|Home | About | Journals | Submit | Contact Us | Français|
Study of the nematode Caenorhabditis elegans has provided important insights in a wide range of fields in biology. The ability to precisely modify genomes is critical to fully realize the utility of model organisms. Here, we report a method to edit the C. elegans genome using the Clustered Regularly Interspersed Short Palindromic Repeats (CRISPR) RNA-guided Cas9 nuclease followed by homologous recombination. We demonstrate that Cas9 is able to induce DNA double-strand breaks with specificity for targeted sites, and that these breaks can be efficiently repaired by homologous recombination. By supplying engineered homologous repair templates, we generated GFP knock-ins and targeted mutations. Together, our results outline a flexible methodology to produce essentially any desired modification in the C. elegans genome quickly and at low cost. This technology is an important addition to the array of genetic techniques already available in this experimentally tractable model organism.
The ability to precisely modify the genome of an organism by adding, deleting or mutating genes is a critical tool for experimental biology. The type II CRISPR-Cas system is a powerful tool for genome editing in a variety of experimental systems. The Cas9 nuclease and two small non-coding RNAs comprise an adaptive immune system in prokaryotes1. A chimeric fusion of the two RNAs, termed a single guide RNA (sgRNA), supports site-specific cleavage of target DNA by Cas9, with target specificity determined by base pairing between the 5’ end of the sgRNA and the target DNA (Fig. 1a)1. The only specific sequence requirement for cleavage is an NGG nucleotide sequence (the “protospacer adjacent motif” or PAM) at the 3’ of the target DNA sequence (Fig. 1a). By changing the targeting sequence at the 5’ end of the sgRNA, an exceptionally broad variety of DNA substrates can be targeted without the need to re-engineer the Cas9 nuclease1-13. Compared to Zinc Finger Nucleases (ZFNs) and TAL-Effector Nucleases (TALENs) that can be used to produce double-strand breaks14, the CRISPR-Cas9 system is substantially less expensive and is easier to program for new target sites. Cas9 has been used to produce targeted insertion or deletion (indel) mutations, which are generated via error-prone repair mechanisms, in a wide range of species2-13. In addition, homologous repair of Cas9-induced double-strand breaks has been demonstrated in bacteria7, yeast3, cultured human and mouse cells5,11, Drosophila melanogaster4, zebrafish10 and mice13.
The nematode Caenorhabditis elegans is a valuable and widely used experimental system due to its rapid growth, ease of handling and transparency (which facilitates microscopy). Recent reports demonstrated that Cas9 can induce double-strand breaks in the C. elegans germline, which led to mutations via error-prone repair mechanisms8,15. Here, we demonstrate that Cas9-induced double-strand breaks can be repaired efficiently by homologous recombination. By supplying engineered homologous repair templates, we generated in-frame GFP insertions and targeted mutations. We refer to this method as Cas9-Triggered Homologous Recombination.
To establish Cas9 as a tool for genome editing in C. elegans, we expressed Cas9 and sgRNA in the C. elegans germline. In mammalian systems, sgRNAs were expressed from a U6 small nuclear RNA promoter, which drives transcription by RNA polymerase III (PolIII)5,11. To our knowledge, no germline PolIII promoters had been described in C. elegans when we initiated these experiments. Aligning ten C. elegans U6 genes revealed a region of 200–300 bp upstream of the transcriptional start site that is partially conserved and may function as the promoter (Fig. 1b and Supplementary Fig. 1). We therefore expressed sgRNAs using 350 bp upstream of the highly expressed R07E5.16 U6 gene16. A different U6 promoter was independently identified by Friedland and colleagues8. To express Cas9 in the germline, we used the eft-3 promoter and tbb-2 3’UTR, which have been used successfully for genome modifications employing the Mos1 transposon17. We built a Cas9-sgRNA plasmid containing both Peft-3::Cas9::tbb-2 3’UTR and PU6::sgRNA (Fig. 1c). This plasmid, available through Addgene, can be engineered to target any desired sequence by using site-directed mutagenesis to insert the appropriate targeting sequence.
Several existing methods for modifying the C. elegans genome rely on homologous repair of double-strand breaks generated by excision of a Mos1 transposon18-21, but these techniques are limited by the relative scarcity of Mos1 insertion sites in the genome22. In principle, a much wider range of genome modifications could be made by using Cas9 to generate double-strand breaks. We therefore tested whether Cas9-induced double-strand breaks could be repaired by homologous recombination in C. elegans. We designed an sgRNA targeting a sequence adjacent to the ttTi5605 Mos1 insertion site on chromosome II, and compared the efficiency of single-copy transgene insertion into this site using either Mos1-mediated single-copy insertion (MosSCI20) or Cas9-triggered homologous recombination (Fig. 2a). The efficiency of both approaches varied between individual experiments, but the overall efficiency of the two methods was similar (Fig. 2b and Supplementary Table 1). We examined single-copy GFP transgenes under the control of the mex-5 promoter and tbb-2 3’UTR. The transgenes were expressed in the germline, and the pattern of expression was indistinguishable regardless of the method used for transgene insertion (Fig. 2c). These data demonstrate that Cas9-induced double-strand breaks can stimulate homologous recombination in the C. elegans germline.
In C. elegans, fluorescent fusion proteins are often expressed by microinjecting DNA into the gonad, generating semi-stable extrachromosomal arrays that contain many copies of the injected DNA23. Transgenes generated in this way are typically overexpressed in somatic tissues and silenced in the germline and early embryo24. Microparticle bombardment can be used to generate low-copy transgenes, which are expressed at closer to endogenous levels25,26, but this approach is expensive and time-consuming. MosSCI can be used to generate single-copy transgenes20, but for many genes, the regulatory sequences needed to recapitulate the native expression pattern are unknown. All of these approaches also leave the endogenous copy of the gene of interest intact, which makes it difficult to assess the function of the fusion protein genetically and can introduce complications in quantitative experiments because less than 100% of molecules of the protein of interest are labeled. Inserting genetically encoded tags such as GFP into endogenous genes ensures 100% labeling and expression under the control of native regulatory elements and in the normal chromatin context.
To test whether Cas9-triggered homologous recombination could be used to insert protein tags into endogenous genes, we targeted the nmy-2 gene, which encodes non-muscle myosin II. We built a homologous repair template comprising the C-terminal 1.5 kb of nmy-2 fused in-frame to GFP, followed by the nmy-2 3’UTR, an unc-119(+) selectable marker, and 1.5 kb of downstream genomic sequence (Fig. 3a). The unc-119(+) gene was flanked by LoxP sites, allowing it to be removed by subsequent expression of Cre recombinase. We also generated a Cas9-sgRNA plasmid targeted to cleave the 3’ end of nmy-2. Co-injection of the Cas9-sgRNA plasmid and homologous repair template into unc-119 worms resulted in integration of GFP and unc-119(+) into the 3’ end of the nmy-2 locus (three independent knock-in alleles from 60 total injected animals). We confirmed the correct integration of GFP at the 3’ end of the nmy-2 gene in all three lines by PCR (Fig. 3b) and sequencing.
We examined the expression and localization of the NMY-2–GFP fusion protein in these three homozygous knock-in lines. For comparison, we analyzed a strain carrying zuIs45, a well-established transgene generated by microparticle bombardment27. The knock-in strains expressed NMY-2–GFP at levels similar to those of endogenous NMY-2, whereas the zuIs45 strain expressed NMY-2–GFP at lower levels (Fig. 3d). The pattern of localization of NMY-2–GFP was indistinguishable in early embryos homozygous for zuIs45 or an nmy-2::gfp knock-in allele (Fig. 3e and Supplementary Video 1), but the knock-in embryos showed consistently brighter fluorescence, consistent with the higher expression of NMY-2–GFP in the knock-ins.
Since nmy-2 is an essential gene28, we tested whether GFP insertion disrupted protein function by assaying for lethality. Two of three nmy-2::gfp knock-in strains were 100% homozygous viable, and a third strain was 99% viable (n>100 embryos for each strain). Animals of all three knock-in strains displayed wild-type movement and had no discernable phenotypes. We conclude that insertion of GFP at the endogenous locus does not affect nmy-2 gene function.
For some applications the insertion of the unc-119(+) selectable marker into the genome may be problematic. We therefore developed a simple procedure to remove the unc-119(+) marker by injecting a plasmid encoding Cre recombinase under the control of the eft-3 promoter and tbb-2 3’UTR (see Methods) and picking uncoordinated (Unc) worms from the F2 progeny (Fig. 3c). We isolated Unc animals in 5/5 independent experiments (12–20 animals injected in each experiment), suggesting that Cre-mediated excision of unc-119(+) is highly efficient. We then outcrossed to wild-type worms to remove the unlinked unc-119 mutation required for unc-119(+) selection. This resulted in a strain containing no known genomic modifications except for insertion of GFP into the nmy-2 gene and a 23 bp LoxP site in the intergenic region downstream of nmy-2 (Fig. 3a). NMY-2– GFP expression in these animals was not altered by removal of the unc-119(+) selection marker (data not shown), suggesting that this marker does not affect fluorescent fusion protein expression when inserted into an intergenic region.
To test whether this strategy is likely to be broadly applicable, we used a similar approach to endogenously GFP-tag the his-72 gene, which encodes a Histone H3. We obtained one knock-in strain from ten successfully injected animals. We were able to amplify the left and right insertion junctions, confirming insertion of GFP into the his-72 locus (Fig. 4a). However, we were unable to amplify across the insertion, and a more detailed PCR characterization showed that a rearrangement involving a duplication of the unc-119(+) cassette had occurred (data not shown). Such rearrangements have been reported to occur in other studies involving homologous recombination20,29, and occurred at a low frequency in our study (1/16 strains generated by Cas9-triggered homologous recombination, 1/7 MosSCI strains). Importantly, his-72 mRNA levels in the his-72::gfp knock in-strain were indistinguishable from wild type, indicating that the rearrangement did not affect the his-72 gene itself (Fig. 4b). These animals were healthy and showed bright nuclear GFP fluorescence in a wide range of tissues including the germline, consistent with labeling of endogenous Histone (Fig. 4c).
Another application of Cas9-triggered homologous recombination is the generation of targeted mutations at endogenous loci. To demonstrate this, we made mutations in lin-31, which encodes a FOXB transcription factor required for vulval development30. LIN-31 forms a complex with a LIN-1 (a homolog of mammalian Ets transcription factor), and this complex is thought to repress the primary vulval fate31. MPK-1 mitogen-activated protein kinase (MAPK) can phosphorylate LIN-31 on four C-terminal threonines, and addition of active MAPK disrupts the interaction between LIN-31 and LIN-1 in vitro31. Overexpression of non-phosphorylatable LIN-31 causes a vulval phenotype31, but whether MAPK phosphorylation affects the function of endogenous LIN-31 has not been tested directly.
We used Cas9-triggered homologous recombination to mutate the four MAPK phosphorylation sites at the C-terminus of LIN-31 to either alanine (non-phosphorylatable) or glutamic acid (phospho-mimetic) (Fig. 5a–b). We used a Cas9-sgRNA plasmid targeting a site 5’ of these four residues to ensure that the Unc-119(+) animals we isolated would contain mutations at all four sites (Fig. 5b). We obtained two independent lin-31(4T→A) alleles from 60 total injected animals, and three independent lin-31(4T→E) alleles from 62 total injected animals. We confirmed the desired mutations by PCR (not shown) and sequencing (Fig. 5c).
Mutation of the four MAPK phosphorylation sites to alanine caused a distorted vulval morphology at the L4 stage and a partially penetrant protruding vulva phenotype in the adult (Fig. 5d–e and Supplementary Table 2). A similar phenotype was observed when these residues were mutated to glutamic acid, mimicking constitutive phosphorylation (Fig. 5d–e and Supplementary Table 2). Identical phenotypes were observed for two independently isolated lin-31(4T→A) lines and for three independently isolated lin-31(4T→E) lines. These results suggest that dynamic regulation of LIN-31 phosphorylation is important for normal vulval development.
Cas9 has been reported to produce off-target mutations in mammalian cells, raising concerns about the specificity of this enzyme in genome editing applications32. To assess the specificity of Cas9 in C. elegans, we identified the genomic sequences most similar to the targeting sequences that we used to modify nmy-2 and lin-31. We focused on candidate off-target sequences that closely matched the 3’ end of our targeting sequences, since Cas9 activity has been shown to be most sensitive to mismatches in the 3’ half of sgRNA targeting sequences1,5,7,32. We PCR-amplified and sequenced ten candidate off-target sites for the nmy-2 sgRNA and four for the lin-31 sgRNA (Supplementary Table 3). We found no mutations at any of these sites, in any of the strains we isolated. We note that while Fu et al. detected off-target activity of Cas9 towards sequences with up to five mismatches to the sgRNA sequence32, the closest matches that existed to our nmy-2 and lin-31 sgRNAs contained six or more mismatches each (Supplementary Table 3). The small size of the C. elegans genome compared to mammalian genomes may reduce the odds of closer off-target matches to particular target sequences.
We have demonstrated a method to efficiently induce essentially any desired modification in C. elegans. This approach was robust and cost-effective: for example, we obtained three independent nmy-2::gfp knock-in lines from a single set of injections, requiring less than four weeks total time (of which less than two days was hands-on time) and less than $200 worth of materials. In six of eight experiments, we obtained multiple independent lines on the first attempt. In the other two cases, the first set of injections failed for trivial technical reasons (see Methods), and we readily obtained homologous recombinants upon re-injection. Thus, we achieved an overall success rate of 100% without extensive re-engineering or optimization, suggesting that our strategy is likely to be broadly applicable. In addition, the ability to remove the selectable marker in a single step using Cre recombinase should facilitate using unc-119 selection, or other selectable markers, for a wide variety of genome editing strategies.
Our method relies on double-strand break repair using an engineered homologous template, similar to earlier methods that used Mos1 transposon excision to generate double-strand breaks18-21. The use of Cas9 in our system overcomes several important limitations of Mos1-based methods. First, Cas9 target sites occur once every 32 bp in random DNA sequence, and we estimate that there are over 1 million potential Cas9 target sites in the C. elegans genome, compared to approximately 14,000 Mos1 insertion sites22,33. This greatly expands the range of modifications that can be made since homologous recombination is most efficient within 0.5 kb of a double-strand break18. Indeed, none of the alleles presented in this study could have been made using Mos1-based methods because appropriate transposon insertion alleles do not exist. Second, by carefully choosing the positions of the Cas9 cleavage site and selectable marker relative to the desired modification(s), the investigator can ensure that the desired genome modification is present in every isolated recombinant (see Fig. 5b and Supplementary Protocol), in contrast to Mos1-based methods where the length of a gene conversion track is stochastic and decays rapidly with increasing distance from the transposon insertion site18. Third, Cas9 could in principle be used in any genetic background, whereas Mos1-based genome modifications must be generated in a strain carrying a Mos1 insertion. Finally, mobilization of Mos1 generates an average of 2–3 new Mos1 insertions at random sites elsewhere in the genome33,34, which could cause undesired phenotypes.
Cas9-triggered homologous recombination also has advantages over conventional transgenic approaches for generating fluorescent protein fusions. Generating and maintaining a knock-in line with our approach is cheaper and less labor-intensive than either microparticle bombardment or extrachromosomal arrays. Also, it is expected to maintain the endogenous pattern of expression in most cases, and fusion protein function can be easily assessed genetically due to the absence of untagged protein. Knock-in strategies are the standard method for generating fluorescent protein fusions in yeast for these reasons, and our methodology will allow widespread application of this approach in C. elegans.
Despite these advantages, there are some potential limitations associated with the use of Cas9-triggered homologous recombination to endogenously tag genes. First, fusion of GFP to the 3’ end of a gene may compromise protein function, though this can be detected if it causes a loss-of-function phenotype. Second, GFP coding sequences are sometimes recognized as “non-self” and silenced by the piRNA pathway35. Such a silencing event should again be detectable if it produces a loss of function phenotype and/or loss of GFP fluorescence. Third, the DNA repair mechanism involved in homologous recombination can sometimes generate complex rearrangements, as we observed for his-72::gfp20,29. Since these events are uncommon, a straightforward solution is to isolate multiple knock-in alleles for each gene of interest. We obtained multiple independent alleles from a single set of injections for seven of the eight Cas9-triggered homologous recombination experiments performed for this study (his-72::gfp was the exception). Finally, for some genes with low endogenous expression levels, fluorescence of GFP knock-ins may be too dim to visualize. Development of brighter fluorescent proteins36-38 may facilitate knock-in approaches for genes with lower endogenous expression levels. If fluorescence in a knock-in strain is too dim to be useful, then the investigator may choose to use an overexpression strategy, accepting the caveat that overexpression artifacts are possible.
In using Cas9-triggered homologous recombination to generate targeted mutations in endogenous genes, we chose to make point mutants, but we expect that insertions, deletions or other modifications could be made with similar ease, as has been done previously using Mos1 excision20,21. Of note, single point mutants have also been made in C. elegans using TALEN cleavage and single-stranded DNA oligo-mediated repair15, but this approach is limited by the length of a synthetic DNA oligo and would have required at least two sequential injection steps to produce the four lin-31 point mutations that we generated in a single step.
Interestingly, the phenotype we observed in lin-31 mutants was different from that reported by Kim and colleagues31, who found that expression of lin-31(4T→A) from an extrachromosomal array inhibited vulval fate specification. We suggest that repression of vulval fate in earlier experiments may have been due to overexpression of LIN-31 protein rather than solely its inability to be phosphorylated by MPK-1. The ability to quickly and efficiently induce mutations in endogenous genomic loci renders the use of multi-copy extrachromosomal arrays unnecessary, and should greatly simplify the interpretation of reverse genetic experiments.
In summary, we have developed a flexible, inexpensive and robust strategy for genome editing in C. elegans using Cas9 targeted cleavage and homologous recombination. Given the ease with which our approach can be adapted to new targets, we suggest that the ability to modify the C. elegans genome is now limited only by the imagination of the investigator.
New genetic nomenclature for genome editing applications has been developed by the WormBase Gene Name Curators (J. Hodgkin and T. Schedl, personal communication). Briefly, edited loci are assigned conventional allele designations, with the nature of the modification described in brackets after the allele name. For example, one of our nmy-2::gfp knock-in alleles is nmy-2(cp7[nmy-2::gfp + LoxP unc-119(+) LoxP]) I, and a lin-31 mutant lin-31(cp2[T145A T200A T218A T220A + LoxP unc-119(+) LoxP]) II. Note that each independently isolated mutant line is given its own allele designation, even if the molecular lesion is identical. When the unc-119(+) selectable marker has been removed using Cre recombinase, a new allele designation is assigned. For example, an nmy-2::gfp knock-in allele with the unc-119(+) cassette removed is designated nmy-2(cp13[nmy-2::gfp + LoxP]) I. The cp13 allele was derived from cp7 by Cre-mediated recombination.
Supplementary Table 4 lists all strains generated and used in this study. All strains were kept at 25°C and fed E. coli strain OP50 except where noted below, and were handled using standard techniques39.
Plasmids have been deposited in Addgene with the following accession numbers: Cas9-sgRNA plasmid targeting a site near ttTi5605, #47550; Cas9-sgRNA plasmid with no targeting sequence, #47549; Peft-3::Cre::tbb-2 3’UTR construct, #47551. All other plasmids used in this study are available from the authors upon request.
To construct the Cas9-sgRNA expression plasmid shown in Fig. 1c, we first designed a synthetic gene encoding Cas9, with C. elegans coding bias and synthetic C. elegans introns, using the C. elegans Codon Adapter40. Our Cas9 sequence includes a Nuclear Localization Signal and an HA tag at the C-terminus. The synthetic gene was produced as a series of overlapping 500 bp gBlocks (Integrated DNA Technologies), assembled using Gibson Assembly (New England BioLabs) and inserted into the vector pCFJ601 (Peft-3::Mos1 Transposase::tbb-2 3’UTR)17 in place of the Mos1 transposase. Next, a gBlock containing the U6 promoter and sgRNA sequence was inserted 3’ of the tbb-2 3’UTR. Genomic targets of Cas9 conform to the target sequence GN19NGG, where N is any base. The initial G is a requirement for transcription initiation by the U6 promoter, and the NGG (PAM) motif is required for Cas9 activity (note that the NGG motif must be present in the genomic target but is not included in the sgRNA sequence). To target Cas9 to different genomic sequences, we inserted the desired targeting sequence into the Cas9 + sgRNA construct using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs) with forward primer 5’-N19GTTTTAGAGCTAGAAATAGCAAGT-3’, where N19 is replaced by the desired 19 bp targeting sequence, and reverse primer 5’-CAAGACATCTCGCAATAGG-3’. Supplementary Table 5 lists the targeting sequences used in this study.
Targeting vectors for single-copy transgene insertion on chromosome II were constructed in the pCFJ150 vector backbone20 using Gateway cloning. We used site-directed mutagenesis with the Q5 site-directed mutagenesis kit (New England Biolabs) to delete a short region of the 3’ recombination arm comprising the Cas9 target sequence, to prevent the homologous repair templates from being cleaved by Cas9.
Homologous repair templates for GFP insertion and lin-31 mutagenesis were constructed in two steps. First, we PCR amplified a 3–4 kb region centered on the desired modification from N2 genomic DNA and cloned the resulting fragment into the pCR-Blunt vector using the ZeroBlunt TOPO Cloning Kit (Life Technologies). Second, we modified this genomic clone by inserting GFP (for GFP knock-ins) or a 3’ exon containing point mutations (for lin-31 mutagenesis), along with the unc-119(+) rescue gene flanked by LoxP sites. GFP and unc-119(+) fragments were generated by PCR, and LoxP sites were included in the unc-119(+) primers. The mutated lin-31 3’ exons were synthesized as gBlocks. These fragments were integrated into the genomic clones using Gibson assembly, which allows for seamless fusion of DNA fragments without the need to include any extra sequence (e.g. restriction sites). To avoid cleavage of the repair templates by Cas9, we deleted or mutated the Cas9 target site in all repair templates. Complete plasmid sequences of all targeting vectors are available from the authors upon request.
To construct the Peft-3::Cre::tbb-2 3’UTR plasmid used for removal of selectable markers with Cre recombinase, we first amplified the Cre ORF from the plasmid pEM3 (ref. 41) and cloned it into the Gateway donor vector pDONR221. We then performed a 3-fragment gateway reaction using our Cre donor vector, pCFJ386 (Peft-3; a gift from Christian Frøkjær-Jensen), pCM1.36 (tbb-2 3’UTR)42 and the destination vector pCFJ212 (ref. 17), which contains an unc-119(+) rescue gene.
Supplementary Table 6 lists all primers used in this study.
We inserted transgenes into the ttTi5605 Mos1 site by following a published MosSCI protocol17. We prepared an injection mix containing 10 ng/μL targeting vector; 50 ng/μL pCFJ601 (Peft-3::Mos1 Transposase); 10 ng/μL pMA122 (heat-shock driven PEEL-1 negative selection); 10 ng/μL pGH8 (Prab-8::mCherry neuronal co-injection marker); 5 ng/μL pCFJ104 (Pmyo-3::mCherry body wall muscle co-injection marker); and 2.5 ng/μL pCFJ90 (Pmyo-2::mCherry pharyngeal co-injection marker). The mixture was microinjected into the gonads of Unc young adults of strain EG6699 (ttTi5605 II; unc-119(ed3) III), which were raised on HB101 bacteria at 15°C. Following injection, single worms were picked to new plates and placed at 25°C until starvation (10–12 days). Plates containing non-Unc worms were counted as successfully injected. Occasionally, a batch of 50–60 injected animals yielded less than five successful injections, and we concluded that the injections had failed (usually for technical reasons, such as a bad needle) and repeated the injections. Following successful injections, plates with non-Unc worms were heat shocked at 34°C for 4 hours in an air incubator to activate the PEEL-1 negative selection marker, which kills animals carrying extrachromosomal arrays. After overnight recovery at 25°C, plates were visually screened to identify non-Unc animals that survived heat shock and did not express the red fluorescent co-injection markers. Single worms from these plates were picked to establish lines, and the presence of single-copy inserts was confirmed by PCR using primers listed in Supplementary Table 6.
To modify the genome using Cas9-triggered homologous recombination, we followed a protocol very similar to that for MosSCI (above). An injection mix containing 10 ng/μL homologous repair template, 50 ng/μL Cas9-sgRNA plasmid, and the negative selection and co-injection markers listed above was injected into young adults of strain DP38 (unc-119(ed3) III)43. Note that although we used DP38 for the experiments reported in this study, we are recommending the use of the outcrossed derivative HT1593 for future experiments (see Supplementary Protocol). The procedure for selecting insertions and eliminating extrachromosomal arrays was identical to that described above. We were able to isolate strains that were either homozygous or heterozygous for all of our modifications (D.J.D., unpublished observations); for the experiments presented in this study, only the homozygous lines were kept.
An injection mix containing 50 ng/μL of pDD104 (Peft-3::Cre::tbb-2 3’UTR) and 2.5 ng/μL pCFJ90 (Pmyo-2::mCherry pharyngeal marker) was injected into the gonads of young adult animals carrying an unc-119(+) cassette flanked by LoxP sites. In each experiment, 15–20 animals were injected and placed at 25°C. We picked single F1 progeny (10–20 per experiment) expressing the red pharyngeal marker, which represent progeny of successful injections. We then selected Unc animals from among the F2 progeny of these mCherry-positive animals. Because our Cre expression construct also carries unc-119(+), only animals that have: 1) Excised both genomic copies of the unc-119(+) cassette; and 2) Lost the extrachromosomal array generated by injecting the Cre expression construct will be Unc. We verified that animals picked during this step were mCherry-negative and segregated only Unc progeny. Excision of unc-119(+) was then confirmed by PCR (see Fig. 3c).
Embryos were isolated from gravid adult worms by bleaching and lysed by sonication for 20 minutes in a bath sonicator filled with boiling water. Lysates were separated on 3-8% NuPAGE Tris-Acetate gels (Life Technologies) and transferred to nitrocellulose. The following antibodies were used: Rabbit anti-NMY-2 (ref. 28) at 1:1000 dilution of crude serum; and Alexa Fluor 790 goat anti-rabbit (Life Technologies catalog number A11369) at 1:1000 dilution. Blots were scanned on an Odyssey imaging system (LI-COR Biosciences).
DIC and fluorescence imaging of whole worms was performed using a Nikon Eclipse E800 microscope equipped with epifluorescence and Nomarski DIC optics. Worms were mounted on 2.5% agar pads containing 10 mM sodium azide as a paralytic.
For NMY-2–GFP imaging, early 1-cell embryos were mounted on poly-L-lysine coated coverslips and gently flattened using 2.5% agar pads. Images were captured using a Nikon Eclipse Ti microscope equipped with a 60X, 1.4NA objective and a Yokogawa CSU-X1 spinning disk head.
Maximum intensity projection and adjustment of brightness and contrast were done using FIJI. No other image manipulations were performed.
RNA was isolated from gravid adult worms as follows. Worms were picked into Trizol reagent (Life Technologies) and lysed by repeated freeze-thaw cycles. 1/5 volume of Chloroform was added to separate phases, and the upper aqueous phase was mixed with an equal volume of ethanol and loaded on a RNeasy spin column (Qiagen). On-column DNase digestion was performed with the Qiagen on-column DNase digestion kit, and then RNA was washed and eluted according to the manufacturer's instructions. 25–75 ng total RNA were used for cDNA synthesis with the Superscript III Reverse Transcriptase kit (Life Technologies). qPCR was performed using a Viia 7 real-time PCR instrument and SYBR Green PCR Master Mix (Life Technologies). his-72 transcripts were detected with forward primer 5’-TCGTTCGTGAGATTGCCCAG-3’ and reverse primer 5’-GAGTCCGACGAGGTATGCTT-3’. Y45F10D.4 was used for normalization44,45. Data were analyzed to determine his-72 expression levels in Viia 7 software, using the default settings for a relative standard curve experiment.
Candidate off-target cleavage sites for each sgRNA were identified by BLAST searches46 against the C. elegans genomic sequence. For each candidate off-target site, we PCR amplified a ~1kb fragment centered on the candidate site from genomic DNA isolated from N2 control animals and from each modified strain we generated. The PCR products were sequenced by Eton Bioscience (Research Triangle Park, NC) using primers binding to each end of the fragment. The sequence reads were aligned to the genomic sequence to look for insertion or deletion mutations at the putative cleavage site.
Our sample sizes were chosen to allow confidence in the results while maintaining feasibility, and are consistent with established norms for C. elegans research and the developmental biology community more broadly. No randomization was necessary for these studies. Phenotypes presented in Fig. 5d-e were scored blindly.
We thank Christian Frøkjær-Jensen (University of Utah) for sharing strains, plasmids and protocols; Ken Kemphues (Cornell University) for antibodies; and Kerry Bloom, Amy Maddox, Gabi Monsalve, Nathalie Pujol, Kevin Slep, Stefan Taubert, Keith Yamamoto and members of the Goldstein lab for helpful suggestions and comments on the manuscript. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by National Institutes of Health (NIH) Office of Research Infrastructure Programs (P40 OD010440). This work was supported by NIH T32 CA009156 and a Howard Hughes postdoctoral fellowship from the Helen Hay Whitney Foundation (D.J.D.); postdoctoral fellowships from the Canadian Institutes of Health Research (award #234765) (J.D.W.); NIH R01 GM085309 (D.J.R.); NIH CA20535 and NSF MCB 1157767 (K. Yamamoto); and NIH R01 GM083071 and U.S. National Science Foundation (NSF) IOS 0917726 (B.G.)
D.J.D. and J.D.W. jointly conceived the project, and all authors discussed and contributed to the experimental design. D.J.D. performed the experiments. D.J.D. and D.J.R. analyzed the data. D.J.D. prepared the manuscript, and all authors discussed and contributed to the final version.
Competing Financial Interests
The authors declare no competing financial interests.