Efficient genome editing using modified Cas9 proteins in zebrafish

  1. Laura Dorner
  2. Benedikt Stratmann
  3. Laura Bader
  4. Marco Podobnik
  5. Uwe Irion

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The zebrafish (Danio rerio) is an important model organism for basic as well as applied bio-medical research. One main advantage is its genetic tractability, which was greatly enhanced by the introduction of the CRISPR/Cas method a decade ago. The generation of loss-of-function alleles via the production of small insertions or deletions in the coding sequences of genes with CRISPR/Cas systems is now routinely achieved with high efficiency. The method is based on the error prone repair of precisely targeted DNA double strand breaks by non-homologous end joining (NHEJ) in the cell nucleus. However, editing the genome with base pair precision, by homology-directed repair (HDR), is by far less efficient and therefore often requires large-scale screening of potential carriers by labour intensive genotyping. Here we confirm that the Cas9 protein variant SpRY, with relaxed PAM requirement, can be used to target some sites in the zebrafish genome. In addition, we demonstrate that the incorporation of an artificial nuclear localisation signal (aNLS) into the Cas9 protein variants not only enhances the efficiency of gene knockout but also the frequency of HDR, thereby facilitating the efficient modification of single base pairs in the genome. Our protocols provide a guide for a cost-effective generation of versatile and potent Cas9 protein variants and efficient gene editing in zebrafish.

Article activity feed

  1. Version published to 10.1242/bio.060401
  7. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Point-by-point response to reviewers’ comments:

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

    Major comments:

    1. Previous studies using HDR and donor templates have shown that mutating the PAM sites in donor templates can enhance repair efficiencies. It would be helpful to add a discussion about the fact that SpRY does not have a PAM sequence that could be mutated and the potential consequences on repair efficiency.

    We find that one mismatch in the target sequence (i.e. albino wt v. albino[b4]) is enough to completely abolish activity of SpRY. We have stated this more clearly in the manuscript.

    It is also unclear how the template for the induction of mutations in kcnj13 was chosen. From the experiment with SpRY it seems that an HDR template equivalent to the sequence of the sgRNA target strand was most efficient, while in this experiment the alternative strand was used. An explanation should be added to the text.

    Oligonucleotides corresponding to both DNA strands were tested, only one of them yielded positive results. We do not know the mechanistic basis for this finding, but amended the manuscript accordingly.

    Minor comments:

    1. It is not directly evident what the difference between the OP2 and OP2* sgRNA is. A short explanation would help clarify this and make it easier for the reader to understand.

    OP2 (now re-named: U6) targets the wild type sequence, whereas OP2* (now: U6*) targets the albino[b4] sequence, which has one mutation leading to a premature stop codon. As this mutation is in the target region, we need adapt the sgRNA accordingly. We have stated this in the text more clearly now.

    Similarly, it would be helpful to add the length of the different donor templates to Figure 2.

    We have added the lengths of the oligonucleotides (in nt) to Fig.2.

    While the PAM sequences and their difference between guides is discussed for two of them (OP2 and U5), it would be helpful to add the PAM sequences for all guides to Table 1 or figure 1.

    We have added a table with all target sites including PAM sequences.

    For people who are unfamiliar with the obelix phenotype/pigment pattern, it would be helpful to add a picture of an obelix mutant to Figure 4, so they would know what the phenotype would look like and how obvious it would be.

    We have added a panel showing an obelix mutant fish to Fig.4.

    Reviewer #2:

    While every new and improved method to generate stable allele swap lines is greatly needed in the community, the results are not sufficient to convince me that the new version is leading to better success than previous methods. While they found one successful founder event, a single one is not enough to calculate efficiencies. Could just be luck that they got one. It is already known that HDR is very locus-specific, so maybe the locus they chose is such a locus.

    This comment is difficult to address; while we found that the improved HDR method we present in the paper leads to better success for the repair of the albinob4 mutation and the one specific allele exchange we performed, we, of course, agree that one founder event is not enough to calculate efficiencies. However, we would like to maintain that one founder will in almost all cases be better than none. We also think that the locus we chose, kcnj13, is not a particularly lucky one yielding positive results easily, because it used to be refractory to editing following published protocols for a long time.

    Overall, the paper suffers from the problem that the authors initially set out to investigate a specific genetic mutation in zebrafish but, upon observing that the resultant mutant exhibited no discernible phenotype, they shifted their focus towards refining and showcasing their methodological approach. This dual identity results in a study that, while informative, lacks the comprehensive exploration typical of dedicated research papers or the focused, technical depth one might expect from methodological publications.

    Overall, we feel that there might be a slight misunderstanding here. The reviewer states that ‘… the paper suffers from the problem that the authors initially set out to investigate a specific genetic mutation in zebrafish but, upon observing that the resultant mutant exhibited no discernible phenotype, they shifted their focus…’, which is quite the opposite of what actually lead to the writing of the manuscript. We had already suspected that the single amino acid difference in the protein sequence between the two sister species might not be responsible for the observed functional divergence of the gene. We had also already found allele-specific differences in expression levels in hybrids, which make cis-regulatory evolution more likely. So, the null-hypothesis of our experiments was that both protein sequences would be functionally equivalent. However, as we had difficulties with the allele exchange due to low HDR efficiencies we needed to improve the method before we could definitively show this.

    We have re-written some parts of the manuscript to make it clearer that we do not claim to have invented a method for HDR that is superior to all previously published ones. Rather, we think that we offer a variation of these published methods, which other researchers, struggling with low editing efficiencies (as we did), might want to try. What we do show in the manuscript is that the addition of an aNLS to Cas9 or SpRY leads to an increase in the efficiency in the generation of albino k.o. alleles and in HDR to repair the albinob4 mutation (see Fig. 3). If this will also be the case when editing other genes in the zebrafish genome needs to be investigated, but is clearly beyond the scope of this manuscript. We investigated one other locus in the zebrafish genome and could get one founder fish for the allele exchange in kcnj13, as opposed to zero we obtained with previously tried methods (conventional Cas9 with long or short donor-DNAs, prime editing). One advantage of ’our method’ is the simplicity of implementation. The Cas9 and SpRY proteins are easy to express in E. coli and the purification using two affinity tags is highly efficient resulting in samples of sufficient purity and high enough concentration for immediate use in injection experiments. So, we think that other researchers could easily try out the aNLS tagged proteins without changing much else of the protocols they usually employ for genome editing in zebrafish.

    Reviewer #3:

    Major comments: • The Cas9SpRY has been previously analyzed for the efficiency in zebrafish (Liang et al, Nat Comm 2022). This becomes only clear after reading the discussion. A comparison of these previously published SpRYCas9 proteins containing the bpNLS is missing, also a comparison of the efficiencies. The same locus (Albino) has been used in the study, are the guides identical? This study has not efficiently put the results in perspective of published results of the afore mentioned paper. And it seems that addition of the aNLS is not providing any benefit, which is good to know for the community.

    We have added information to the introduction making it clear that the SpRY protein has previously been used in zebrafish. We also expanded the discussion and added more details comparing our results to previously published ones. However, this comparison is not always easy because the evaluation methods are different, sequencing v. phenotypic read-out. While the addition of the aNLS to the SpRY protein did not significantly enhance the (already high) k.o. efficiency for the albino locus, it did result in a significant boost of the repair efficiency of the albino[b4] mutation (see Fig.3C). Therefore, we think that the general statement it ’is not providing any benefit’ might not be entirely accurate. We think that the use of SpRY could be beneficial in some instances, but it must be assessed one a case-by-case basis.

    The HDR numbers is relying on 1 germline founder fish and might not be representative. More loci and higher numbers would be desirable.

    We completely agree with the reviewer on this point. However, we feel that this is beyond the scope of this manuscript; we are looking forward to seeing other labs using the aNLS tagged proteins and finding out about their experiences.

    The allele exchange in Obelix is an interesting approach to use HDR but should be explained a little bit more. The motivation behind this experiments rains unclear.

    We have added some information on obelix to provide more context

    minor points: • All y axes require a labeling: % of what?!

    We have changed the labels to % of larvae.

    When showing the specific classes of phenotpes the reader would benefit if the classes were written directly into the fish picture rather than using B, C, D, etc...

    We have added this information directly to the pictures.

    OP2 should be called U6 to avoid unnecessary confusion, or is there anything special about it, why does it have another name?

    We have changed OP2 to U6, as requested. The naming was completely due to historic reasons, there is nothing special about this target site / sgRNA.

    Differences in efficiency could potentially attributed to the PAM sequence as discussed. Please list the different PAM sequences and discuss in more detail. Why are so many gRNAs not efficient in the KO approach (Figure1)?

    We added a table with the different target sites and the corresponding PAM sequences.

    While we cannot provide a satisfactory explanation for the low efficiencies of five from six sgRNAs in our experiments, we notice that in the published data from Liang et al., 2022, a sizeable proportion of the tested sgRNAs with the SpRY protein also show low efficiency or no activity at all (see Fig. 2B, Liang et al., 2022, https://doi.org/10.1038/s41467-022-31034-8). This phenomenon is likely to be locus-specific and more data will be needed to come to a mechanistic understanding. We also do acknowledge that there is the possibility that our assay, the albino mutant phenotype in larvae, is likely not as sensitive as sequencing-based approaches. For one, we rely on the bi-allelic k.o. of the target gene, and we only assess a small proportion of all larval cells. However, we think that our approach with a phenotypic read-out is still valid, as it will reflect the practical requirements for an HDR method in many laboratories, where low efficiencies will result in no or weak and variable F0 phenotypes and in very low probabilities for germ-line transmission, which in most cases researchers will want to avoid.

    Line 217: correct co.injected to co-injected

    done

    The scientific advancement is not clear. Readers would benefit if the advancement can be worked out better. Most readers would like to decide if it is worth changing their Cas9 design for genome editing in zebrafish and what efficiencies to expect.

    We have modified the manuscript to better convey the scientific advancement it presents. We think it lies mainly in the fact that no other changes to the design of genome editing experiments is required, but to exchange the Cas9 protein usually employed for the aNLS tagged proteins. Both proteins, aNLS tagged Cas9 or SpRY, can easily be produced and purified in the lab following standard protocols. In less than one week enough protein for several hundred or thousands of injection experiments can be purified and aliquoted. We suggest that everybody uses their tried and tested method to produce knock-in alleles, and, as long as it works for them, don’t change it. If, however, the efficiencies are too low to get the desired allele, it will be very quick and simple to try our method. This is what we wanted to demonstrate with the editing of the obelix locus. In all cases we can envisage identifying one founder fish will be considerably better than not finding a single one.

    Reviewer #4:

    Major

    1. The authors use a mutated version of the widely used Cas9 protein from Streptococcus pyogenes, SpRY which basically does not rely on a PAM motif adjacent to the sgRNA target site. While this has certain advantages which are properly described, lowering stringency also comes with disadvantages, i.e. enhanced off target site activity. While assessing these is of the scope of the paper, these considerations should be properly discussed. Under which circumstances do the authors suggest to use SpRY and at which the conventional Cas9 or TALENs?

    This is an important point and we have expanded on this. We think that SpRY offers a possibility to target sites that are not accessible to conventional Cas9, but it should not be expected to work as well as Cas9 for all loci (see also Liang et al., 2022 Fig.2). Whether the reduced stringency leads to more off-target effects is unclear; we did not experience higher rates of deformations or mortality in the injected larvae. This is, admittedly, a very crude measure for potential off-target effects, but is also in good accordance with the findings of Liang et al., 2022. In contrast to this, all labs that produce their own Cas9 protein could easily switch to the aNLS tagged version. It does not seem to have any disadvantages.

    The authors designed 6 guides against slc42a2/alb according to the text and to Fig1 U1-U5+OP. Table 1 contains 16 sequences fitting these criteria. Which ones where used? Why are they named differently (U vs OP)? What method was used to design them? Does their design include PAM requirements? Have these guides been used previously and confirmed to work efficiently using CAS9? If the authors intend to provide an improved method that can widely and easily be adopted by other labs, they should put special emphasis in describing the procedure properly possibly including a supplemental figure detailing the workflow.

    We have added a table with the target sites and the corresponding PAMs (see response to reviewer #1). The oligonucleotides shown in Table 1, which is now Table 2, are the ones used to generate the plasmid templates for the in vitro transcription of the sgRNAs.

    The naming of the target sites, which was solely due to ’historic’ reasons, has been changed to U1 - U6.

    They were designed (basically by hand) to allow in vitro transcription with T7 RNA polymerase (i.e. 5’ with GG), to have a G/C content of 50 - 65% and to represent a variety of different PAM sequences, that should potentially result in high activity (according to the data published by Walton et al., 2020 DOI: 10.1126/science.aba8853).

    These sgRNAs could not be tested with Cas9 as they lack the PAM (NGG) required for activity of this protein.

    We think that the main advantage of ’our’ method lies in the fact that aNLS-Cas9 (and aNLS-SpRY) can easily incorporated into the experimental procedures and workflows already in place in other laboratories. There is no need to follow exactly our protocol, eg. regarding sgRNA production or target site selection. We think that we showed that SpRY can be as effective as conventional Cas9, but not for all target sites, and that the addition of an aNLS sequence to Cas9 or SpRY is beneficial for genome editing in zebrafish, even when the aNLS is not combined with a myc-tag, as is the case shown by Thumberger et al., 2022, i.e. hei-tag.

    The authors use a recessive pigment mutant (albino) to validate and quantify precise genome editing by HDR applying their toolbox. This is very clever and probably the most robust readout possible. The authors found that adding an aNLS to CAS9 and SpRY improves rescue efficiency, possibly also for germ line transmission. The authors should compare their efficiency for accurate editing with that of other papers in the field to allow for a better comparison.

    We have now included a more detailed comparison of our results with previously published data in the discussion. However, this comparison is not always easy because the evaluation methods are different, sequencing v. phenotypic read-out. In terms of accuracy of the methods, we found that the majority of the HDR events we detected were associated with additional mutations. Some of these were possibly due to synthesis errors in the donor oligonucleotides, which might be alleviated by different purification methods. Other mutations, however, most likely occurred during cellular repair of dsDNA breaks and are therefore not easily avoided, unless double strand breaks are avoided, which would be the case if base editors are used. However, with base editors it is so far not possible to introduce every possible DNA change, making HDR methods still useful.

    Minor:

    1. Fig.1A: Please indicate orientation of the gene

    done

    Line 168: ... tested sperm... à Method not explained in the methods section

    The sperm samples extracted from anaesthetized males were used in exactly the same way as larvae were in other genotyping experiments; as is mentioned in the methods section. We have re-phrased this section a bit to make it clearer that we used larvae or sperm in exactly the same way for genotyping.

    Kcnj13 editing. Explain obelix pigment phenotype to the non expert reader in pigmentation possibly illustrating D. aesculapii. This is a very powerful method allowing such comparisons, however it is not properly explained.

    We have added some information on the obelix phenotype and included a panel of a mutant zebrafish in Fig.4.

    Line 130: 'hei-tag' not properly explained

    The hei-tag, published by Thumberger et al., 2022, consists of a myc-tag, a flexible linker and an aNLS in exactly that order. We have added some more details on the hei-tag to the text.

    The co-editing of a restriction site for later identification of the edited allele is clever. However precise editing should be performed carefully and include splice site prediction algorithms to avoid enabling ectopic splice sites by silent mutagenesis. Also, an example of the analysis would be benefitial to Fig.4 or in the supplement.

    We agree that this is an important point. We originally designed the edit in a way that would not result in the generation of a strong ectopic splice site by avoiding the creation of AG or GT di-nucleotide sequences.

    We now also performed analysis with spliceator (http://www.lbgi.fr/spliceator/), a splice site prediction tool using convolutional neural networks, which confirmed that no ectopic splice site should be generated.

    We could include this into a supplementary figure, if deemed necessary.

    The manuscript is well written, the data are presented in an accessible way and the results look convincing. The work clearly shows a path to improvement of a fundamental method of gene editing in zebrafish and other species and clearly provides essential data on the topic. However, some aspects of the work are not properly described for the non-expert. Given the nature of the work which aims to improve an important, established method a more precisely described workflow in form of a table and workflow chart would certainly help the reader to focus on the essentials of the procedure.

    As mentioned above, we think that it will be easy for other labs to incorporate our improvements into their existing protocols by exchanging normal Cas9 for aNLS-Cas9 or aNLS-SpRY. There should not be the need to strictly follow our protocols, e.g., for target site selection or sgRNA synthesis. The proteins can easily be expressed in bacteria and purified by standard methods using the His- and Strep-tags, as we published previously for conventional Cas9 (Podobnik et al. 2023).

  8. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #4

    Evidence, reproducibility and clarity

    In this manuscript, Dorner, Stratmann et al. developed a new variant of the homologous directed repair mediated genome editing technique in zebrafish using modified Cas9 proteins. They focus on the SpRY Cas9 protein variant, which offers a more relaxed PAM requirement for gene targeting. The requirement of a PAM has particularly hampered the feasabilty of HDR in the zebrafish model as the genomic sites of interest often do not meet the PAM requirements for conventional Cas9. Their improved method enhances the versatility of CRISPR/Cas methods in zebrafish, a crucial model organism in biomedical research. The authors also demonstrate that integrating an artificial nuclear localization signal (aNLS) into Cas9 variants not only improves gene knockout efficiency but also boosts homology-directed repair (HDR) frequency. This advancement allows for more precise genetic modifications, including single base pair changes, offering significant potential for research and other applications in genetics.

    The manuscript is well written, the data are presented in an accessible way and the results look convincing. The work clearly shows a path to improvement of a fundamental method of gene editing in zebrafish and other species and clearly provides essential data on the topic. However, some aspects of the work are not properly described for the non-expert. Given the nature of the work which aims to improve an important, established method a more precisely described workflow in form of a table and workflow chart would certainly help the reader to focus on the essentials of the procedure.

    Major comments:

    1. The authors use a mutated version of the widely used Cas9 protein from Streptococcus pyogenes, SpRY which basically does not rely on a PAM motif adjacent to the sgRNA target site. While this has certain advantages which are properly described, lowering stringency also comes with disadvantages, i.e. enhanced off target site activity. While assessing these is of the scope of the paper, these considerations should be properly discussed. Under which circumstances do the authors suggest to use SpRY and at which the conventional Cas9 or TALENs?

    2. The authors designed 6 guides against slc42a2/alb according to the text and to Fig1 U1-U5+OP. Table 1 contains 16 sequences fitting these criteria. Which ones where used? Why are they named differently (U vs OP)? What method was used to design them? Does their design include PAM requirements? Have these guides been used previously and confirmed to work efficiently using CAS9? If the authors intend to provide an improved method that can widely and easily be adopted by other labs, they should put special emphasis in describing the procedure properly possibly including a supplemental figure detailing the workflow.

    3. The authors use a recessive pigment mutant (albino) to validate and quantify precise genome editing by HDR applying their toolbox. This is very clever and probably the most robust readout possible. The authors found that adding an aNLS to CAS9 and SpRY improves rescue efficiency, possibly also for germ line transmission. The authors should compare their efficiency for accurate editing with that of other papers in the field to allow for a better comparison.

    Minor comments:

    1. Fig.1A: Please indicate orientation of the gene

    2. Line 168: ... tested sperm...  Method not explained in the methods section

    3. Kcnj13 editing. Explain obelix pigment phenotype to the non expert reader in pigmentation possibly illustrating D. aesculapii. This is a very powerful method allowing such comparisons, however it is not properly explained.

    4. Line 130: 'hei-tag' not properly explained

    5. The co-editing of a restriction site for later identification of the edited allele is clever. However precise editing should be performed carefully and include splice site prediction algorithms to avoid enabling ectopic splice sites by silent mutagenesis. Also, an example of the analysis would be benefitial to Fig.4 or in the supplement.

    Significance

    The manuscript is well written, the data are presented in an accessible way and the results look convincing. The work clearly shows a path to improvement of a fundamental method of gene editing in zebrafish and other species and clearly provides essential data on the topic. However, some aspects of the work are not properly described for the non-expert. Given the nature of the work which aims to improve an important, established method a more precisely described workflow in form of a table and workflow chart would certainly help the reader to focus on the essentials of the procedure.

  9. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    The manuscript "efficient genome editing using modified Cas9 Proteins in zebrafish" by Dorner et al. is using the modified SpRY Cas9 protein with the addition of an artificial nuclear localization sequence (aNLS) and compares its efficiency in the generation of making KO animals with other Cas9 proteins and its use to generate a HDR mediated KI in zebrafish. The paper shows that the SpPY Cas9 works efficiently and that addition of aNLS can increase the HDR mediated efficiency in one locus.

    Major comments:

    • The Cas9SpRY has been previously analyzed for the efficiency in zebrafish (Liang et al, Nat Comm 2022). This becomes only clear after reading the discussion. A comparison of these previously published SpRYCas9 proteins containing the bpNLS is missing, also a comparison of the efficiencies. The same locus (Albino) has been used in the study, are the guides identical? This study has not efficiently put the results in perspective of published results of the afore mentioned paper. And it seems that addition of the aNLS is not providing any benefit, which is good to know for the community.

    • The HDR numbers is relying on 1 germline founder fish and might not be representative. More loci and higher numbers would be desirable.

    • The allele exchange in Obelix is an interesting approach to use HDR but should be explained a little bit more. The motivation behind this experiments rains unclear. minor points:

    • All y axes require a labeling: % of what?!

    • When showing the specific classes of phenotpes the reader would benefit if the classes were written directly into the fish picture rather than using B, C, D, etc...

    • OP2 should be called U6 to avoid unnecessary confusion, or is there anything special about it, why does it have another name?

    • Differences in efficiency could potentially attributed to the PAM sequence as discussed. Please list the different PAM sequences and discuss in more detail. Why are so many gRNAs not efficient in the KO approach (Figure1)?

    • Line 217: correct co.injected to co-injected

    Significance

    The scientific advancement is not clear. Readers would benefit if the advancement can be worked out better. Most readers would like to decide if it is worth changing their Cas9 design for genome editing in zebrafish and what efficiencies to expect.

  10. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The manuscript: Efficient genome editing using modified Cas9 proteins in zebrafish by Dorner and Stratmann et al. provides a putative improved method of modifying single base pairs in the genome of zebrafish through homology-directed repair. The authors use a modified Cas9 protein called SpRY in zebrafish. The SpRY protein has fewer restrictions on the PAM sequence it requires, which broadens its genome targeting potential. The paper presents experiments on zebrafish using SpRY for efficient genome editing. However, not all target sites are equally efficient, and the authors suggest that individual sequences may need to be evaluated on a case-by-case basis.

    The paper also explores the benefits of using an optimized artificial nuclear localization signal (aNLS) for the Cas9 protein, which significantly increases genome editing efficiency in zebrafish. Using this improved method, the researchers demonstrate precise editing of the kcnj13 gene in zebrafish to match the sequence found in Danio aesculapii, their closest sister species. The edited zebrafish do not show any visible phenotypic changes, suggesting that the Kcnj13 proteins from both species are functionally equivalent.

    While every new and improved method to generate stable allele swap lines is greatly needed in the community, the results are not sufficient to convince me that the new version is leading to better success than previous methods. While they found one successful founder event, a single one is not enough to calculate efficiencies. Could just be luck that they got one. It is already known that HDR is very locus-specific, so maybe the locus they chose is such a locus.

    Significance

    Overall, the paper suffers from the problem that the authors initially set out to investigate a specific genetic mutation in zebrafish but, upon observing that the resultant mutant exhibited no discernible phenotype, they shifted their focus towards refining and showcasing their methodological approach. This dual identity results in a study that, while informative, lacks the comprehensive exploration typical of dedicated research papers or the focused, technical depth one might expect from methodological publications.

  11. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary:

    The manuscript "Efficient genome editing using modified Cas9 proteins in zebrafish" by Dorner et al. describes the utility of combining Cas9 variants with an alternative NLS to improve the efficiency of induction of single base pair changes in zebrafish through HDR. The authors confirm findings from previous studies, that the SpRY variant of Cas9 can be used in zebrafish to induce knockouts at efficiencies similar to Cas9 and additionally show the ability of SpRY to induce point mutations via HDR. They further confirm other studies that showed that the addition of an artificial NLS can significantly increase targeting efficiency of Cas9 in zebrafish. They extend their studies to show that the increased efficiency through addition of this alternative NLS also enhances HDR mediated induction of single base-pair changes using donor templates for both Cas9 variants. Most conclusions of the paper are well supported by the presented data and the experiments are explained in sufficient detail for replication. The paper could benefit from a few modifications/additions to the text to clarify a few details.

    Major comments:

    Previous studies using HDR and donor templates have shown that mutating the PAM sites in donor templates can enhance repair efficiencies. It would be helpful to add a discussion about the fact that SpRY does not have a PAM sequence that could be mutated and the potential consequences on repair efficiency. It is also unclear how the template for the induction of mutations in kcnj13 was chosen. From the experiment with SpRY it seems that an HDR template equivalent to the sequence of the sgRNA target strand was most efficient, while in this experiment the alternative strand was used. An explanation should be added to the text.

    Minor comments:

    1. It is not directly evident what the difference between the OP2 and OP2* sgRNA is. A short explanation would help clarify this and make it easier for the reader to understand.

    2. Similarly, it would be helpful to add the length of the different donor templates to Figure 2.

    3. While the PAM sequences and their difference between guides is discussed for two of them (OP2 and U5), it would be helpful to add the PAM sequences for all guides to Table 1 or figure 1.

    4. For people who are unfamiliar with the obelix phenotype/pigment pattern, it would be helpful to add a picture of an obelix mutant to Figure 4, so they would know what the phenotype would look like and how obvious it would be.

    Significance

    General assessment: Strength of the study are the use of multiple, independent injection experiments for each group to test guide and repair efficiency. Clear presentation of methods which will allow replication of experiments and also production of reagents. While the study clearly shows that SpRY does work for HDR mediated repair, only one specific repair template design (single stranded oligo) was tested. This study could be enhanced through testing of additional HDR template designs and a direct comparison of repair efficiency between Cas9 and SpRY.

    Advance: This study does provide a minor advance in our understanding of how the efficiency of Cas9 and its variants can be optimized and how these modifications can enhance repair efficiency. The use of the SpRY variant in zebrafish as well as the enhancement of Cas9 efficiency through the use of the aNLS has been shown before (Liang et al. 2022, Thumberger et al. 2022). Novelties in this study are the use of SpRY for HDR mediated repair and the improvement of the repair efficiency through the addition of the aNLS to SpRY and Cas9.

    Audience: The methods described in this paper will be of interest to the zebrafish and Medaka communities as well as people using HDR mediated repair for the induction of mutations.

    Expertise: I am a geneticist who has worked with the zebrafish model for over 20 years and uses CRISPR/Cas for genome editing in zebrafish routinely.

  12. Version published to 10.1101/2023.11.12.566744v1 on bioRxiv