Tissue-specific genome editing in Ciona embryos by CRISPR/Cas9

Recent advances have harnessed the CRISPR/Cas (Clustered Regularly Interspaced Short Palindromic Repeats/CRISPR-associated proteins) system for targeted genome editing (Ran et al., 2013). This prokaryotic immune system functions through short RNAs that guide Cas to foreign DNA (Barrangou et al., 2007; Brouns et al., 2008; Marraffini and Sontheimer, 2008). Modified variants of this system have been used for genome editing applications in various organisms (Cong et al., 2013; Dickinson et al., 2013; Friedland et al., 2013; Hwang et al., 2013; Mali et al., 2013; Qi et al., 2013; Wang et al., 2013a; Yang et al., 2013).

Ascidians of the genus Ciona are model organisms for chordate developmental genomics (Satoh, 2014). Hundreds of synchronized Ciona embryos can be simultaneously electroporated with plasmid DNA for high-throughput transgenesis (Corbo et al., 1997). This technique is used in gain-of-function experiments for overexpression of protein-coding genes (Stolfi and Christiaen, 2012). However, few options exist for scalable, tissue-specific loss-of-function experiments in Ciona.

Targeted mutagenesis in Ciona was reported using transcription activator-like effectors (TALEs) and zinc-finger nucleases (Kawai et al., 2012; Treen et al., 2014; Yoshida et al., 2014). Here, we report the use of CRISPR/Cas9 to induce site-specific double-stranded breaks (DSBs) in the C. intestinalis genome. By targeting a crucial region of the Ebf gene, we demonstrate the power of a simple electroporation-based transfection technique to deliver CRISPR/Cas9 components for tissue-specific, targeted mutagenesis in large batches of F0-generation Ciona embryos.

To test Cas9 expression in Ciona, we found that Cas9 flanked by two nuclear localization signals (nls) and a C-terminal eGFP tag (Chen et al., 2013; Fig. 1A) was strongly expressed and localized to nuclei (Fig. 1B). We avoided the use of Cas9 variants bearing an N-terminal hemagglutinin (HA) epitope tag, as these were found to be excluded from nuclei (supplementary material Fig. S1A-C).

We used the U6 promoter (Nishiyama and Fujiwara, 2008) to drive RNA polymerase III-dependent constitutive expression of single guide RNAs (sgRNAs) (Jinek et al., 2012; Mali et al., 2013). A recent study indicated that four consecutive Ts in the sgRNA hairpin result in premature transcriptional termination of T-rich sgRNAs (Wu et al., 2014). We used a modified sgRNA scaffold (‘F+E’, Fig. 1C) (Chen et al., 2013) to circumvent this problem. Indeed, by in situ hybridization we were able to detect transcription through the F+E backbone (Fig. 1D), but not through the original backbone (supplementary material Fig. S1D,E). This suggests that sgRNA constructs using the original backbone might not be expressed efficiently in Ciona embryos due to early termination in the hairpin.

Collier/Olf/EBF genes are transcriptional regulators of cell fate and differentiation in diverse tissues (Dubois et al., 1998; Crozatier and Vincent, 1999). We used CRISPR/Cas9 to induce mutations in exon 9 of Ebf (previously known as COE), the sole Ciona homolog of mammalian EBF1/2/3/4. This gene plays crucial roles in specification of motor ganglion neurons (Kratsios et al., 2012) and pharyngeal muscle precursors (Razy-Krajka et al., 2014; Stolfi et al., 2010). Ebf exon 9 codes for part of the IPT domain (immunoglobulin-like, plexin, transcription factor), situated between the DNA-binding and helix-loop-helix (HLH) domains (Fig. 2A).

We designed sgRNA vectors targeting exon 9 (Ebf.774 and Ebf.813) and a vector driving nls::Cas9::nls expression using the ubiquitous Eef1a (EF1α) promoter (Sasakura et al., 2010). Fertilized eggs (n>100) were pooled and electroporated with sgRNA/Cas9 plasmids. At 16 h postfertilization (hpf), genomic DNA extracted from the entire batch of transfected embryos (the ‘F0’ generation) was pooled. Ebf exon 9 was PCR-amplified and the resulting products TOPO-cloned. When individual clones were sequenced, three out of seven sequences carried deletions in exon 9 (Fig. 2B), suggesting that Cas9 had specifically induced DSBs that were then imperfectly repaired. Targeting with Cas9+Ebf.774 sgRNA (not in combination with Ebf.813 sgRNA) yielded six additional unique Ebf mutations (Fig. 2C, see below). We also generated targeted mutations in 5′ flanking regions of Foxf and Hand-related (supplementary material Fig. S2), suggesting that CRISPR/Cas9 can be harnessed for targeted mutagenesis of a variety of loci in the Ciona genome.

To verify that CRISPR/Cas9-targeted mutagenesis of the Ebf locus results in mutant Ebf transcripts, RNA was isolated from magnetic-activated cell sorting (MACS)-selected Ebf-expressing cells from pooled, dissociated embryos electroporated with Ebf>hCD4::mCherry, EF1α>nls::Cas9::nls and U6>Ebf.774. Partial Ebf cDNA fragments were amplified by RT-PCR and TOPO-cloned. Four out of eight sequenced clones had indels in the target sequence (supplementary material Fig. S3).

We used GeneArt Genomic Cleavage Detection to evaluate CRISPR/Cas9-induced mutagenesis. Cleavage assays were performed on Ebf exon 9 PCR products from batches of embryos electroporated with EF1α>nls::Cas9::nls and U6>Ebf.774. This resulted in a cleavage efficiency of 31.5% according to the formula provided by the manufacturer (Fig. 3A; supplementary material Fig. S4).

It should be noted that we found that the cleavage assay detected a large number of naturally occurring polymorphisms at other loci (supplementary material Fig. S5), potentially limiting its usefulness for studies using highly polymorphic Ciona populations (Tsagkogeorga et al., 2012).

A cleavage assay is likely to underestimate actual mutagenesis efficiency due to mosaicism of electroporated plasmids in Ciona embryos (Zeller et al., 2006). Electroporated embryos contain both transfected and non-transfected cells. In order to enrich for transfected cells, we performed MACS on dissociated cells from pooled embryos co-electroporated with 10 μg each of EF1α>hCD4::mCherry, EF1α>nls::Cas9::nls and U6>Ebf.774. The cleavage assay indicated a 27.1% efficiency in hCD4⁺-sorted cells, whereas cleavage bands from hCD4⁻ flow-through or unsorted cells were barely visible (Fig. 3B). These findings indicate that transfection mosaicism can mask mutagenesis efficiency, a problem that is overcome by MACS selection of transfected cells.

To test whether cleavage efficiencies could be improved by increasing CRISPR/Cas9 plasmid concentrations, we electroporated embryos with 10 μg EF1α>hCD4::mCherry, 25 μg EF1α>nls::Cas9::nls and 75 μg U6>Ebf.774. This resulted in a cleavage efficiency of 45.1% in unsorted cells and 66.2% in hCD4⁺ cells (Fig. 3C; supplementary material Fig. S6). Moreover, when TOPO-cloned, Ebf exon 9 amplicons from hCD4⁺ cells yielded six out of 13 sequences bearing novel mutations (Fig. 2C). Taken together, these data suggest that sgRNA vector distribution and concentration are limiting factors for efficient CRISPR/Cas9-mediated mutagenesis in electroporated embryos.

We tested how early in the development of Ciona embryos could CRISPR/Cas9-mediated targeted mutations be detected. We performed cleavage assays on PCR products amplified from DNA samples extracted at different time points from the same pool of transfected embryos. We found that mutations were not detectable at 4 hpf, but appeared at 8 hpf (Fig. 3D). Cleavage efficiency did not increase between 8 and 12 hpf, suggesting that CRISPR/Cas9 is limited by the onset of EF1α promoter activity (∼5 hpf, supplementary material Fig. S7), and that target sequence mutagenesis rapidly reaches saturation.

We next tested whether CRISPR/Cas9 could disrupt Ebf function in electroporated embryos. The Ciona motor ganglion is composed of neurons that innervate the tail to drive swimming behavior in the larva (Horie et al., 2010). Previous studies suggested that Ebf is located upstream of the conserved motor neuron (MN) regulatory factor Islet (Isl) in A10.57 MNs (Imai et al., 2009). To express Cas9 in early ectoderm, we chose the Sox1/2/3 (also known as SoxB1) promoter. We batch-electroporated embryos with Sox1/2/3>nls::Cas9::nls+U6>Ebf.774 and assayed MN-specific Isl reporter expression (Stolfi et al., 2010) at the larval stage. Isl reporter expression in MNs was seen in only 23% (n=100) of larvae that had been electroporated with Sox1/2/3>nls::Cas9::nls+U6>Ebf.774 (Fig. 4B,H), compared with 68% (n=100) of control larvae (Fig. 4A,H), suggesting that Ebf is required for Isl activation in MNs.

Off-target effects of CRISPR/Cas9 in eukaryotic cells have raised concerns about the specificity of purported targeted mutations (Hsu et al., 2013). In order to verify that downregulation of Isl was specifically attributable to Ebf loss-of-function, we designed a rescue construct. It consisted of an Ebf promoter driving an Ebf-coding sequence that contained synonymous substitutions in the Ebf.774-targeted seed sequence (Ebf>Ebf^m774). As predicted, co-electroporation of Ebf>Ebf^m774 rescued Isl reporter expression (Fig. 4C,H). We therefore conclude that the observed loss of Isl expression was not due to non-specific effects of CRISPR/Cas9, and that the majority of disrupted Ebf sequences are in fact loss-of-function alleles.

To target Ebf specifically in the B7.5 lineage, we used the Mesp promoter to express Cas9 in B7.5 blastomeres. The B7.5 lineage gives rise to three distinct muscle cells: tail, heart and atrial siphon (pharyngeal) muscles (ASMs) (Hirano and Nishida, 1997). Ciona Ebf is crucial for ASM specification and is also upstream of Isl expression in these cells (Razy-Krajka et al., 2014; Stolfi et al., 2010).

Lateral trunk ventral cells (TVCs) are normally specified as ASM precursors and express Isl (Fig. 4D) and the pharyngeal muscle markers Mrf (also known as MyoD) and Ncam (Fig. 4F; supplementary material Fig. S8) (Razy-Krajka et al., 2014). They detach from their sister cells (the heart precursors), activate Isl and migrate dorsally to surround the atrial siphon primordia (Fig. 4D). In larvae electroporated with Mesp>nls::Cas9::nls and U6>Ebf.774, lateral TVCs migrate and activate the Isl reporter in only 22% and 3% of larvae, respectively (n=100 each, Fig. 4E,I). A similar loss of Mrf and Ncam expression was also observed in these larvae (Fig. 4F,G,I; supplementary material Fig. S8). Moreover, 19 out of 30 such larvae raised to juveniles completely lacked ASMs (supplementary material Fig. S9). We conclude that Ebf is necessary in the B7.5 lineage for ASMP specification and migration, consistent with previous results using dominant repressor forms of Ebf (Razy-Krajka et al., 2014; Stolfi et al., 2010).

Using the CRISRPR/Cas9 system to generate targeted mutations, we have demonstrated a requirement for Ebf in specification of Isl⁺ MNs and pharyngeal muscles. We believe that the efficiency of CRISPR/Cas9, allied to the ease of transfection of Ciona embryos, promises for rapid scaling of genome editing in this model chordate.

While our paper was in press, Sasaki et al. reported CRISPR/Cas9-mediated gene knockout in C. intestinalis embryos (Sasaki et al., 2014).

Putative (N)21-GG targets were screened for polymorphisms and off-targets. sgRNA vectors were designed according to Mali et al. (2013) with scaffold modification (‘F+E’) (Chen et al., 2013). Annealed oligonucleotides were ligated into U6>sgRNA(F+E) linearized with BsaI, downstream of the Ciona U6 promoter (Nishiyama and Fujiwara, 2008). See supplementary material protocol for further information. For sgRNA transcription assays, eGFP sequence was inserted downstream of Ebf.34 sgRNA in F+E or original vectors, with or without termination sequences.

nls::Cas9::nls and nls::Cas9::nls::eGFP were derived from Sp-dCas9 (Qi et al., 2013). Cas9 variants were inserted downstream of the promoters Mesp (Davidson et al., 2005), Eef1a (EF1α) (Sasakura et al., 2010) and Sox1/2/3 (Stolfi et al., 2014). Isl and Ebf drivers/reporters have previously been published (Stolfi and Levine, 2011; Stolfi et al., 2010; Wang et al., 2013b).

For RT-PCR, RNA was extracted using RNAqueous kit (Ambion) and first strand-synthesized by Sensiscript-RT (Qiagen) primed with (dT)15 oligonucleotides. All targeted genomic or cDNAs amplified were cloned into pCRII-TOPO Dual-Promoter (Invitrogen).

Detailed primer, sgRNA and probe sequence information can be found in supplementary material Table S1. Our sgRNA and Cas9 vectors can be obtained through Addgene (www.addgene.org/browse/article/9026/)

Quikchange (Agilent) was used to generate a synonymous mutation in the target ‘seed’ of Ebf.774 from ACAGG to ACCGG to create Ebf^m774, which was subcloned downstream of the Ebf cis-regulatory sequences (−3348/exon1b).

DNA electroporation was performed on fertilized, dechorionated eggs from C. intestinalis (Type A) obtained commercially (M-REP) as described (Christiaen et al., 2009a,,b). Electroporated plasmid amounts (e.g. 10 μg) were per 700 μl of total volume. In situ hybridization was carried out as described (Beh et al., 2007; Christiaen et al., 2009c). Probes were transcribed in vitro from linearized plasmid. PCR products with flanking T7 promoters were used as template for eGFP probes. Rat anti-HA (Roche, Cat#1867423), mouse monoclonal anti-β-Gal (Promega, Z3781), rabbit polyclonal anti-GFP (Abcam, AB6556), Alexa Fluor-conjugated secondary antibodies (Molecular Probes, A212428, A21422, A10037, A21434, A21042) were diluted 1:500-1:1000. Embryos were fixed in 4% (para)formaldehyde-MEM buffer and mounted in 2% DABCO/50% glycerol/PBS. Images were taken on a Leica inverted TCS SP8 X confocal microscope or a DM2500 epifluorescence microscope.

Pooled embryos or cells were lysed in buffer from GeneArt Genomic Cleavage Detection Assay Kit (Invitrogen). PCR was performed from lysate using AmpliTaq supplied with the kit, except for assays on MACS-selected cells, for which Pfx platinum polymerase (Invitrogen) was used. For AmpliTaq we used the kit protocol with 60°C annealing temperature. For Pfx, we used 30 cycles, 60°C annealing temperature and 68°C extension temperature. For experiments comparing sgRNA vector concentrations, 50-cycle reactions were required to amplify sufficient material for assays and cloning. PCR products were purified using QIAquick PCR purification kit (Qiagen) or Nucleospin gel cleanup kit (Macherey-Nagel). Denaturing, re-annealing and incubation with detection enzyme were performed according to the manual. Reactions were run on a 2% agarose gel at 70 V for 30 min and imaged using Molecular Imager Gel Doc XR+ (Bio-Rad). Analysis was performed with ImageJ (National Institutes of Health). Cleavage efficiency was determined according to kit guidelines (Guschin et al., 2010).

Embryos were dissociated as described (Christiaen et al., 2009d). OctoMACS starting kit (Miltenyi Biotec) was used according to the manufacturer's recommendations, with modifications. Prior to sorting, dissociated cells were resuspended in 90 μl 0.05% BSA in calcium/magnesium-free artificial sea water and incubated with 10 μl anti-hCD4 antibody-coated microbeads (Miltenyi Biotec) at 4°C for 1 h. After applying diluted cells to pre-filter and column, flow-through was collected as hCD4⁻ cells. Columns were washed three times with the same buffer and hCD4⁺ cells were flushed with the kit plunger.

We thank Christiaen lab members for thoughtful discussions.