Abstract
The Clustered Regularly Interspaced Short Palindromic Repeat/CRISPR associated (CRISPR/Cas) technology allows rapid, site-specific genome modification in a wide variety of organisms . Proof-of-principle studies in Drosophila melanogaster have used various CRISPR/Cas tools and experimental designs, leading to significant uncertainty in the community about how to put this technology into practice. Moreover, it is unclear what proportion of genomic target sites can be modified with high efficiency. Here, we address these issues by systematically evaluating available CRISPR/Cas reagents and methods in Drosophila. Our findings allow evidence-based choices of Cas9 sources and strategies for generating knock-in alleles. We perform gene editing at a large number of target sites using a highly active Cas9 line and a collection of transgenic gRNA strains. The vast majority of target sites can be mutated with remarkable efficiency using these tools. We contrast our method to recently developed autonomous gene drive technology for somatic and germline genome engineering and conclude that optimized CRISPR with independent transgenes is as efficient, more versatile, and does not represent a biosafety risk.
The Clustered Regularly Interspaced Short Palindromic Repeat/CRISPR associated (CRISPR/Cas) genome engineering system is currently revolutionizing experimental and applied biology (Hsu et al. 2014; Doudna and Charpentier 2014). The system consists of the bacterial endonuclease Cas9 and a small chimeric guide RNA (gRNA), which directs Cas9 to a genomic target site 5′ to an NGG protospacer adjacent motif (PAM) (Jinek et al. 2012). Repair of CRISPR/Cas-mediated DNA double-strand breaks by endogenous proteins can result in insertions and deletions (indels) that disrupt gene function or—together with a supplied piece of donor DNA—precise sequence alterations. This methodology has been used successfully to edit the genome of a variety of organisms, although mutagenesis efficiencies have varied widely between different target sites, even in individual studies.
Several studies have demonstrated the feasibility of CRISPR/Cas genome engineering in the soma and the germline of Drosophila melanogaster, a major model organism in biomedical research (Bassett et al. 2013; Gratz et al. 2013, 2014; Sebo et al. 2013; Kondo and Ueda 2013; Ren et al. 2013, 2014; Yu et al. 2014; Port et al. 2014; Gokcezade et al. 2014; Zhang et al. 2014). These proof-of-principle studies have suggested a number of strategies to modify the fly genome, differing mostly in the way the Cas9 and gRNA are delivered. A popular method is centered on transgenic cas9 Drosophila strains, which are injected either with gRNA-encoding plasmids (Ren et al. 2013; Gratz et al. 2014) or crossed to strains with gRNA transgenes (Kondo and Ueda 2013; Port et al. 2014; Chen et al. 2015) [we now refer to this latter method as CRISPR with independent transgenes (CRISPR-it)]. These methods benefit from high success rates, easy generation of the necessary reagents, and the possibility to outsource gene targeting to commercial providers. A number of different transgenic cas9 strains have been described and are publicly available. However, because there has been no assessment of their comparative performance, there is considerable uncertainty in the field about which strains to use for which applications. Different published CRISPR/Cas strategies for introducing knock-in mutations with homology-directed repair (HDR)—a key application of this technology in Drosophila— also have not been compared systematically.
It is also not known how many genomic target sites can be effectively edited in CRISPR/Cas experiments because most previous studies have used a small number of gRNAs. gRNAs delivered by injection of RNA or a plasmid can differ substantially in their activity (Bassett et al. 2013; Lee et al. 2014; Ren et al. 2014; Zhang et al. 2014), prompting some to recommend screening for active gRNAs in cell culture before performing whole animal experiments (Bassett and Liu 2014; Zhang et al. 2014).
Gantz and Bier recently described a novel CRISPR-based method, the mutagenic chain reaction (MCR), which in principle allows highly efficient somatic and germline mutagenesis of the Drosophila genome (Gantz and Bier 2015). The authors demonstrated using the yellow (y) gene that inserting a cassette encoding Cas9 and gRNA into the endogenous target site of a gRNA will autocatalyze its integration into the homologous allele, thus converting heterozygous into homozygous cells. Autocatalytic homing endonucleases have been created previously and are commonly referred to as “gene drives” (Chan et al. 2011; Windbichler et al. 2011; Simoni et al. 2014), but Gantz and Bier’s method works particularly efficiently because of the use of CRISPR/Cas. CRISPR/Cas gene drive systems could be used potentially to spread traits within wild populations of plants and animals to address global problems in public health, sustainable agriculture, and environmental management (Burt 2003; Sinkins and Gould 2006; Esvelt et al. 2014). Gantz and Bier (2015) proposed that their MCR technology also could be used to accelerate laboratory genome engineering and reveal homozygous mutant phenotypes in genetic screens. However, there are major biosafety concerns about the use of autonomous gene drive technology in the laboratory because of the risk of accidental infiltration of autocatalytic alleles into wild populations (Esvelt et al. 2014; Oye et al. 2014).
Here we describe the systematic evaluation of Drosophila CRISPR/Cas tools and experimental designs that do not create gene drives. Our findings allow evidence-based choices of Cas9-expressing lines and methods for generating knock-in alleles. Our work also reveals that optimized CRISPR with independent, integrated Cas9 and gRNA transgenes can achieve remarkably efficient germline and somatic gene targeting at a very large proportion of genomic target sites. The performance of these methods is comparable with that reported for MCR technology. We conclude that transgenic CRISPR/Cas is a safe and consistently efficient method for somatic and germline modification of the fly genome and encourage others to apply similar technology to other experimental model organisms.
Materials and Methods
Plasmid construction
All primer sequences [purchased from Integrated DNA Technologies (IDT)] are listed in the Supporting Information, File S1. Unless stated otherwise, enzymatic reactions were performed according to the manufacturers’ guidelines. To generate gRNA expression vectors, pCFD3 (Port et al. 2014) (Addgene 49410) or pDCC6 (Gokcezade et al. 2014) (Addgene 59985) were cut with BbsI and dephosphorylated with alkaline phosphatase, followed by gel purification of the linear plasmid. To introduce the target-specific spacer sequence, two oligonucleotides containing the spacer sequence and reverse complement spacer sequence, as well as appropriate overhangs, were mixed [1 μL of each oligo (100 μM stock solutions)] together with 1 μL of 10× T4 Ligation Buffer [New England Biolabs (NEB)], 6.5 μL of dH2O and 0.5 μL of T4 polynucleotide kinase (NEB). Phosphorylation and annealing of oligos was performed in a thermo cycler (30 min at 37°, 5 min at 95°, followed by ramping down to 25° at 5°/min). Annealed oligos were diluted 1:200 in dH2O and ligated into the linear expression plasmids with T4 DNA ligase (NEB), followed by transformation into chemically competent bacteria. A step-by-step protocol is available from www.crisprflydesign.org.
Drosophila culture
Flies were maintained at 25° and 50% humidity with a 12-hr light/dark cycle.
Assessing activity of cas9 lines
To assess cas9 line performance in targeting ebony (e), virgin females from the various cas9 lines were mated to U6:3-gRNA-e transgenic males. The resulting double transgenic offspring were examined for CRISPR/Cas-induced somatic phenotypes, with germline mutagenesis assessed by mating randomly selected virgin females to e mutant males (w;;TM3/TM6b). The number of progeny from these crosses with ebony pigmentation was recorded. At least five independent crosses were analyzed for each cas9 line. For these and other experiments, information on sample sizes is given in the appropriate figure legend.
To analyze the ability of the various cas9 lines to mediate mutagenesis of wntless (wls), virgin cas9 females were crossed to U6:3-gRNA-wls transgenic males. At least three independent crosses were performed for each genotype. In cases in which the resulting cas9 U6:3-gRNA-wls animals had significant rates of pupal lethality, the number of dead vs. total pupae was recorded. Males from the remaining, viable cas9 U6:3-gRNA-wls genotypes were crossed to y w hs-FLP;;MKRS/TM6b virgins. To determine the rate of germline transmission, genomic DNA was isolated from some of the offspring [using 10 μL of microLysis-Plus (Microzone) according to the manufacturer’s instructions] and used for PCR analysis. Primers wls_geno_fwd and wls_geno_rev were used to amplify the regions flanking the gRNA-wls target site. The genomic sequence from 10 to 12 flies from two independent crosses was analyzed for each genotype. To monitor embryonic viability of the offspring, virgins heterozygous for both cas9 and U6:3-gRNA-wls were crossed to wild-type males, and the cross was transferred to cages mounted on apple juice agar plates. All embryos from a 1- to 2-hr egg collection were counted and kept at 25° for 48 hr to allow embryonic development to be completed. After 48 hr, the number of embryos that hatched was recorded. Where inspected, arrested embryos had segmentation defects. This analysis was carried out in quadruplicate for each genotype. Intercrosses of act-cas9 U6:3-gRNA-e flies were used as a control.
Embryo injections and transgenesis
Microinjection into embryos was performed via standard procedures as described previously (Port et al. 2014). To generate transgenic gRNA lines, plasmids were often injected in pools (Bischof et al. 2013). Equal amounts of gRNA plasmids (5–20 per pool) were mixed and diluted to a final concentration of 150 ng/μL in dH2O and injected into y v nos-PhiC31; attP40 [Bloomington Stock (BL)25709] embryos. A small proportion of microinjections for Drosophila transgenesis were performed by the fly facility injection service of the Department of Genetics, University of Cambridge, United Kingdom. Single transgenic offspring, selected by the v+ marker encoded by the gRNA plasmid, were mated with v; Sp/CyO flies for several days and then squashed in 10 μL of microLysis-Plus (Microzone) to extract genomic DNA. To identify the integrated gRNA plasmid a region of the transgene was amplified by PCR using primers U63fwd1 and CFD4seqrev3, and the resulting PCR product was submitted for Sanger sequencing (Source Bioscience) using the former primer.
Assessing activity of the 66 gRNA transgenes
Transgenic gRNA males were crossed to act-cas9 virgin females. Crosses with gRNAs targeting essential genes were monitored daily, with the proportion of dead offspring estimated at each developmental stage. The activity of gRNAs targeting nonessential genes was monitored in adult offspring of cas9 x gRNA crosses (note that we did not notice significant lethality before adulthood in any of these crosses). Flies expressing act-cas9 and gRNAs targeting sepia (se) were analyzed 1 wk after eclosion, when the difference in eye pigmentation between wild-type and se mutant tissue was most obvious. The percentage of eye tissue that was se mutant was analyzed in at least 20 act-cas9 gRNA-se flies by three researchers blind to the genotype. The results were very similar for each researcher and the average of the three independent recordings is presented in Figure 3. Flies expressing gRNAs targeting y or e and act-cas9 were inspected visually for their respective pigmentation phenotype between 3 and 6 d after eclosion. Male flies were selected independently of the severity of the somatic pigmentation phenotype and crossed to either y or e mutant partners. The number of progeny with yellow or ebony pigmentation was recorded. Data presented in Figure 3 are the average of three independent crosses. The percentage of yellow offspring was normalized to account for the 50% of flies that were expected to be yellow without CRISPR mutagenesis due to the parental genotype.
Image acquisition and processing
Flies were anesthetized with CO2 and submerged in 90% ethanol/10% glycerol for at least 4 hr, followed by mounting on Sylgard plates (Dow Corning) in 50% ethanol/50% glycerol. Images were captured with a Canon 550D camera equipped with a Canon 24 mm f1.8 lens mounted on a stereomicroscope (Leica MZFLIII). Camera settings were entirely manual and constant illumination was used in each session. Images presented in the same figure were captured in a single session. Brightness and contrast was adjusted with Adobe Photoshop software, with identical manipulations for each image within a series.
Sequence analysis of CRISPR/Cas-induced mutations
Genomic DNA from single flies was extracted with 10 μL of microLysis-Plus. Then, 0.75 μL of DNA solution was used as a template in 25 μL of polymerase chain reactions (PCRs) using the Q5 Hot Start 2x master mix (NEB). Sequences of primers used for the individual target genes are listed in the File S1. PCR products were purified with the QIAGEN PCR purification kit and submitted to Sanger sequencing using the forward PCR primer. Sequencing chromatograms from flies that inherited a CRISPR/Cas-induced indel and are consequently heterozygous at the gRNA target site contained an overlay of sequence from both alleles. Chromatograms were analyzed with Tide [http://tide.nki.nl/ (Brinkman et al. 2014)] and re-examined manually to correct for possible mistakes by the software, which occurred occasionally for alleles involving insertions.
Assessing different HDR strategies
Primers pBSwg3′HA_fwd and pBS-wg5′HA_rev were used to amplify the 5′ and 3′ wingless (wg) homology arms and the pBluescript SK-(+) vector backbone from the previously described wg::GFP donor plasmid (Port et al. 2014). The 3xP3-RFP-αtub-3′UTR (a gift from Nick Lowe and Daniel St Johnston) sequence was amplified by PCR using primers 3xP3-RFP_fwd and 3xP3-RFP_rev. Both fragments were then joined by Gibson assembly to generate the red fluorescent protein (RFP) donor construct. Sequence-verified plasmid DNA was purified using the MinElute PCR purification kit from QIAGEN. To test HDR efficiency using transgenic cas9 and gRNA, the purified circular donor plasmid was injected into embryos resulting from crosses between nos-cas9 CFD2 or nos-cas9 TH_attP2 virgin females and U6:3-gRNA-wg males. For the protocol based on transgenic cas9 stocks, both the donor and U6:3-gRNA-wg plasmid were injected into embryos either hemizygous (CFD2 males) or homozygous (CFD2 females or TH_attP2 males and females) for the nos-cas9 transgene. To test HDR efficiency using a derivative of the pDCC6 plasmid (Gokcezade et al. 2014) encoding cas9 and U6:2-gRNA-wg (hereafter called pDCC6-wg), both donor and pDCC6-wg were injected into embryos of an isogenized w1118 stock (Bloomington Stock Centre ((BL)5905)). The concentrations of injected plasmids for the HDR experiments are given in Figure 2. Injected embryos were kept at 18° for ∼48 hr and then moved to 25°. G0 males were crossed to yw hs-FLP; Sp/CyO females, and the resulting male offspring was screened for the presence of RFP. To test whether the 3xP3-RFP cassette was inserted into the wg locus, genomic DNA was isolated from some of the RFP positive male offspring for PCR analysis as described above, with primer pairs wgHRgeno_fwd1/wgHRgeno_rev1 and wgHRgeno_fwd2/wgHRgeno_rev2 used to amplify the regions flanking each homology arm.
Results
Publicly available Cas9-expressing strains vary widely in their pattern and level of activity
We and others have shown that strains with integrated cas9 transgenes can be used to modify the Drosophila genome in somatic and germline cells (Sebo et al. 2013; Kondo and Ueda 2013; Ren et al. 2013, 2014; Gratz et al. 2014; Port et al. 2014; Zhang et al. 2014; Xue et al. 2014; Chen et al. 2015). However, there is significant uncertainty in the fly community about which of the many published cas9 strains are most effective for genome engineering experiments.
We evaluated the performance of all transgenic cas9 lines available at the time of study. The collection consisted of six lines that we assessed previously (Port et al. 2014) and eight others (Table 1). The transgenes differed in regulatory elements, cas9 codon usage, and genomic integration sites (Table 1). We crossed cas9 flies to a previously validated, ubiquitously expressed gRNA transgene (U6:3-gRNA-e), which targets the 5′ end of the coding sequence of the pigmentation gene ebony (e) (Port et al. 2014). Phenotypic assays using this gRNA typically report only on out-of-frame indels, as most in-frame mutations at the gRNA-e target site retain protein function (Port et al. 2014). All adult progeny expressing U6:3-gRNA-e and cas9 under the control of either the actin5c (act) or vasa promoter had a large proportion of cuticle that was dark, demonstrating efficient biallelic disruption of e in somatic cells (Figure 1A). In contrast, seven of the nine lines tested that express cas9 under nanos (nos) regulatory elements did not have detectable somatic activity when crossed to U6:3-gRNA-e [Figure 1A; the other two lines (CFD3_nos and CFD8) resulted in small somatic mutant clones as previously reported (Port et al. 2014)]. Collectively, these findings confirm and extend our previous findings on differential somatic activity of Cas9 with these classes of regulatory elements (Port et al. 2014).
Large variation in somatic and germline activity of publicly available transgenic cas9 strains. (A) Representative examples of female flies expressing one copy of the different cas9 transgenes and one copy of U6:3-gRNA-e transgene (>100 adults of each genotype were examined). Darker body coloration indicates biallelic disruption of e in epidermal cells (arrowheads: sporadic biallelic targeting of e). (B) Assessment of germline transmission of nonfunctional e alleles from animals expressing different cas9 transgenes and the U6:3-gRNA-e transgene. In (B–D), dots show data from individual crosses with mean ± SD represented by the underlying bar chart. In (B), typically between 30 and 140 progeny were analyzed for each cross. (C) Frequency of pupal lethality caused by cas9 transgenes with somatic activity in combination with the U6:3-gRNA-wls transgene. Flies that eclosed from the pupal case had leg and wing defects consistent with reduced wls function and died shortly afterward. In (C) and (D), control flies were of the genotype act-cas9 U6:3-gRNA-e, which have widespread induction of double-strand breaks. Typically, 50–150 animals were examined for each cross. (D) Percentage of nonviable embryos laid by females expressing different germline restricted nos-cas9 transgenes and the U6:3-gRNA-wls transgene following mating to wild-type males. Typically, 100 – 300 embryos were examined for each cross. (E) Germline transmission rates of CRISPR/Cas-induced wls mutations. Offspring from nos-cas9 U6:3-gRNA-wls male flies crossed to wild-type females were genotyped by polymerase chain reaction of a region of the wls locus containing the target site, followed by sequencing. Each square represents an individual genotyped fly.
We next tested to what extent each cas9 line could induce heritable mutations in the germline. cas9 U6:3-gRNA-e flies were crossed to e mutants, allowing transmission of CRISPR/Cas-induced e loss-of-function alleles to be evaluated by the pigmentation of the offspring. Mean germline transmission rates of nonfunctional e alleles varied widely for the different cas9 lines, with values between 0.5 and 49% (Figure 1B). act-cas9, vasa-cas9 BL51323, and nos-cas9 TH00787.N (which we refer to as TH_attP2) gave rise to similar, high rates of mutagenesis.
We also evaluated the cas9 lines with a second gRNA transgene (U6:3-gRNA-wls) targeting the essential gene wntless (wls) (Port et al. 2014). As expected, cas9 transgenic strains that had somatic activity in combination with U6:3-gRNA-e resulted in nonviable progeny when crossed to U6:3-gRNA-wls (Figure 1C). The progeny of the other seven nos-cas9 lines were viable in combination with U6:3-gRNA-wls, although a small proportion of adults from these crosses had wing defects indicative of gene targeting in a small subset of somatic cells (Figure S1). Targeting of wls in the germline was assessed by crossing nos-cas9 U6:3-gRNA-wls females to wild-type males. All but two of the nos-cas9 transgenes resulted in a high proportion of embryos arresting with defective segmentation (Figure 1D), a phenotype associated with biallelic disruption of wls in the female germline (Bänziger et al. 2006; Bartscherer et al. 2006). Thus, the majority of nos-cas9 lines can be used to assess the embryonic phenotype associated with disruption of an essential gene in the female germline.
All crosses of nos-cas9 U6:3-gRNA-wls females to wild-type males gave rise to some viable offspring, with sequencing of the wls locus from these flies revealing substantial variation in the frequency of germline transmission of CRISPR/Cas-induced mutations (Figure 1E). In two cases, CFD2 and TH_attP2, all analyzed offspring (10/10) had an indel at the wls target site. TH_attP40 also resulted in very efficient targeting, with 9/11 progeny inheriting a modified wls allele. Many inherited wls mutations were out-of-frame, presumably disrupting gene function. For females expressing U6:3-gRNA-wls and the nos-cas9 CAS series of lines there was not a strong correlation between the rate of embryonic lethality of the progeny (Figure 1D) and the rate of germline transmission of mutated wls alleles (Figure 1E). Germline transmission depends on a mutation of wls in the transcriptionally quiescent oocyte nucleus, whereas embryonic lethality arises from mutagenesis of the gene in the auxiliary nurse cells, which supply protein products to the egg through cytoplasmic bridges. These nos-cas9 lines may therefore be differentially active in targeting wls in the nurse cells and oocyte within the female germline.
Together our results reveal that the cas9 lines differ substantially in the pattern and level of their activity. Whereas act-cas9 and vasa-cas9 lines can directly reveal null mutant phenotypes in the soma when combined with gRNA transgenes, a subset of nos-cas9 lines (e.g., CFD2, TH00788.N (referred to as TH_attP40) and TH_attP2) are ideally suited to efficiently generate indel mutations in both essential and nonessential genes in the germline.
Assessment of methods for generating knock-in alleles using CRISPR/Cas
Another important application of CRISPR/Cas is the precise modification of the genome by HDR. This involves Cas9-mediated induction of double-strand breaks at the target site in the presence of exogenous donor DNA. Three experimental strategies are particularly appealing in Drosophila because of their relative simplicity. Embryos that are transgenic for both cas9 and gRNA can be injected with donor DNA (Port et al. 2014), transgenic cas9 embryos can be injected with a mixture of donor DNA and gRNA-encoding plasmid (Gratz et al. 2014; Ren et al. 2014; Zhang et al. 2014) or nontransgenic embryos can be injected with a donor plasmid together with a plasmid encoding Cas9 and gRNA (Gokcezade et al. 2014). No study has directly compared the efficiency with which all three methods facilitate HDR with the same gRNA and donor construct.
To address this issue, we constructed a donor plasmid designed to knock-in a cassette expressing RFP under an eye specific promoter into the essential wg gene (Figure S2A). The plasmid, which contained wg homology arms of 1.4 and 1.7 kb, was used in circular form in combination with a previously validated gRNA [gRNA-wg (Port et al. 2014)]. The 3xP3-RFP cassette allows rapid identification of integration events by screening adults for red fluorescent eyes (Figure S2B). For technical reasons (see the legend of Figure 2) we only followed integration of donor DNA using male flies.
Assessing strategies for generating knock-in alleles with CRISPR/Cas-mediated HDR. (A) nos-cas9 female flies of the indicated strain were crossed to U6:3-gRNA-wg males and a donor plasmid encoding RFP was injected into the embryonic progeny at the indicated concentrations. Due to a pre-existing X-linked RFP insertion in the CFD2 stock only male G0 and F1 flies could be analyzed. For consistency we only analyzed male flies in all other HDR experiments. G0 males that give rise to RFP-positive offspring were designated “founders.” The percentage of male progeny with RFP expression relative to all offspring (i.e., from fertile founder and nonfounder G0 males) is indicated below. N, total number of males analyzed. (B) Donor and U6:3-gRNA-wg plasmids were injected into nos-cas9 embryos. In (B) and (C), plasmids were mixed to give an injection solution containing the concentrations shown. (C) Nontransgenic w1118 embryos were injected with donor DNA and a single plasmid containing both hsp70-cas9 and U6:2-gRNA-wg. The attention sign indicates that unintended integration of this plasmid at the gRNA target site could create an autonomous gene drive. Propagation of a gene drive is not possible in our experiment as integration at the gRNA-wg target site would create a lethal allele. Table S1 contains detailed results for all HDR experiments.
For the experiments involving transgenic cas9 supply we used different plasmid concentrations and two different nos-cas9 strains (CFD2 and TH_attP2; Figure 2, A and B). Injection of the donor plasmid into cas9 gRNA-wg double transgenic embryos resulted in 19–25% of all offspring from G0 flies having integration of the RFP construct, compared with 5–15% when the U6:3-gRNA-wg plasmid was injected with donor DNA into nos-cas9 embryos (Figure 2, A and B and Table S1).
We assessed the efficiency of plasmid-based delivery of Cas9 and gRNA using a published vector that encodes both components (Gokcezade et al. 2014). Although expected to be a very rare event, accidental integration of such a plasmid at the gRNA target site could create an autonomous gene drive. We were able to safely test this plasmid as insertion at the gRNA-wg target site would be homozygous lethal, preventing the allele from being propagated. After coinjection of the dual cas9/gRNA plasmid and donor plasmid into nontransgenic embryos only 3% of all progeny from G0 flies had genomic integration of the RFP cassette (Figure 2C and Table S1). Integration of the donor DNA within the wg locus could be confirmed for 87 of 98 RFP-positive flies tested from our entire set of HDR experiments (Figure S2, C and D and Table S1). In summary, the methods in which a transgenic cas9 source was used resulted in the greatest frequency of knock-in alleles, with injection into embryos that are also transgenic for the gRNA giving the most efficient targeting.
CRISPR-it with a highly active Cas9 line results in remarkably efficient somatic and germline mutagenesis at a large proportion of target sites
The aforementioned results, together with previous literature (Kondo and Ueda 2013; Port et al. 2014), provide evidence that transgenic supply of both Cas9 and gRNA can lead to very high rates of mutagenesis. An important outstanding question is whether the high rates of efficiency observed for a small number of genomic target sites in these studies are a general feature of this system.
To determine what fraction of gRNAs is functional in the CRISPR-it system we generated a set of 66 transgenic fly strains each expressing a different gRNA. The gRNAs were designed to target seven genes, including essential genes [wingless (wg), wls, Lissencephaly-1 (Lis1), Dynein heavy chain at 64C (Dhc64C)] and nonessential pigmentation genes [e, y; also controlling coloration of the cuticle, and sepia (se; controlling coloration of the eye)]. gRNA target sites were based on the reference sequence of the Drosophila genome (Flybase release 6.02). They were located in the 5′ half of the coding sequence of each gene, such that out-of-frame indels are likely to create loss-of-function alleles, but were otherwise selected at random. gRNAs were expressed under the control of the strong, ubiquitous U6:3 promoter (Port et al., 2014), with all constructs integrated at the same genomic location (attP40 on chromosome 2L) to ensure comparable expression.
The 66 individual gRNA strains were crossed to the X-linked act-cas9 line, which has ubiquitous, high activity (Figure 1, A and B). Strikingly, analysis of the progeny revealed that all but one of the gRNAs gave rise to a somatic mutant phenotype (Figure 3 and Table S2). Sequencing data from our strains revealed a single-nucleotide polymorphism in the target site of the inactive gRNA, designed to target se, providing a likely explanation for its inactivity. Most of the gRNAs targeting the essential genes caused lethality at the same developmental stage as reported for the respective homozygous null-mutant animals, with the remainder leading to developmental arrest at later stages (Figure 3). With the exception of the single inactive gRNA-se transgene, all of the gRNAs targeting nonessential genes gave adults with regions of tissue with the null-mutant pigmentation phenotype (Figure 3, Figure S3, and Table S2). For almost all of these gRNAs, biallelic gene disruption was usually observed in more than 50% of the affected tissue, with half of the gRNAs, including all five targeting y, leading to more than 80% mutant tissue (Table S2). Nonspecific phenotypes were not observed in act-cas9 adults coexpressing any of the gRNA transgenes, consistent with previous evidence that CRISPR/Cas operates with substantial fidelity in Drosophila (Bassett et al. 2013; Gratz et al. 2014; Ren et al. 2014). The aforementioned results demonstrate that the vast majority of randomly selected gRNAs efficiently induce biallelic loss-of-function mutations in target genes, thereby directly revealing the somatic mutant phenotype. There was, however, variation in the strength of the phenotypes observed for gRNAs targeting the same gene (Figure 3 and Table S2). Weaker phenotypes could arise either through a reduced frequency of induced indel mutations or because functional in-frame mutations at the specific gRNA target site provide a significant fraction of cells with functional protein.
Almost all transgenic gRNAs are highly active with a strong transgenic source of Cas9. Male flies transgenic for individual gRNAs were crossed to act-cas9 females. Somatic phenotypes of act-cas9 gRNA progeny are summarized for gRNAs targeting Dhc64C, Lis1, wg, wls, and se. The gRNA-se with no detectable activity was found to contain a polymorphism. Activity in the germline is shown for gRNAs targeting y and e [data show mean percentage of progeny that inherited a nonfunctional allele from three independent crosses (see Table S2 for more details and somatic phenotypes induced by these gRNAs)]. gRNAs highlighted by black diamonds frequently give rise to functional in-frame mutations (Table S2), suggesting that their efficiency in creating indel mutations often approaches 100%. The gRNA additionally highlighted by an asterisk gives rise to an unexpected large number of in-frame mutations in functional alleles analyzed (8/8 flies), suggesting a micro-homology-mediated bias in nonhomologous end joining. Since transcription initiation from U6 promoters is proposed to require a G nucleotide, we extended gRNAs by a mismatched G where necessary (lower case). gRNAs with spacer sequences of 18 nt, 19 nt, and 20 nt can support efficient mutagenesis, consistent with previous evidence that short truncations of spacers usually do not strongly affect activity of gRNAs (Fu et al. 2014; Ren et al. 2014).
To directly test the hypothesis that in-frame mutations can mask high rates of indel induction by some gRNAs, we characterized CRISPR/Cas-induced mutations transmitted to the next generation. Flies expressing act-cas9 and a transgenic gRNA targeting y or e were crossed to y or e mutant flies, respectively, followed by phenotypic analysis of the offspring. The rates of transmission of nonfunctional alleles across target sites was generally greater than reported when gRNAs were delivered by plasmid injection into transgenic cas9 embryos (Ren et al. 2014) (Figure 3 and Figure S4, A and B), providing further evidence of the relatively high activity of CRISPR-it. All five gRNA-y transgenes and six of 18 gRNA-e transgenes transmitted nonfunctional alleles to >93% of their offspring (Figure 3). The high frequency of nonfunctional alleles strongly suggests that in-frame mutations at these target sites usually disrupt protein function. Transmission rates for nonfunctional mutations ranged from 13 to 90% for the other 12 gRNA-e lines (Figure 3). To explore the basis of this variation, we determined the sequence of e around the target sites in some of the phenotypically wild-type progeny from the 12 lines. For nine of the gRNAs, in-frame indel mutations were detected in all, or almost all, analyzed animals (Table S2 and Figure 3). Thus these gRNAs appear to create indels in the vast majority of e alleles, with differences in the functional effects of in-frame mutations at their target sites influencing the outcome of phenotypic assays. We confirmed that this effect was not specific for target sites in e by molecular analysis of the se locus in offspring of act-cas9 gRNA-se flies, which again revealed in-frame mutations in the majority of functional alleles (Table S2 and Figure 3).
Indel mutations were not found readily in functional alleles inherited from flies expressing act-cas9 and the three gRNA-e transgenes that were the least efficient in the phenotypic assays (transmission of nonfunctional alleles in 13–35% of cases) or one poorly active se gRNA (7% of phenotypically mutant eye tissue) (Table S2 and Figure 3). Thus, these gRNAs induce indels with relatively low frequency. Sequencing of the target site for these gRNAs excluded the possibility that polymorphisms were responsible for their reduced activity. gRNAs with lower activity were not readily predicted by available online gRNA design tools and their target sites did not have a low GC content in the region proximal to the PAM (Figure S4, C–J and Table S3), a parameter that strongly predicted less active gRNAs encoded by plasmids injected into transgenic cas9 embryos (Ren et al. 2014). Several gRNAs targeting sites with relatively low PAM proximal GC content had very high activity in our experiments (Figure S4, A, F, and J). Collectively, these results demonstrate that optimized CRISPR-it is a very robust genome engineering system, leading to highly effective mutagenesis at the vast majority of target sites.
Discussion
We set out to provide a systematic evaluation of tools and design parameters for transgenic CRISPR/Cas genome engineering in Drosophila melanogaster. We show that selection of the cas9 strain is critical for the successful application of this methodology, as publicly available lines vary widely in their activity in somatic and germline cells. act-cas9 and all available vasa-cas9 lines can be used with transgenic gRNAs to reveal null mutant phenotypes in somatic tissues. However, somatic expression of Cas9 in these lines means that they are not suitable for germline transmission of mutations in essential genes using gRNA transgenes. nos-cas9 TH_attP2, CFD2, and TH_attP40 are highly effective for germline mutagenesis of both the nonessential gene e and the essential gene wls, suggesting that they are the most versatile transgenic Cas9 sources for creating heritable indel mutations with transgenic gRNAs. These three nos-cas9 transgenes are inserted on different chromosomes (Table 1) and the choice between them will depend on the specific crossing scheme of the user. CFD2 is often used in our laboratory because its location on the X chromosome means that it can easily be selected against in subsequent generations. Our work defines for the first time nos-cas9 transgenic lines that have relatively low activity. Although these lines are less suitable for most genome engineering experiments, they may be useful when widespread biallelic targeting in the germline is problematic, such as when the target gene is essential for cell survival.
We find that success rates vary substantially between different CRISPR/Cas-based protocols for generating knock-in alleles. High efficiency is particularly important for HDR without selectable markers (which introduce unwanted sequences into the target locus and often require additional cloning steps) as in most cases integration must be detected by PCR-based analysis of genomic DNA. The rates of HDR that we and others (Yu et al. 2014; Gratz et al. 2014; Ren et al. 2014; Zhang et al. 2014) have reported when injecting donor DNA and gRNA plasmids into cas9 transgenic embryos are sufficient to make PCR-based screens practical. Because this method is also relatively quick and can be easily be performed by commercial injection services, it is likely to be the method of choice for most applications. However, the greater efficiency achieved by injection of donor DNA into embryos transgenic for both cas9 and the gRNA will be attractive to users who want to maximize their chances of success and minimize the amount of work that has to be spent screening for the desired insertion. It should be noted that, once generated, the transgenic gRNA strain can also be used in other Cas9-based applications, including phenotypic analysis in somatic cells either in a stochastic or tissue-specific manner (Port et al. 2014; Xue et al. 2014).
We show that use of transgenic gRNAs with a highly active Cas9-line allows highly efficient mutagenesis across a large proportion of target sites, with our analysis of germline transmission of mutations in pigmentation genes suggesting that the vast majority of gRNAs cause close to maximal rates of indel induction. Incomplete penetrance of phenotypes induced by some transgenic gRNAs is usually due to functional in-frame mutations. This problem can be circumvented in cases in which previous knowledge highlights codons that are likely to be crucial for protein function. A more general approach is to coexpress more than one gRNA using a dual gRNA vector (e.g., Port et al. 2014). This will increases the probability of creating a nonfunctional mutant allele, in some cases by creating deletions between the gRNA target sites (Gratz et al. 2013; Kondo and Ueda 2013).
Importantly, our study highlights that CRISPR-it is a safe, efficient, and robust alternative to the recently reported MCR, which uses an autonomous gene drive for functional studies (Gantz and Bier 2015) (Table 2). MCR has so far only been demonstrated at a single target site in the Drosophila y gene, a locus often used in pioneer studies and which appears to have an unusually high susceptibility for gene targeting (Rong and Golic 2000; Gratz et al. 2013). MCR can rapidly reveal recessive somatic phenotypes as heterozygous mutations are efficiently converted to the homozygous state. We show here using many target sites, including five in y, that CRISPR-it can induce biallelic gene disruption in the soma with similar efficiency to that reported for Gantz and Bier’s MCR experiment (Figure 3 and Table S2). Because CRISPR-it uses separate integrated cas9 and gRNA transgenes, mutagenesis is inactivated by breeding. This is not the case for the MCR system due to linkage of the cas9 and gRNA sequences at the target locus. This means that escape of flies from the laboratory could conceivably result in rapid spread of an MCR allele in the wild, with unpredictable ecological consequences. In fact, a recent study that theoretically modeled the spread of MCR alleles under different parameters (Unckless et al. 2015) concluded that “there are conditions in which accidental introductions of a single individual can lead to fixation of the MCR allele even with significant fitness consequences to the individual.” A risk of creating autocatalytic alleles is in principle also associated with other plasmids encoding both Cas9 and gRNA (e.g., Gokcezade et al. 2014), although the probability of insertion at the gRNA target site is much lower due to the absence of homology arms.
Working with flies containing an MCR cassette has to be performed using very strict biosafety procedures (Table 2) (Gantz and Bier 2015). In contrast, work with independent cas9 and gRNA transgenes can be performed in a regular fly room suitable for work with standard transgenic animals; this means that experiments can be performed much more conveniently and rapidly. Furthermore, CRISPR-it can readily reveal mutant phenotypes of essential genes and can be employed in a tissue specific manner (Port et al. 2014; Xue et al. 2014), applications that are not easy to achieve with MCR. The ability of CRISPR-it to efficiently reveal recessive mutant phenotypes associated with a large proportion of genomic target sites also makes it a highly attractive system for large scale F1 mutagenesis screens.
Gene drive technology has great promise for applications such as ecosystem management and pest control and may well be adopted for basic research applications (Burt 2003; Sinkins and Gould 2006; Esvelt et al. 2014). However, responsible use of this technology will require development of robust molecular containment strategies (Esvelt et al. 2014). We therefore urge researchers not to apply gene drives without these safeguards in place. Our experiments illustrate safe and effective genome engineering strategies that are already available for basic research and can be applied to any genetically tractable organism.
Acknowledgments
We thank Sean Munro for encouragement and support; Peter Duchek, Nick Lowe, and Daniel St Johnston for materials; Kevin Esvelt for discussions on gene drives; Neil Grant for help with photography; Mohammad Mofatteh for help with scoring eye phenotypes; users of www.crisprflydesign.org for feedback; and the Drosophila CRISPR community for freely sharing stocks and reagents. This study was supported by a Marie-Curie IntraEuropean Fellowship (to F.P.), a Wellcome Trust Strategic Award (095927/B/11/A) (N.M. as part of a collaboration with Sean Munro) and a UK Medical Research Council Project U105178790 (to S.B.).
F.P. and S.B. are inventors of CFD fly strains that have been licensed by the MRC to commercial providers of Drosophila injection services.
F.P. and S.L.B. conceived the study; F.P., N.M. and S.L.B. designed experiments; F.P. and N.M. performed experiments; F.P., N.M. and S.L.B. analyzed data; F.P. and S.L.B. wrote the manuscript; N.M. edited the manuscript. F.P. and N.M. contributed equally to experimental work. F.P. and S.L.B. contributed equally to study design and writing the manuscript.
Footnotes
Supporting information is available online at www.g3journal.org/lookup/suppl/doi:10.1534/g3.115.019083/-/DC1
Communicating editor: B. J. Andrews
- Received April 16, 2015.
- Accepted May 16, 2015.
- Copyright © 2015 Port et al.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
