Two strategies to generate cellular chain reactions using nCas9-CDA
We designed two systems to generate chain reactions using nCas9-CDA. In the first system, the T>C mutations are introduced in the scaffold region of sgRNA to disrupt its structure. The T>C mutations in the sgRNA (sgRNA-2) are corrected by nCas9-CDA and another sgRNA (sgRNA-1). The corrected sgRNA-2 in turn corrects the T>C mutations in the third sgRNA (sgRNA-3) in cooperation with nCas9-CDA (Fig. 1A). Because the sequence and the length of the stem-loop region in the sgRNA scaffold are not restrictive4,5, we can design multiple sgRNAs by introducing various sequences in the stem-loop region (XXXXX in sgRNA-2-MT, YYYYY in sgRNA-3-MT) (Fig. 1B).
In the second system, the T>C mutations are introduced in the TATAloxP sequence (TATAloxP-MT). TATAloxP forms a 13-8-13 structure, which is composed of an 8-bp spacer sequence containing TATA box flanked on either side by two 13-bp inverted repeats (Fig. 1C). Cre recombinase binds to the 13 bp repeats of the TATAloxP, and importantly, this TATA box containing sequence can be incorporated into the U6 promoter to induce sgRNA expression6. The T>C mutations in the TATAloxP are corrected by the nCas9-CDA and the sgRNA (sgRNA-1) targeting the mutations and adjacent sequences (XXXXX in Fig. 1D). After the correction, a loxP flanked stop cassette is excised by Cre recombinase, which drives expression of the second sgRNA (sgRNA-2). The sgRNA-2 targets the next mutatiosn and adjacent sequences (YYYYY in Fig. 1D), resulting in expression of the third sgRNA (sgRNA-3) (Fig. 1D). Because there is no restriction for the adjacent sequences, we can design multiple sgRNAs with various adjacent sequences that can be activated during the chain reaction.
Generation of inactive sgRNAs with T>C mutations in the scaffold region
To establish the first system, we introduced T>C mutations in the template sequence of the sgRNA scaffold region for its inactivation. Because nCas9-CDA was shown to induce C>T substitution in DNA efficiently when the cytosine is located 18bp upstream from PAM sequence3, we also introduced PAM sequence (NGG) 18 bp downstream from the T>C mutation in each sgRNA template (Fig. 2A and 2B). We then assessed activity of each sgRNA with the T>C mutations and PAM insertion using canavanine assay. Canavanine is a toxic analogue of arginine and is imported into yeast cells via a transporter Can1. Therefore, expression of Cas9 and Can1-targeting sgRNA induces Can1 depletion and decreases the sensitivity of yeast to canavanine7. We used this assay to evaluate impact of each T>C mutation for inactivation of the sgRNA, and found that the T>C mutation at 4th base resulted in sgRNA inactivation in yeast. Insertion of PAM sequence did not inhibit the sgRNA function in yeast (Fig. 2D and 2E).
We then assessed the effect of the T>C mutations on sgRNA function in mammalian cells. Mutant EGFP in which the start codon was mutated from ATG to GTG was used8. In this mutant EGFP, the template strand has a T>C mutation at start codon and its 18bp-downstream PAM sequence. Therefore, expression of nCas9-CDA and active sgRNA can correct the T>C mutation in the mutant EGFP to induce GFP expression (Fig. 3A and 3B). Unfortunately, unlike the results obtained by the canavanine assay, insertion of PAM sequence itself inactivated sgRNA in mammalian cells (Fig. 3C). Because the weak interaction between G and U is known to be important for maintaining RNA structure9, we speculated that the loss of G-U interaction at 7th base induced by the PAM insertion might impair sgRNA function (Fig. 3D). Therefore, we introduced additional sequences near to the PAM sequence to maintain the G-U interaction at 7th base. Elongation of the stem-loop sequences in the sgRNA scaffold was shown to have little influence on sgRNA function4,5. As expected, this optimized PAM-inserted sgRNA retained the G-U interaction at 7th base (Fig. 3D) and efficiently induced GFP expression when it was expressed with nCas9-CDA and mutant EGFP in 293T cells (Fig. 3E).
We next examined if the T>C mutation at 4th base, which diminished sgRNA activity in yeast, also inactivates sgRNA in mammalian cells. Again, we got different results in mammalian cells from those obtained by the canavanine assay. The sgRNA with single T>C mutation at 4th base efficiently induced GFP expression, suggesting that it retained normal function. Therefore, we introduced T>C mutations at 2, 3, or 5th base to the sgRNA template in addition to the 4th base. These sgRNAs with double T>C mutations at 2/4th, 3/4th, 4/5th bases lost the function to recover GFP expression in 293T cells (Fig. 3F). We then performed similar experiments using the PAM-inserted sgRNAs and confirmed that the double T>C mutations at 3/4th bases, but not single T>C mutation at 4th base, resulted in loss of sgRNA function (Fig. 3G). Thus, we generated an inactive sgRNA with T>C mutations and PAM sequence in the scaffold region, whose function can theoretically be recovered by the nCas9-CDA-induced base editing in mammalian cells.
Generation of non-responsive TATAloxP sequences with T>C mutations
To establish the second system, we introduced T>C mutations in the TATAloxP sequences to make it non-responsive to Cre-induced recombination. We generated several TATAloxP mutants in which one base of C in the 13 bp repeat sequence (Fig. 1C) was replaced with T. These DNAs with various TATAloxP sequences at either 5' or 3' sides (approximately 3500 bp) were linearized and incubated with Cre recombinase. In this in vitro assay, Cre-induced recombination produces approximately 7000 bp linear DNA when the Cre recombinase can recognize the TATAloxP sites (Fig. 4A). As shown in Fig. 4B and 4C, the T>C mutation at the 13th base resulted in the reduced recombination in both mutant/mutant and mutant/wild-type incubations.
Next, we examined if the TATAloxP sequences with the T>C mutations are resistant to Cre-mediated recombination in mammalian cells. We generated an expression vector containing polyA signal flanked with two TATAloxP sites and a downstream EGFP cassette under EF1a promoter, in which one of the TATAloxP site has the T>C mutations. If the mutant TATAloxP site is resistant to Cre-mediated recombination, GFP is not expressed even in the presence of Cre-recombinase (Fig. 4D). We expressed the TATAloxP (wild-type or mutant)-EGFP constructs in 293T cells together with Cre-R32V, a mutant Cre recombinase with improved accuracy10 (Fig. 4E). However, unlike the results of in vitro experiments, the T>C mutation at the 13th base in one or both 13-bp arms did not prevent Cre-R32V-induced GFP expression (Fig. 4F). Thus, the T>C mutation at 13th base was not sufficient to make the TATAloxP sequence non-responsive to Cre recombinase in mammalian cells. Because a previous report showed that double mutations in the both 13-bp arms, in particular at the 7th, 8th, 11th, 12th and 13th base, disrupt loxP structure efficiently11, we then assessed the effect of double T>C mutations at 11th and/or 13th base on their responsiveness to Cre recombinase. Among the various combinations, we found that double T>C mutations at 11th and 13th bases in both arms became resistant to Cre-induced GFP expression (Fig. 4G). Thus, we generated a non-responsive TATAloxP sequence with T>C mutations, whose responsiveness to Cre can theoretically be recovered by the nCas9-CDA-induced base editing in mammalian cells.
Establishment of a chain reaction in mammalian cells
Finally, we examined whether the inactive sgRNA or the non-responsive TATAloxP with T>C mutations can be converted to the active form by the nCas9-CDA-induced base editing in mammalian cells. First, we transduced nCas9-CDA, the mutant EGFP, PAM-inserted control or inactive sgRNA targeting the EGFP mutation with or without the second sgRNA into 293T cells. The second sgRNA was designed to correct the T>C mutations in the EGFP-targeting sgRNA to make it an active form (Fig. 5A and 5B). As expected, the optimized version of PAM-inserted sgRNA efficiently corrected the EGFP mutation and induced GFP expression, while that with T>C mutation did not. Importantly, coexpression of the second sgRNA recovered the function of the inactive sgRNA to drive robust GFP expression in 293T cells (Fig.5C). Thus, the chain reaction was successfully achieved in mammalian cells using nCas9-CDA and two sgRNAs targeting the T>C mutation or EGFP mutation.
Second, we transduced wild-type and the non-responsive TATAloxP-EGFP, Cre-R32V and nCas9-CDA with or without the second sgRNA into 293T cells. The second sgRNA was designed to correct the T>C mutations in the non-responsive TATAloxP to convert it to a responsive form (Fig. 5D and 5E). Consistent with earlier results, Cre-R32V induced recombination of only wild-type TATAloxP, but not mutant TATAloxP, to induce GFP expression in 293T cells (Fig. 5F). Unfortunately, coexpression of the second sgRNA did not recover the GFP expression, indicating that the mutant TATAloxP with double T>C mutations at 11th and 13th bases is resistant to the nCas9-DNA-mediated base editing. Thus, this TATAloxP system is not suitable for cellular chain reactions in its current form.