Gene editing and scalable functional genomic screening in Leishmania species using the CRISPR/Cas9 cytosine base editor toolbox LeishBASEedit

  1. Markus Engstler
  2. Tom Beneke

  • Curated by eLife

    eLife logo

    eLife assessment

    Only few species of Leishmania, an important human pathogen, have an RNAi machinery, alternative methods are needed for genetic screens. The authors resent and validate a valuable method, based on the introduction of premature stop codons, that can be used for several different species. The results are very convincing, the data are solid, and the approach will be of interest to researchers studying any eukaryote that lacks the RNAi machinery.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

CRISPR/Cas9 gene editing has revolutionised loss-of-function experiments in Leishmania , the causative agent of leishmaniasis. As Leishmania lack a functional non-homologous DNA end joining pathway however, obtaining null mutants typically requires additional donor DNA, selection of drug resistance-associated edits or time-consuming isolation of clones. Genome-wide loss-of-function screens across different conditions and across multiple Leishmania species are therefore unfeasible at present. Here, we report a CRISPR/Cas9 cytosine base editor (CBE) toolbox that overcomes these limitations. We employed CBEs in Leishmania to introduce STOP codons by converting cytosine into thymine and created http://www.leishbaseedit.net/ for CBE primer design in kinetoplastids. Through reporter assays and by targeting single- and multi-copy genes in L. mexicana , L. major , L. donovani , and L. infantum , we demonstrate how this tool can efficiently generate functional null mutants by expressing just one single-guide RNA, reaching up to 100% editing rate in non-clonal populations. We then generated a Leishmania -optimised CBE and successfully targeted an essential gene in a plasmid library delivered loss-of-function screen in L. mexicana . Since our method does not require DNA double-strand breaks, homologous recombination, donor DNA, or isolation of clones, we believe that this enables for the first time functional genetic screens in Leishmania via delivery of plasmid libraries.

Article activity feed

  1. Version published to 10.7554/elife.85605 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    We would like to thank reviewer #1 for her helpful comments and would like to respond to these as follows:

    1. “Editing efficiencies were variable (99% to 0%) depending on the species, being worst for L. major.”

    It is true that the editing efficiency was different in each species and worst for L. major. However, it is important to note that these efficiencies varied not only for each species but also amongst genes and especially chosen sgRNA sequences. Variations in efficiency across sgRNAs targeting the same gene and locus is a common problem in any CRISPR approach. We made this clearer in our revised manuscript (line 670 – 673).

    1. “The use of premature termination codons also clearly raises issues for false positives and negatives, especially as there is no evidence for nonsense-mediated mRNA decay in Leishmania.”

    We have now included in our revised manuscript that it is currently unclear whether a classical nonsense-mediated decay pathway is present in Leishmania or not. If such a pathway would be present, mutant mRNAs in which a termination codon is present within the normal open reading frame would be removed (Clayton, Open Biology 2019; Delhi et al., PLoS One 2011). But if not, remaining N-terminal protein parts could be functional and may lead to false positive and negative results. However, as reviewer #2 pointed out, this may also provide extra information about functional domains of the targeted protein and highlights that our tool can not only be used to create functional null mutants by inserting premature STOP codons but also to pursue targeted mutagenesis screens (line 674 - 683).

    1. “There are already two genome-wide screening options for Leishmania, so the advantages and disadvantages of the method proposed here need to be discussed in a much more detailed and balanced way.”

    We have revised our manuscript to include in our introduction (line 36 - 73) and discussion (line 658 - 697) a better comparison of all potential tools for genome-wide screening in Leishmania, including RNAi, bar-seq and base editing screening. We highlight why we think that base editing has unique advantages.

    1. “In the "LeishGEM" project (http://www.leishgem.org) all Leishmania mexicana genes will be knocked out and each KO will be bar-coded. At the end, 170 pooled populations of 48 bar-coded mutants will be publicly available. The only real reason the authors of the current paper give for not using this approach is that it is labour-intensive. However, LeishGEM is funded and underway, with several centres involved, so that argument is weak.”

    In our original manuscript we gave multiple reasons why we think that the LeishGEdit method, which is being used for the LeishGEM screen and has been developed by the lead author of our here presented study, has clear disadvantages compared to base editing.

    As written in our original manuscript (line 709 – 716): “However, for a bar-seq screen, each barcoded mutant needs to be created individually by replacing target genes with drug selectable marker cassettes (20,21), making them extremely labour intensive and most likely “one-offs” on a genome-wide scale. Furthermore, aneuploidy in some Leishmania species can be a major challenge for gene replacement strategies as multiple rounds of transfection or isolation of clones may be required to target genes on multi-copy chromosomes. Using gene replacement approaches it is also not feasible to study multi-copy genes that have copies on multiple chromosomes. These are major disadvantages of bar-seq screening.”

    Therefore, we still think that the main disadvantage of bar-seq screening is that it is labour-intensive as each mutant needs to be created individually. The fact that LeishGEM requires five years and several research centres to knockout all genes in just one Leishmania species is proof for this argument.

    However, to clarify our position about this further, we have listed other disadvantages of the LeishGEM screen, including difficulties of sharing mutant pools between labs, possible problems in expanding mutant pools without losing uniformity, no ability to change the composition of generated pools and limited ability to distinguish between technical failures and essentiality. If any of these problems would occur, it would require a de novo generation of barcoded mutants and therefore this is an extremely labour-intensive method for large-scale screening. We also added that bar-seq screens are not feasible in Leishmania species that display extreme cases of aneuploidy, such as L. donovani (line 59 – 73).

    Despite all these disadvantages of the LeishGEdit approach for the LeishGEM project, there are of course also clear advantages, which we also point out in our introduction (line 52 – 55).

    1. “There is also a preprint describing RNAi for functional analysis in Leishmania braziliensis.”

    Although our original manuscript included the pre-print about RNAi screening in Leishmania braziliensis already (line 706-709), we understand that this deserves a stronger discussion. We have therefore highlighted now RNAi as a possible tool for genome-wide screening in selected Leishmania species in our revised introduction (line 36 - 43). However, we also argue that RNAi approaches are at the moment only available to Leishmania of the Viannia subgenus and that RNAi activity greatly varies between the species (line 36 – 43 and 665 - 669). In addition, we discuss that the use of RNAi genome-wide screens is much less specific, as usually randomly sheared genomic DNA is used to generate RNAi libraries (line 687 - 689). Since the pre-print is now published, we have replaced the pre-print publication with the peer-reviewed one.

    Reviewer #2 (Public Review):

    We would like to thank reviewer #2 for helpful comments and would like to respond to those as follows:

    1. “Line 482 - the authors wrote 'As expected, the proportion of cells showing a motility phenotype in the IFT88 targeted L. infantum population decreased further' Why is this result expected? Presumably, this is due to the fact that cells without a functional IFT system lack flagella and grow slower so can be outcompeted by faster-growing mutants. This speaks to the major caveat highlighted by the authors in the discussion and the final small-scale screen. In a population of cells, those with deleterious mutations in an essential gene or one whose disruption results in slower growth will be outcompeted by cells in which a non-deleterious mutation has occurred, which feeds into the issue of timing.”

    As the reviewer highlighted himself, deleterious mutations that result in slower growth will be outcompeted by cells in which a non-deleterious mutation has occurred. We have stated that the complete deletion of IFT88 in Leishmania mexicana has been shown to have reduced doubling time (Beneke et al., PLoS Pathogens 2019) and are therefore most likely outcompeted from the pool (line 529 – 532 and 767 - 769).

    1. “The authors show with CRK3 this process of non-deleterious mutants outcompeting deleterious mutants does result in a detectable drop in the number of parasites with specific CRK3 guides but not in those with IFT88. Is this due to the fact that the outgrowth of the non-deleterious IFT88 mutants occurs rapidly or that the mutation of the targets in IFT88 was ineffective? The data presented in Figure 5 shows that for some species at least a mutation of the IFT88 gene was possible. This might mean that for certain genes the outgrowth occurs within the first 12 days after transfections so will not be seen using this approach, without a wider study, which is beyond the scope of this manuscript it will be difficult to know.”

    As we stated in our discussion, we did not test IFT88 guides individually in L. mexicana. Therefore, the editing rate observed for the IFT88 guides in L. major and L. infantum (Fig. 5) may differ from the editing rate in L. mexicana, which is the species we used for the pooled transfection screen. It is therefore difficult to conclude why IFT88 was not depleted from the pool. This may be due to lower guide activity in L. mexicana or rapid selection of non-deleterious mutations (line 769 - 774). We are therefore planning to further optimize our system by streamlining the editing efficiency and eliminating species-specifics effects (line 735 - 745). As the reviewer highlighted, this is beyond the scope of this study.

    However, the reviewer raises a fair point about the exact timing of isolating DNA from pools, which might influence when exactly parasites with a deleterious mutation are depleted from the pool. This may differ between guides and may even be gene specific. We have added this point to our discussion (776 - 780).

    1. “The authors highlight that this base editing approach will leave potentially functional regions of the NT of proteins, which is true and may mean genes are missed. However, this may also provide extra information about the protein's function/domain structure if STOP codons in certain positions showed an effect on function whereas those in others don't.”

    We thank reviewer #2 for pointing out that functional parts of truncated proteins following base editing may actually allow to draw additional conclusions. We have included this in the manuscript (681 - 683).

  3. eLife

    eLife assessment

    Only few species of Leishmania, an important human pathogen, have an RNAi machinery, alternative methods are needed for genetic screens. The authors resent and validate a valuable method, based on the introduction of premature stop codons, that can be used for several different species. The results are very convincing, the data are solid, and the approach will be of interest to researchers studying any eukaryote that lacks the RNAi machinery.

  4. eLife

    Reviewer #1 (Public Review):

    The basis of this method is to clone guides into a Crispr-based editing plasmid, transfect pools into Leishmania, maintain them as episomes, then look at phenotypes. The guides are designed to cause editing that converts codons to stop codons, and the authors have designed a computational tool that enables the design of guides that work for the first half of each gene. Selection for the episome is necessary and editing efficiencies were variable (99% to 0%) depending on the species, being worst for L. major. The use of premature termination codons also clearly raises issues for false positives and negatives, especially as there is no evidence for nonsense-mediated mRNA decay in Leishmania.

    There are already two genome-wide screening options for Leishmania, so the advantages and disadvantages of the method proposed here need to be discussed in a much more detailed and balanced way.
    In the "LeishGEM" project (http://www.leishgem.org) all Leishmania mexicana genes will be knocked out and each KO will be bar-coded. At the end, 170 pooled populations of 48 bar-coded mutants will be publicly available. The only real reason the authors of the current paper give for not using this approach is that it is labour-intensive. However, LeishGEM is funded and underway, with several centres involved, so that argument is weak.
    There is also a preprint describing RNAi for functional analysis in Leishmania braziliensis.

  5. eLife

    Reviewer #2 (Public Review):

    This is a well-written and clear manuscript, in which the authors describe the stepwise development of an approach for loss of function screens in a range of different Leishmania species, culminating in a small-scale screen. The method relies on CRSIPR/Cas9 directed mutation of cytosine bases to generate premature STOP codons. The conclusions of the manuscript are well supported by the data presented and this approach appears to have great potential to facilitate functional studies and discovery biology in a range of different species.

    The authors have presented the development of their base editing toolbox in a stepwise manner, showing the optimisation steps they took. They initially used a tdTomato expressing cell line to optimise which base editor to use and examine constitutive versus episomal expression approaches. Before analysing specific proteins - PFR2, IFT88, PF16, MFT. This systematic approach gives confidence in their results and the utility of the system. The primer design resource with primer effectiveness score is great to see and will aid the adoption of this approach.

    Line 482 - the authors wrote 'As expected, the proportion of cells showing a motility phenotype in the IFT88 targeted L. infantum population decreased further' Why is this result expected? Presumably, this is due to the fact that cells without a functional IFT system lack flagella and grow slower so can be outcompeted by faster-growing mutants. This speaks to the major caveat highlighted by the authors in the discussion and the final small-scale screen. In a population of cells, those with deleterious mutations in an essential gene or one whose disruption results in slower growth will be outcompeted by cells in which a non-deleterious mutation has occurred, which feeds into the issue of timing.

    The authors show with CRK3 this process of non-deleterious mutants outcompeting deleterious mutants does result in a detectable drop in the number of parasites with specific CRK3 guides but not in those with IFT88. Is this due to the fact that the outgrowth of the non-deleterious IFT88 mutants occurs rapidly or that the mutation of the targets in IFT88 was ineffective? The data presented in Figure 5 shows that for some species at least a mutation of the IFT88 gene was possible. This might mean that for certain genes the outgrowth occurs within the first 12 days after transfections so will not be seen using this approach, without a wider study, which is beyond the scope of this manuscript it will be difficult to know.

    The ability to readily generate cells resistant to miltefosine, highlight the strength of this approach in identifying the mode of actions/resistance mechanisms for anti-leishmanial drugs. Moreover, any screens using this base editing approach, in which cells expressing proteins without a changed functionality/expression are killed will likely be effective in identifying genes of interest. This could mirror the success that the genome-wide RNAi screens have had in Trypanosoma brucei.

    This base editing approach now sits alongside using CRISPR/Cas9 to generate full gene deletion mutants and RNAi to help understand gene function in Leishmania. As discussed by the authors in their balanced discussion there are merits. A major advantage of this approach is the ability to simply generate a library of plasmids that will target the entire genome, whereas both full gene deletions and RNAi in L. braziliensis are more time-consuming and the latter lacks inducible control. However, as part of the LeishGEM project pools of barcoded deletion mutants are being generated, which have the potential to be used in other screens. Moreover, this base-editing approach has the potential to identify the function of essential genes, which is not possible when trying to generate stable deletion cell lines. However, this has only been demonstrated for one gene to date and the ability to detect slower-growing mutants varied greatly between different species.

    The authors highlight that this base editing approach will leave potentially functional regions of the NT of proteins, which is true and may mean genes are missed. However, this may also provide extra information about the protein's function/domain structure if STOP codons in certain positions showed an effect on function whereas those in others don't.

    Overall, the base editing approach in this manuscript looks to have great utility and in reality, is a complementary approach to the genetic tools we already have to study gene function in Leishmania. However, only time will tell how effective this method is through its adoption and effective use by different researchers.

  6. Version published to 10.1101/2022.12.08.519658v1 on bioRxiv