locus cold be a fortunate pick). These results should be extended to other genes as it's standard in the field (see the two publications above).

Finally the base editor experiment are equally strongly biased. The comparison with BE4-Gam is not representative of the current state of the art were several reports using ancBE4Max (See for instance Carrington et al., 2020, Zao et al., 2020 and Rosello et al., 2021) show highly improved results. In lights of these papers the statement "Notably, in heiBE4-Gam injections, for each of the three cytosines in the base editing window, the C-to-T transition rate was higher than 60%, a level never observed in BE4-Gam injected embryos" is not true as similar or higher level of C to T conversion have been reported. Again this a comparison with ancBE4Max should be extended to multiple loci.

Reviewer #3 (Recommendations for the authors):

1. In the abstract, the authors introduced factors that leave room for the improvement of gene editing efficiency in CRISPR/Cas9 tools: (1) nuclear localization signal -citing Cong et al., 2013; and (2) protein tags for "immediate detection or straight-forward purification" and linkers to "avoid steric hinderance impacting on activity" -citing Zhang et al., 2014. However, such conclusions were never made in either of the originally cited papers. Cong et al., did not compare the editing activity with and without a NLS signal. On the contrary, there are partial evidence indicating that Cas9 protein may not require an NLS to assist import into the nucleus (Hu et al., G3, 2018). In Zhang et al., it was suspected that the addition of a flag or myc tag changed the charge distribution of the Cas9 protein, thus increasing its specificity and efficacy. No statement was made about the purification or the linker. It is OK to introduce the relevant background, but I find it problematic to misinterpret the cited literature to show conclusions they did not make.

2. The authors used modified version of NLS in the hei-tag construct to facilitate early nuclear targeting, while a straight-forward way to make nuclear-targeted Cas9 available in the cell is to directly inject the (nuclear-targeting) Cas9 protein. The authors should either provide this control experiment, or clarify why they only chose to build RNA-based systems.

3. It is widely known that the gRNA design is a critical factor affecting the gene editing efficiency. While hei-tag shows an increased bi-allelic editing efficiency than the control constructs, it is not clear whether this boost is universal with different targeted genes and different sgRNA designs.

4. JDS246-Cas9 was chosen as the baseline construct to evaluate the boost of editing efficiency. Given this was a construct originally made for mammalian cells, it is not clear whether it represents the state-of-art editing techniques, especially in zebrafish. While it is unrealistic to test all the available tools, other systems have been reported with high bi-allelic editing efficiency specifically in zebrafish should be introduced as a control (e.g. Jao et al., 2013, PNAS -disclaimer: the reviewer was not a maker of this tool).

5. In the base-editing experiments, the injected fish showed various level of eye-pigmentation colors, in contrast to the knockout experiment where cells devoid of pigment appear in patches on the eye. The authors should provide an explanation of why this is the case and justify why the pigmentation level has to be quantified differently in Figure 3 than Figure 1.

Essential revisions:

1) Benchmark hei-tag activity against the state-of-art genome engineering tool in each experiment (see reviewer comments – also comparison to RNP use in fish is highly desirable).

In the revised version of the manuscript we now extensively compare the hei-tagged Cas9 (heiCas9) against state-of-the-art Cas variants, including RNPs.

Strikingly, heiCas9 clearly outperforms the state-of-the-art Cas variants and RNPs. Only when applied at highest concentrations (highly viscous injection mix) RNPs show a slightly increased activity. Following the reviewer’s suggestion, we now use the zebrafish codon optimized nls-zCas9-nls from Jao et al., 2013 as reference in both, medaka and zebrafish. heiCas9 mRNA in both cases massively outperforms the reference. This is now presented in the new Figure 1.

Given that all those comparisons were performed at saturating conditions in the plateau of Cas activity, we subsequently diluted the concentrations of the respective Cas variants by a factor of 10 and compared their activity in the exponential phase for four different target loci in a multiplexing approach. Here, the massive performance boost provided by the hei-tag becomes fully apparent, which was partially masked at saturating conditions.

We quantified the results by Illumina sequencing of the respective multiplexed loci and show a high specific activity of the heiCas9 in contrast to the zCas9. Strikingly, the allele variance per locus is clearly reduced in the heiCas9 experiment (even under multiplexing conditions) in comparison to the state-of-the-art Cas9 variant. This argues for an early (in development) and precise activity of heiCas9.

Those findings are now presented in the all new Figure 2 in the revised version of the manuscript.

2) Provide heiCas9 editing data across multiple targets/gRNAs, ideally showing diverse characteristics (e.g. genomic location, sequence, activity).

To address this point, we have been comparing the activity of the state-of-the-art Cas9 variant (zCas9) with heiCas9 on four different genomic loci (oca2, rx2, rx3, cryaa) employing sgRNAs with different activities in a multiplexing approach.

Editing efficiency on the respective loci in this multiplexing experiment was quantified by Illumina sequencing.

This analysis revealed a different editing efficiency for the different loci of both, the nls-zCas9-nls (Jao et al., 2013) as well as the heiCas9. In all cases, however, heiCas9 performed better than the state-of-the-art Cas9 variant. Interestingly, besides a massive increase in editing events, heiCas9 editing resulted in a clear drop of the allele variance. This indicates that heiCas9 edited more efficiently and acted earlier in development (than the zCas9) resulting in a lower number of different alleles. This is now presented in a new Figure 2 in the revised version of the manuscript.

3) Edit manuscript to clearly describe the strengths and weaknesses of the "hei-tagging" method.

We followed this advice and have included a paragraph in the discussion to clearly describe the strengths and weaknesses of “hei-tagging”.

Reviewer #1 (Recommendations for the authors):

I have a few concerns regarding the way the authors structure and present their claims:

1) One of the main take home messages (based on the Abstract) is that the hei-tag boosts the activity of a "wide variety" of genome editing tools. Even though the BE data are promising, the manuscript could use more examples (e.g. prime editors, that have been used in zebrafish – Petri et al., 2021). Can the hei-tag make editing possible in cases where it is undetectable with the state of the art tools?

We toned down this statement in the abstract. In our new Figure 2 we now show multiplexed editing by using sgRNAs of different efficiencies. In all cases, the heiCas9 performed better (more overall editing, fewer different alleles) compared to the reference zCas9 (Jao et al., 2013).

2) Throughout the manuscript, the authors use only a handful of guides. Adding more examples, especially using guides that show a range of efficiencies with the non-hei-tag Cas9 would provide a better description of the strengths and limitations of the hei-tag.

We followed the referee’s suggestion and have chosen additional sgRNAs of which one targets an intron (rx3) and three sgRNAs with different editing activity targeting exonic sequences (oca2, rx2 and cryaa). The respective Illumina sequencing analysis is now included as new main Figure 2 and Supplementary file 3.

3) What is the rationale of using the Cas9 from Hwang et al., and not the zebrafish codon optimised Cas9 from Jao et al?

We thank the referee for the suggestion and now use the zCas9 from Jao et al., as reference in our oca2 analysis pipeline (Figure 1d, g) for which we performed an entirely new set of experiments (transcription, injection and analysis) in medaka and zebrafish. We also use the zCas9 from Jao et al., 2013 as reference in the multiplexing approach in new Figure 2.

4) Why do the authors use the GeneArt nuclease as a "state of the art" for the cell culture? What other tags are included in this nuclease? We can see that the worse score in Fig1d comes from JDS246. Not having the hei-tag is only one difference between JDS246 and heiCas9. One other is that heiCas9 lacks the FLAG (Myc-Cas performs already pretty well, however zebrafish data are not shown). How does GeneArt Cas9 compare in that respect? Such information should be presented to readers.

For cell culture assays, there is only a limited number of commercially available Cas9 nuclease mRNAs to choose from. The exact sequence composition of tags/NLSs of the GeneArt Cas9 mRNA is not further specified by the manufacturer.

5) In Figure 1, what do the single guide results look like? Boosting deletion outcomes using two guides is not the same as boosting activity of single guide editing. Is the activity of both guides improved? Are they just being brought at the same level? Did the authors use sequencing to analyse what happened at the DNA level?

We now analyze the edited alleles with Illumina sequencing (new Figure 2). Since in this approach we followed other suggestions by the referee, we multiplexed different sgRNAs. Therefore, we decided to pick only one sgRNA for oca2 (T2). We can show that targeting efficiency using heiCas9 clearly improves over the zCas9 from Jao et al., 2013 not only in oca2 but all targeted loci.

6) The hei-tag could be affecting editing outcomes, rather than ON target activity. In Figure S1, it is clear that the frequency of the in frame -15 allele is reduced, while overall efficiency in not dramatically affected. Changing the editing profile by using different nucleases (variants) is not the same as boosting efficiency in KO. The in-frame indels could be favored in other sites, thus reducing KO-score (if out of frame are regarded as desired outcome).

We agree with the referee that the increased out-of frame repair seen with the heiCas9 may be locus specific. Nevertheless, the induction of indels using the heiCas9 was dramatically higher compared to the other nucleases employed.

7) Generally, it is advised to use NGS data for strong claims about editing efficiency.

We thank the referee and include Illumina sequencing data now in new Figure 2 and Supplementary file 3.

8) If ON-target activity is boosted, increased OFF-target activity is of concern, especially in cell culture when editing accuracy and precision are key. Since the main focus of the authors is not mammalian cell culture or therapeutic applications, they should at least be clear about this limitation.

We discuss this important point now in the main text where we clearly state this potential limitation.

Reviewer #2 (Recommendations for the authors):

Several reports in zebrafish have shown that Cas9/gRNA complexes injections can reach gene editing up to 100% in F0 embryos including when targeting multiple loci at the same time (See for instance Wu et al., 2018 and Kroll et al., 2021). In light of these impressive results the utility of the present method for the fish community is overstated at best without mentioning these points anywhere in the discussion.

The authors should attempt to compare the injection of heiCAS9 protein as this is the current preferred and most efficient method in the fish community.

We want to kindly point out that especially in the medaka community, protein injection of effectors of most kinds is not the common protocol. In case of targeted genome editing, also Cas9 and BaseEditors are usually provided as mRNA.

Following the referees suggestion, we have now compared the most recent variant of the IDT Cas9 protein (Alt-R S.p.Cas9 V3) in similar concentrations as used by Wu et al., 2018, Dev Cell (5µM) and Kroll et al., 2021, eLife (24µM) and equimolar to the mRNA molecule concentration we use (1.765µM) to heiCas9 mRNA injections (new Figure 1d). This clearly demonstrates that in medaka, mRNA injection with the heiCas9 variant is equally effective compared to protein injections.

The benefits of using mRNA over protein injections are: cost-efficiency, flexibility in generating variants, easier handling of the injection mix (24µM and 5µM protein concentrations are very viscous, often leading to needle clogging and hence affecting injection throughput and success).

In the present manuscript is not clear what is the purpose of the Myc tag in the construct. This is not used at all to visualize protein localization or purification. Do the authors believe that the addition of Myc increases Cas9 activity? If not giving that this is a method paper they should demonstrate the utility in this specific contest.

The Myc-tag was initially included to facilitate straight forward protein purification. Strikingly, the injection of mRNA of this construct already dramatically increased knock-out efficiencies eliminating the need for protein purification.

We performed a permutation screen of NLSs, Myc-tag and linkers and addressed the impact on knock-out efficiency. This resulted in the identification of the hei-tag (myc, flexilinker, oNLS) described here. To provide this information to the community, we now include this screen in Figure 1—figure supplement 1.

The zebrafish experiment lacks the comparison of the heicas9 with the most efficient myc-Cas9 leading to an overestimation to the improved efficiency of the construct.

By suggestion of referee #1 we now have changed the reference Cas9 to “the zebrafish codon optimised Cas9 from Jao et al., 2013” in the oca2 knock-out screen and have adapted the medaka injections as well. Interestingly in zebrafish, zCas9 was not significantly better than the earliest Cas9 version (“JDS246”) – see updated Figure 1. In medaka however, the zCas9 performed well but clearly less efficient than the heiCas9.

Finally methodological papers based on a single locus are difficult to appreciate as they may be influenced by "the gambler's luck" (i.e. the chosen locus cold be a fortunate pick). These results should be extended to other genes as it's standard in the field (see the two publications above).

We thank the referee for the point raised here and now included 3 more sgRNAs with different activity and position (intronic/exonic). In all cases, heiCas9 not only performed more efficiently compared to the zCas9 from Jao et al., but also generated fewer different alleles over all modified alleles detected in the Illumina sequencing (see new Figure 2 and Supplementary file 3).

Finally the base editor experiment are equally strongly biased. The comparison with BE4-Gam is not representative of the current state of the art were several reports using ancBE4Max (See for instance Carrington et al., 2020, Zao et al., 2020 and Rosello et al., 2021) show highly improved results. In lights of these papers the statement "Notably, in heiBE4-Gam injections, for each of the three cytosines in the base editing window, the C-to-T transition rate was higher than 60%, a level never observed in BE4-Gam injected embryos" is not true as similar or higher level of C to T conversion have been reported. Again this a comparison with ancBE4Max should be extended to multiple loci.

The referee is right and we changed the indicated sentence. We feel that we need to clarify our rationale here (as we did in the amended main text): it was not our intention to generate “the best editor”. The activity of most recent editors clearly plateaus at the standard concentrations. Our intention rather was to show that a little tag can boost activity, which is easier detected in the exponential phase. We therefore used a version of the BaseEditor that has not reached its activity plateau to address this point. We now clearly state this in the revised version of the manuscript.

Reviewer #3 (Recommendations for the authors):

1. In the abstract, the authors introduced factors that leave room for the improvement of gene editing efficiency in CRISPR/Cas9 tools: (1) nuclear localization signal -citing Cong et al., 2013; and (2) protein tags for "immediate detection or straight-forward purification" and linkers to "avoid steric hinderance impacting on activity" -citing Zhang et al., 2014. However, such conclusions were never made in either of the originally cited papers. Cong et al., did not compare the editing activity with and without a NLS signal. On the contrary, there are partial evidence indicating that Cas9 protein may not require an NLS to assist import into the nucleus (Hu et al., G3, 2018). In Zhang et al., it was suspected that the addition of a flag or myc tag changed the charge distribution of the Cas9 protein, thus increasing its specificity and efficacy. No statement was made about the purification or the linker. It is OK to introduce the relevant background, but I find it problematic to misinterpret the cited literature to show conclusions they did not make.

We thank the referee for the comment and have carefully revised the manuscript to avoid the problems indicated.

2. The authors used modified version of NLS in the hei-tag construct to facilitate early nuclear targeting, while a straight-forward way to make nuclear-targeted Cas9 available in the cell is to directly inject the (nuclear-targeting) Cas9 protein. The authors should either provide this control experiment, or clarify why they only chose to build RNA-based systems.

We now have included a comparison to protein/sgRNA complex injections in Figure 1. Following the referees suggestion, we have now compared the most recent variant of the IDT Cas9 protein (Alt-R S.p.Cas9 V3) in similar concentrations as used by Wu et al., 2018, Dev Cell (5µM) and Kroll et al., 2021, eLife (24µM) and equimolar to the mRNA molecule concentration we use (1.765µM) to heiCas9 mRNA injections (new figure 1d). This clearly demonstrates that in medaka, mRNA injection with the heiCas9 variant is equally effective compared to protein injections.

The benefits of using mRNA over protein injections are: cost-efficiency, flexibility in generating variants, easier handling of the injection mix (24µM and 5µM protein concentrations are very viscous, often leading to needle clogging and hence affecting injection throughput).

3. It is widely known that the gRNA design is a critical factor affecting the gene editing efficiency. While hei-tag shows an increased bi-allelic editing efficiency than the control constructs, it is not clear whether this boost is universal with different targeted genes and different sgRNA designs.

We followed the referee’s suggestion and have chosen additional sgRNAs of which one targets an intron (rx3) and three sgRNAs with different editing activity targeting exonic sequences (oca2, rx2 and cryaa). The respective Illumina sequencing analysis is now included as new main figure 2 (and Supplementary file 3). In all cases, heiCas9 outperformed the reference.

4. JDS246-Cas9 was chosen as the baseline construct to evaluate the boost of editing efficiency. Given this was a construct originally made for mammalian cells, it is not clear whether it represents the state-of-art editing techniques, especially in zebrafish. While it is unrealistic to test all the available tools, other systems have been reported with high bi-allelic editing efficiency specifically in zebrafish should be introduced as a control (e.g. Jao et al., 2013, PNAS -disclaimer: the reviewer was not a maker of this tool).

We now have included the comparison to “the zebrafish codon optimised Cas9 from Jao et al.” and have adapted the medaka injections as well (Figure 1). Interestingly, it was not significantly better than the earliest mammalian optimized Cas9 version (“JDS246”) – see updated Figure 1. In medaka however, the zCas9 performed very well but clearly less efficient than the heiCas9. The Cas9 we use in our constructs is mammalian codon optimized, and stellar also in zebrafish.

5. In the base-editing experiments, the injected fish showed various level of eye-pigmentation colors, in contrast to the knockout experiment where cells devoid of pigment appear in patches on the eye. The authors should provide an explanation of why this is the case and justify why the pigmentation level has to be quantified differently in Figure 3 than Figure 1.

In both experiments, the loss of pigmentation phenotype is identical – loss of pigmentation appears in patches. We repeated the BaseEditor experiments and quantified according to the procedure demonstrated in Figure 1.

Author details

1. Thomas Thumberger

   Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Software, Validation, Visualization, Writing – original draft, Writing – review and editing

   Contributed equally with

   Tinatini Tavhelidse-Suck and Jose Arturo Gutierrez-Triana

   Competing interests

   patent application pending (EP21166099.8) related to the findings described

   "This ORCID iD identifies the author of this article:" 0000-0001-8485-457X

2. Tinatini Tavhelidse-Suck

   1. Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany
   2. Heidelberg Biosciences International Graduate School (HBIGS), Heidelberg, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Validation, Writing – original draft, Writing – review and editing

   Contributed equally with

   Thomas Thumberger and Jose Arturo Gutierrez-Triana

   Competing interests

   patent application pending (EP21166099.8) related to the findings described

   "This ORCID iD identifies the author of this article:" 0000-0002-6103-9019

3. Jose Arturo Gutierrez-Triana

   Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany

   Present address

   Escuela de Microbiología, Facultad de Salud, Universidad Industrial de Santander, Bucaramanga, Colombia

   Contribution

   Conceptualization, Formal analysis, Investigation, Validation, Writing – original draft, Writing – review and editing

   Contributed equally with

   Thomas Thumberger and Tinatini Tavhelidse-Suck

   Competing interests

   patent application pending (EP21166099.8) related to the findings described

4. Alex Cornean

   1. Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany
   2. Heidelberg Biosciences International Graduate School (HBIGS), Heidelberg, Germany

   Contribution

   Data curation, Formal analysis, Investigation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3727-7057

5. Rebekka Medert

   1. Heidelberg Biosciences International Graduate School (HBIGS), Heidelberg, Germany
   2. Institute of Pharmacology, Heidelberg University, Heidelberg, Germany
   3. DZHK (German Centre for Cardiovascular Research), partner site Heidelberg/Mannheim, Heidelberg, Germany

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

6. Bettina Welz

   1. Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany
   2. Heidelberg Biosciences International Graduate School (HBIGS), Heidelberg, Germany
   3. DZHK (German Centre for Cardiovascular Research), partner site Heidelberg/Mannheim, Heidelberg, Germany

   Contribution

   Investigation

   Competing interests

   No competing interests declared

7. Marc Freichel

   1. Institute of Pharmacology, Heidelberg University, Heidelberg, Germany
   2. DZHK (German Centre for Cardiovascular Research), partner site Heidelberg/Mannheim, Heidelberg, Germany

   Contribution

   Resources, Supervision

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1387-2636

8. Joachim Wittbrodt

   1. Centre for Organismal Studies (COS), Heidelberg University, Heidelberg, Germany
   2. DZHK (German Centre for Cardiovascular Research), partner site Heidelberg/Mannheim, Heidelberg, Germany

   Contribution

   Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – original draft, Writing – review and editing

   For correspondence

   jochen.wittbrodt@cos.uni-heidelberg.de

   Competing interests

   patent application pending (EP21166099.8) related to the findings described

   "This ORCID iD identifies the author of this article:" 0000-0001-8550-7377

• 2,517

  Page views

• 519

  Downloads

• 4

  Citations

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