Ribozyme Mediated gRNA Generation for In Vitro and In Vivo CRISPR/Cas9 Mutagenesis

CRISPR/Cas9 is now regularly used for targeted mutagenesis in a wide variety of systems. Here we report the use of ribozymes for the generation of gRNAs both in vitro and in zebrafish embryos. We show that incorporation of ribozymes increases the types of promoters and number of target sites available for mutagenesis without compromising mutagenesis efficiency. We have tested this by comparing the efficiency of mutagenesis of gRNA constructs with and without ribozymes and also generated a transgenic zebrafish expressing gRNA using a heat shock promoter (RNA polymerase II-dependent promoter) that was able to induce mutagenesis of its target. Our method provides a streamlined approach to test gRNA efficiency as well as increasing the versatility of conditional gene knock out in zebrafish.


Introduction
Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9 technology has revolutionized biology by greatly simplifying targeted mutagenesis [1]. The method relies on selection of the target sequence by the Cas9 DNA endonuclease, using a guide RNA (gRNA). The gRNA usually contains 20 nucleotides that are complementary to the target site as well as sequences that are recognized by Cas9 itself. Cleavage of DNA is mediated by Cas9 and requires the presence of a protospacer adjacent motif (PAM) that is downstream of the target sequence (N20NGG). Mutagenesis of the target site is mediated by aberrant DNA repair after cleavage.
Zebrafish researchers have been quick to adopt CRISPR/Cas9 technology for the generation of germ line mutations in their model system [1,2]. Typically mutations are induced by injection of both Cas9 mRNA and the relevant gRNA into newly fertilized egg. Target selection is limited by the nucleotide sequence of the transcriptional start site in bacteriophage promoters used to generate the gRNAs. Tolerance of mismatches in the first two nucleotides of the target site by Cas9 has been exploited to increase the number of target sites; however, some studies have shown that this decreases the mutagenesis efficiency [3,4]. Although Csy4, a sequence specific RNA endonuclease has been used successfully to increase the number of available target sites by incorporation of its recognition sites within the RNA transcript, however its toxicity precludes its application to zebrafish [4].
A potentially powerful extension of the application of CRISPR/Cas9 technology is the generation of somatic mutations by the in-vivo expression of gRNAs. Current methods use the zebrafish RNA polymerase (RNAP) III-dependent U6 promoter, which produces transcripts with well-defined start and end sites, to express gRNA in zebrafish [5,6]. The utility of this approach is limited by the fact it is believed that U6 transcripts start with a G, constraining the available target sites [5]. However a study using mouse U6 promoter have shown that transcripts start with a purine (A or G) [7]. Moreover, the U6 promoter is expressed ubiquitously; so temporal and spatial specificity of in-vivo mutagenesis can only be achieved by expressing the Cas9 endonuclease under the control of tissue-specific RNAP II promoters. The use of similar promoters to drive gRNA expression would greatly increase the versatility of CRISPR/ Cas9 for conditional gene knock out; however, unlike the bacteriophage and RNAP III promoters currently used to generate gRNAs in vitro and in vivo respectively [5,8] the extensive processing of transcripts produced from RNAP II dependent promoters has precluded their use for gRNA expression [9][10][11].
Ribozymes are RNA molecules with catalytic activity and catalysis occurs utilizing sequence specific interactions within the RNA molecule [12]. In particular, Hammerhead (HH) and hepatitis delta virus (HDV) ribozymes are small and mediate sequence-specific intramolecular RNA cleavage [13]. Hammerhead and HDV ribozymes mediate cleavage of the RNA at the 3' and 5' end respectively. Incorporation of these ribozymes into RNA allows the generation of transcripts with precisely defined ends, an approach that has been successfully used for the invivo expression of gRNAs using RNAP-II dependent promoters both in yeast and mammalian cell lines [10,14]. In this study, we describe the application of ribozymes to generate gRNA both in-vitro and in-vivo for genome modification in zebrafish. Our method streamlines the generation of generation of gRNAs and increases the versatility of CRISPR/Cas9 mediated conditional mutagenesis in the zebrafish.

Fish husbandry and microinjection
Fish were maintained in the IMCB zebrafish facility and embryos were obtained through natural crosses and staged according to Kimmel et al. [15]. All experiments in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of the Biological Resource Centre (BRC) (IACUC # 110638/151020). This was in accordance with guidelines set by National Advisory Committee for Laboratory Animal Research (NACLAR) and Agri-Food and Veterinary Authority of Singapore (AVA). All experiments were limited to 0-5 dpf zebrafish embryos. The Tg(smyhc1:GFP) i104 transgenic line has been previously described [16].
For CRISPR mutagenesis, embryos were injected with 200 fg Cas9 mRNA and 50 fg gRNA or 300 fg mRNA containing mCherry and gRNA (this reflects the amount of RNA injected in each embryo). To test the efficiency of DNA constructs expressing gRNA, 30 fg of plasmid and 20 fg Tol2 RNA was injected into embryos. Tg(smhyc1:GFP) i104 was made using PI-SceI therefore Tol2 transposase should not remobilize the transgene. Transgenic zebrafish were generated by injecting embryos with 30 fg Tol2 and 30 fg Plasmid DNA.

Construction of transgenic and RNA synthesis
PCR amplification for cloning was done using PrimeSTAR Max (Takara, Clontech). gBlocks containing the HH-gRNA and HH-gRNA-HDV constructs were synthesized by IDT (Integrated DNA technologies) and final constructs were assembled using Gibson assembly (E2611, NEB). Middle entry vector containing Cas9 was cloned from pCS2-nCas9n [18]. Larger constructs containing Cas9 and Ubiquitin promoter [19] were constructed by Gateway cloning using components from the Tol2 kit [20]. HSP: smo gRNA was made using the hsp70l promoter and the Tol2 destination vector containing cmlc2:GFP marker from the Tol2 kit. Plasmids will be made available in Addgene.
mRNA was transcribed using Sp6 mMessage mMachine kit (Ambion, Invitrogen). gRNA was synthesized using the MEGAshortscript T7 kit (Ambion, Invitrogen). RNA was purified by lithium chloride and ammonium acetate for the mMessage mMachine kit and MEGAshortscript kit respectively. Further purification of RNA containing ribozyme is not necessary. Ribozyme cleavage of RNA was determined by denaturing 10% PAGE. Briefly, samples were resuspended in 50% formamide, 10mM EDTA and heated to 70˚C for 5 min and chilled on ice. Denaturing PAGE gels contain 1xTBE and 7 M urea. PAGE Gel was pre-run at 20 W for 15 min before loading samples. Electrophoresis was carried out at 10 W.

Genotyping and PAGE
PCR for genotyping was performed using GoTaq (Promega). Quantification of mutagenesis efficiency of various gRNA was performed as described [21]. Briefly, heteroduplex/homoduplex were induced by heating PCR products to 95˚C for 5 min, 85˚C for 10 s (ramp -2˚C/s), 25˚C for 10 s (ramp -0.1˚C/s) and finally stored at 4˚C. PAGE analysis was performed using 7.2% polyacrylamide gels in 1x TBE. Primers used for genotyping are listed in S1 Table. smo 5' UTR F and smo ATG R were used to detect truncations in smo 5' UTR.
For quantification of mutagenesis efficiency by cloning and sequencing, genomic DNA was extracted from twenty injected embryos and the gRNA target site was amplified by PCR (Smo 5' UTR F and Smo 5' UTR R) and cloned into pGEM-Teasy (Promega). Approximately twenty colonies were picked for colony PCR and amplified DNA was used for subsequent sequencing using Smo 5' UTR R.

Heat shock of HSP:smo gRNA
After injection of 250 fg Cas9 mRNA, embryos were shifted to 34.5˚C when they reached 64 cell stage. Embryos were heat-shocked twice beginning at High stage for 2 h at 37˚C with a 2 h interval at 34.5˚C. After heat shock embryos were incubated overnight at 34.5˚C.

Results
In-vitro generation of gRNA using Cis-acting hammerhead ribozyme Hammerhead (HH) ribozymes are comprised of a RNA sequence motif that catalyzes selfcleavage from the 3' end of the RNA molecule and they have been used to remove 5' end heterogeneity of RNA transcripts. Cleavage of the HH ribozyme occurs at stem loop one (Fig 1A). To make use of a HH ribozyme to generate gRNA, the first six nucleotides of the HH ribozyme is made complementary to the first six nucleotides of the target site of the gRNA [14]. We refer to this as HH-gRNA.
It was previously shown that in-vitro transcribed (IVT) gRNA with ribozymes was able to guide Cas9 cleavage of its target in an in-vitro cleavage assay [14]. To verify whether HH could mediate release of the gRNA without affecting the in-vivo mutagenesis efficiency, a construct encoding HH-gRNA targeting the 5' UTR of mRNA encoded by the zebrafish smoothened (smo) gene was generated. IVT of smo HH-gRNA showed the presence of a band corresponding to gRNA after electrophoresis (Fig 1B). Sequencing of individual embryos injected with this HH-gRNA and Cas9 mRNA, revealed mutations at the target site (Fig 1C).
In-vitro generation of biscistronic RNA containing a fluorescent protein and gRNA To mimic transcript processing following transcription by RNAPol II, a new construct was generated with H2a mCherry and HDV ribozyme upstream and downstream of HH-gRNA (Fig 2A). This new gRNA-containing construct, referred to as mCherryHH-gRNA-HDV, was transcribed using the mMessage Kit (kit produces 5' capped RNA for efficient translation). H2a mCherry was added as an independent assay for upstream sequence removal by the HH ribozyme as well as a visible marker of transgene-expressing cells in subsequent experiments. MALAT sequence forms a triple helical structure that stabilizes the H2a mCherry mRNA due to the removal of the polyA tail by ribozymes (Fig 2B and 2C) [10,22]. The HDV ribozyme does not require any base pairing with upstream sequences and mediates cleavage at its 5' end.
A previous study has shown that the sequence "UUUUU" can also terminate RNAP II transcripts [23] but this was not tested. IVT of the construct resulted in the detection of a band corresponding to gRNA following electrophoresis (Fig 1B). Zebrafish embryos injected with this transcript together with Cas9 mRNA showed nuclear accumulation of mCherry protein and harboured mutations of the target site in the smo gene (Fig 2B).
To validate general applicability of this method of ribozyme-mediated gRNA production, we tested another well-established gRNA that targets GFP [24]. Embryos carrying the smyhc1: GFP transgene that drives expression of GFP in slow-twitch muscle fibres were injected with mCherry-HH-gRNA-HDV targeting GFP along with Cas9 mRNA; all embryos (n = 34) showed a dramatic loss of GFP expression in their slow-twitch muscle fibres, reflecting efficient mutagenesis of the GFP target site (Fig 2C; S1B Fig, on average 6 GFP+ slow fibres remained in each somite compared to~24 in wild type). This demonstrates the general utility of the HH ribozyme method for generating functional gRNA.

Ribozyme mediated processing does not compromise gRNA efficiency
To compare the efficiency of gRNA targeting the smo 5'UTR generated by ribozymes with that generated by the conventional method, we injected equimolar (equivalent to 20 fg of gRNA) of each gRNA together with 200 fg of Cas9 mRNA. Previously it was shown that polyacrylamide gel electrophoresis (PAGE) was a reliable method for the detection as well as quantification of subtle allelic alterations [21]. This method relies on the formation of heteroduplexes between different alleles, which migrate slower than homoduplexes during electrophoresis. Genomic DNA was extracted from single embryos and PCR was used to amplify the smo 5'UTR. To determine the efficiency of mutagenesis, genomic DNA was extracted from single embryos and smo 5'UTR was amplified by PCR. Efficiency of mutagenesis was determined by comparing the ratio of heteroduplex and homoduplex. We found that gRNA generated using ribozymes is as efficient as gRNA generated using conventional methods, both show~60% efficiency (Fig 3). We also further evaluated this by cloning and sequencing gRNA target sites from pooled samples of twenty injected embryos. Sequencing of twenty clones from each sample showed that approximately 45% of these clones showed mutagenesis of its target site (S1A Fig). Therefore ribozyme mediated gRNA generation does not affect the efficiency of gRNA targeting.

In-vivo expression of gRNA and Cas9 can induce target site mutation
As a preliminary test for the in-vivo expression of gRNA using ribozymes and RNAPol II dependent promoters, we generated a construct (UBI:Cas9 UBI:GFP gRNA) that drives ubiquitous expression of Cas9 and mCherryHH-gRNA-HDV targeting GFP (Fig 4A). This construct was injected into Tg(smyhc1:GFP) i104 embryos and expression of H2a mCherry was monitored to confirm expression of gRNA. We found a number of embryos (n = 24) with  ~24 in wild type). However, loss of GFP expression was less extensive than that induced by RNA (14 compared with 6 in DNA and RNA injections respectively, Figs 2C and 4B; S1B Fig) implying that the in-vivo gRNA production from DNA expression constructs is less efficient. PAGE analysis confirmed that mutagenesis of GFP occurred (Fig 4C).
To explore further the expression of gRNAs by RNAPol II dependent promoters, another construct (HSP: Smo gRNA) expressing two gRNAs targeting Smo ATG and Smo 5' UTR driven by the heat shock promoter (hsp70l) was generated ( Fig 4D). As expected, heat shock of Mutagenesis of gRNA target sites was confirmed using PAGE analysis and truncations of Smo 5' UTR could also be detected (Fig 4F). Loss of smo activity leads to defects in muscle specification, including loss of Engrailed positive muscle pioneer cells and medial fast fibres [25,26]. Immunohistochemical analysis revealed effects on both cell types in the injected embryos (S3B Fig).

Discussion
We have demonstrated efficient expression of gRNA using RNAP II dependent promoters without loss of targeted mutagenesis capacity. Conditional CRISPR/Cas9 mediated mutagenesis requires in-vivo expression of both Cas9 and gRNA; the ability to drive both using RNAP II dependent promoters should expand the versatility of the method significantly. For example, intersectional expression by different promoters for Cas9 and gRNA can be used to restrict gene targeting to very precise populations of cells as well as specific developmental stages. Moreover, the small size of ribozymes, allows expression of multiple gRNAs to increase the efficiency of mutagenesis or target multiple genes simultaneously.
Owing to the mosaic nature of DNA injections, far fewer heteroduplexes are seen following DNA injections compared to RNA injections (Fig 4C). This highlights the difficulty in validating gRNA targets using DNA injections. Low efficiency of mutagenesis might reflect low number of cells expressing constructs or failure of gRNA to mutagenize its target site. This also hampers comparisons between different gRNAs. By contrast, we find that RNA injections of gRNA and Cas9 are much more efficient. Combining ribozyme incorporation and IVT using bacteriophage promoters, allows testing of gRNA efficiency using RNA injections, before generating transgenic lines for conditional mutagenesis (S4 Fig). In most cases where a single gRNA is used for mutagenesis of a gene of interest, mutations leading to frame-shifts or premature termination of the gene of interest are analysed for gene loss of function. There have been reports showing the low diversity of recovered mutant alleles, a number of which do not generate frame-shift mutations [3]. Therefore, it is possible that conditional alleles generated using CRISPR/Cas9 might lead to the formation of genetic mosaics in the tissue of interest. To circumvent this problem multiple gRNAs targeting the same gene might be employed to increase the likelihood of inactivation. Alternatively, generating conditional CRISPR/Cas9 transgenic in a heterozygous mutant background should increase the efficiency of inducing complete loss of function of the targeted gene.
Overall, our method provides a streamlined approach for generating conditional mutants, as the same gRNA can first be transcribed for testing mutagenesis efficiency and subsequently used for generation of stable zebrafish lines. Our method also provides a simple way to increase the repertoire of available gRNA targets.