Versatile CRISPR/Cas9-mediated mosaic analysis by gRNA-induced crossing-over for unmodified genomes

Mosaic animals have provided the platform for many fundamental discoveries in developmental biology, cell biology, and other fields. Techniques to produce mosaic animals by mitotic recombination have been extensively developed in Drosophila melanogaster but are less common for other laboratory organisms. Here, we report mosaic analysis by gRNA-induced crossing-over (MAGIC), a new technique for generating mosaic animals based on DNA double-strand breaks produced by CRISPR/Cas9. MAGIC efficiently produces mosaic clones in both somatic tissues and the germline of Drosophila. Further, by developing a MAGIC toolkit for 1 chromosome arm, we demonstrate the method’s application in characterizing gene function in neural development and in generating fluorescently marked clones in wild-derived Drosophila strains. Eliminating the need to introduce recombinase-recognition sites into the genome, this simple and versatile system simplifies mosaic analysis in Drosophila and can in principle be applied in any organism that is compatible with CRISPR/Cas9.

4. We listed potential gRNA sequences for applying MAGIC to other major chromosome arms in Table S2. Although we were unable to establish full MAGIC toolkits for all chromosome arms within the timeframe allowed for the revision, we hope other researchers can generate their own MAGIC reagents based on these recommendations using our or their own vectors while we create full toolkits for all arms and make them publicly available. 5. We revised our figures and texts to address all issues raised by both reviewers. We also made other small changes to the manuscript to improve its readability.

Reviewer #1:
Allen and colleagues describe a novel CRISPR/Cas9-based technique in Drosophila enabling the generation of genetic mosaics through interchromosomal recombination. The authors validate their approach in a number of somatic tissues and the female germline. The authors also provide a tool set for the genetic mosaic-based functional study of candidate genes on chromosome arm 2L. Lastly, the authors cross their transgenic lines to fly lines with wild-derived genomes and provide proof-of-principle.
Overall the authors present a potentially useful method for the study of candidate genes in genetic mosaic flies. Currently, the manuscript presents mainly proof-of-principle and the resource is still limited. While the data presentation is sound, the writing of the manuscript could be improved. At many passages the writing is not precise. . Thus they will never know when the DSB was induced and interchromosomal recombination happened. As a consequence they also will not know if more than one event happened and in more than one progenitor stem cell.

Response:
We thank the reviewer for pointing out the important distinction between mosaic analysis and clonal analysis, and we apologize for using them loosely in the original submission. As we show in the revised manuscript (Figure 3), MAGIC can be made inducible by heat shock. Therefore, in principle it can be used in clonal analysis, even though we did not specifically distinguish whether the labeled cells were derived from single stem cells in our experiments. We revised our text for a more stringent use of these terms and to better reflect the intent of our experiments.
3. Line 82 -the authors state: '…, a 50% chance exists for identical distal chromosome segments to sort into the same daughter cell, generating …' The authors should please precise their writing and explain the entire spectrum of segregation possibilities in more detail. If recombination happens in G2 phase of the cell cycle, there are two segregation possibilities, G2-X (recombinant chromosomes segregate away from each other) and G2-Z (recombinant chromosomes 'sort' together into the same cell). The authors should elaborate for the nonspecialist reader and clarify the schematic in Figure 1.

Response:
We thank the reviewer for this suggestion. We revised the text to better explain all outcomes of chromosome segregation illustrated in Figure 1, as suggested.

The authors state on line 119 -'…label clones homozygous … either negatively or positively…'. How can something be negatively labeled. Please be precise the writing.
Response: Thanks to the reviewer for pointing this error. We have changed our text to remove the phrase "negative labeling". Response: We agree with the reviewer that unique target sites do not exclude off-targeting. We used published algorithms to predict the probability of off-targeting. The off-targeting scores are now included in the gRNA table (Table S2). We revised our discussion to mention potential offtargeting effects, as the reviewer suggested.

Line
6. The authors state on line 146: '… suggests that an efficient gRNA construct for one tissue will likely perform well in other tissues also…'. Please remove such speculation or show the data with quantitative assessment. Also in the discussion line 195/6.

Response:
We removed this statement. Response: When we say that analogous toolkits could easily be made for other chromosome arms, we meant that making reagents for other arms is intellectually straightforward and the procedures for making them are standard molecular cloning and transgenic steps. It is, in fact, our goal to make the entire MAGIC toolkits for all major chromosome arms. However, this task requires far longer than the 3-month turnaround time allowed for our revision. It requires molecular cloning, establishing, stabilizing, and verifying 24 different transgenic lines, and then functionally validating and comparing the transgenes in multiple tissues. The fly generation time alone means that simply generating and stabilizing the transgenes and fly-lines would take a minimum of 3 months, if all went well and optimally. The current pandemic makes the situation even worse. Both of our labs are small and resource-limited. The personnel on whom we rely to build these reagents currently have very limited accessibility to our partially-operating labs. Therefore, we expect that completing the 4 additional full kits requested will likely take much longer than we originally planned, well beyond 3 months.
We understand that it would have been ideal to have all MAGIC kits for all chromosome arms in this paper. Indeed, the original MARCM paper was an exemplar of that, providing complete reagents for all contingencies. But we are hoping that generating and validating one kit would be sufficient proof-of-principle for this paper, given the time and labor required to make all of the remaining kits. We plan to build the remaining kits in the near future, and will deposit them into stock centers as soon as they are ready and verified. Meanwhile, we have added in Table S2 the gRNAs we propose for the other chromosome arms, and will deposit the cloning vectors into Addgene. This will allow interested labs to start taking advantage of this new method by establishing transgenes for their own studies, while we build the kit for general use.

Line 169/170 -'…demonstrating the potential of MAGIC for clonal analysis of the function of natural alleles residing…'. The authors did not do any functional analysis of natural alleles. Please tone down the claims or at least discuss such issues in a more balanced manner.
Response: Thanks to the reviewer for this suggestion. We revised our text to tone down our claim.
9. The entire Discussion aims to sell the method, which in principle is fine. However, the entire Discussion should be written in a more balanced manner. Please do not discredit previous work based on FRT/Flp-mediated recombination. There are thousands of studies that exploit such technique and even once MAGIC is available for the community, many people will continue

using FRT/Flp systems for functional gene analysis. In fact, FRT sites have been inserted and reliably used on (almost) all chromosome arms. MAGIC currently is only enabling functional gene analysis on 2L.
Response: We apologize for giving the reviewer the impression of discrediting previous work based on FRT/Flp-mediated recombination -this was certainly not our intention. We revised the discussion in the hope that the reviewer and future readers will find our comparison of the two approaches to be objective and balanced.

Reviewer #2: [identifies himself as Liqun Luo]
In this manuscript, the authors described a new method for mosaic analysis in Drosophila.

Response:
We appreciate the reviewer's valid concern on the efficiency of MAGIC and thank him for suggesting comparisons between MAGIC and FRT/Flp. In the revised manuscript (Figure 3), we compared the efficiency of our 2L nMAGIC reagent in generating clones in the wing imaginal disc to that of FRT 40A using Cas9 and Flp lines, respectively, driven by the same enhancers. FRT 40A was chosen for the comparison because ubiquitously expressed fluorescent markers are available on FRT 40A chromosomes for easy identification and quantification of mosaic clones. Surprisingly, we found that MAGIC generated larger and more frequent clones than FRT/Flp. Although the low efficiency of the FRT/Flp method in these experiments may be due to the property of this particular FRT site, our results nevertheless demonstrate that MAGIC can be reliably used to generate mosaic clones in imaginal discs. Similarly, when we examined LOF phenotypes of Sec5, Rab5, and Syx5 using SOP-Cas9 to induce mutant clones in da neurons, we were able to identify multiple clones in every larva. Although we did not compare the efficiencies of clone induction between MAGIC and FRT/Flp in da neurons, our experience suggests that MAGIC can be very usable in the nervous system as well, as long as appropriate gRNA target sites and Cas9 lines have been identified.
2. The author should demonstrate a temporally inducible way of making clones-such as using a heat shock promoter to drive Cas9. This is one of the most widely used ways to induce clones in the field. This is especially important in light of the caveat the authors raised in their discussion: "For the cell type in question, an ideal Cas9 should be expressed in the precursor cells, as too early expression can mutate gRNA target sites prematurely and too late expression will lead to unproductive DSBs."

Response:
We thank the reviewer for suggesting this experiment. We were fortunate to obtain from Dr. Tzumin Lee a HS-Cas9 line recently made by his lab. In our heat shock experiments using this line (Figure 3), we show that a single one-hour heat shock at 37°C at 72 hours after egg laying was sufficient to generate many clones in the wing imaginal disc, while larvae of the same genotype grown at 25°C showed almost no clones. Therefore, we conclude that MAGIC clones can be temporally induced.
3. Mosaic analysis technique papers in Drosophila have typically included resources that allow researchers to use the tools right away, rather than having to create the tools AND apply the tools. This will speed up the adoption of new techniques. The authors have produced tools for analysis of genes located on 2L, which covers only 20% of the genome. The paper will be greatly improved if the authors can also provide tools for other chromosomal arms. These will also serve to further validate the generality of the approach. I understand that this is a substantial amount of work (creating new transgenes) in particular during the pandemic, so I will leave it up to the journal editors to decide whether it is an option or a requirement.

Response:
We appreciate the reviewer's comment and agree that it would be ideal to provide the full MAGIC kit to the fly community at the time of publication. It is, in fact, also our goal to make the entire MAGIC toolkits for all major chromosome arms. However, this task requires far longer than the 3-month turnaround time allowed for our revision. It requires molecular cloning, establishing, stabilizing, and verifying 24 different transgenic lines, and then functionally validating and comparing the transgenes in multiple tissues. Although the kits are intellectually straightforward to generate, and the fly manipulations are simple and standard too, the fly generation time alone means that simply generating and stabilizing the transgenes and fly-lines would take a minimum of 3 months, if all went well and optimally. The current pandemic makes the situation even worse. Both our labs are small and resource-limited. The personnel on whom we rely to build these reagents currently have very limited accessibility to our partially-operating labs. Therefore, we expect that completing the 4 additional full kits requested will likely take much longer than we originally planned, and well beyond 3 months.
We do understand that it would have been ideal to have all MAGIC kits for all chromosome arms in this paper. Indeed Dr. Luo's MARCM paper was an exemplar of that, providing complete reagents for all contingencies. But we are hoping that generating and validating one kit would be sufficient proof-of-principle for this paper, given the time and labor required to make all of the remaining kits. We plan to build the remaining kits in the near future, and will deposit them into stock centers as soon as they are ready and verified. Meanwhile, we have added in Table S2 the gRNAs we propose for the other chromosome arms and will deposit the cloning vectors into Addgene. This will allow interested labs to start taking advantage of this new method by establishing transgenes for their own studies, while we build the kit for general use.
Minor issues: 4. The statement on line 82 (a 50% chance…) is incorrect. G2-X (Fig. 1A top) and G2-Z (Fig.  1A, bottom) segregations are known to be unequal. There is a literature on this in Drosophila and in other organisms.
Response: Thanks for pointing out this error. We have included in the discussion that G2-X segregation is predominant in Drosophila, as shown in Beumer et al., Genetics 1998.