mCherry was well-regulated by fused ecDHFR DD
ecDHFR DD can regulate the expression of the fused proteins with the effect of TMP4. To assess the versatility of the ecDHFR DD system, two ecDHFR DD-regulated fluorescence reporters, namely mCherry-ecDHFR DD and ecDHFR DD-mCherry, were designed and ecDHFR DD was fused at the C- and N-terminal of mCherry (Figure 1a). Three constructs, including mCherry, mCherry-ecDHFR DD, and ecDHFR DD-mCherry, were used to analyze the mCherry expression under the control of the ecDHFR DD-TMP combination. TMP was expected to prevent the destabilizing effect of ecDHFR DD, thereby protecting mCherry from degradation (Figure 1b). The three fluorescence reporters were transfected into 293T cells, and different doses of TMP (0, 1, 2, 4, 6, and 8 ng/μL) were added 24 h post-transfection. At 48 h post-transfection, lower mCherry expression was observed in mCherry-ecDHFR DD- and ecDHFR DD-mCherry-transfected cells without TMP treatment compared with the mCherry-transfected group, whereas it was significantly increased in mCherry-ecDHFR DD-transfected cells upon TMP treatment, reaching the maximum mCherry effective concentration at 2 ng/μL TMP (Figures 1c and S1). Therefore, the optimal concentration of 2 ng/μL TMP was selected for the following experiments. Of note, the inhibitory effect could be restored after TMP removal for 48 h. However, the mCherry expression did not recover after the TMP treatment in ecDHFR-mCherry-transfected cells, suggesting that the N-terminal ecDHFR DD fusion could be insensitive to all TMP concentrations (Figures 1c and S1). These results showed that mCherry expression could be effectively regulated by the ecDHFR DD-TMP combination, under the control of the ecDHFR DD fused at the C-terminal of mCherry.
ecDHFR DD well-regulated ecRESCUE system expression
Next, based on above-described results, the ability of ecDHFR DD to regulate the expression of the two main functional elements of RESCUE RNA editor (dRanCas13b and human ADAR2), which reduced the RESCUE effective time and off-target events, was evaluated (Figure 2b). Two mCherry expression RNA base editors were used; RESCUE-mCherry and RESCUE-mCherry-ecDHFR RNA base editors, with mCherry and ecDHFR DD fused at the C-terminal of the RESCUE and RESCUE-mCherry, respectively, were considered for further experiments (Figure 2a). The two RNA base editors, RESCUE and ecRESCUE, were transfected into 293T cells, 2 ng/μL TMP was added to the cells 24 h post-transfection, and fluorescence analysis was performed to detect the expression of dRanCas13b and ADAR2. The results showed that mCherry levels were markedly decreased in ecRESCUE-transfected cells without TMP treatment compared to the RESCUE-transfected group. Moreover, 2 ng/μL TMP treatment could inhibit the ecDHFR DD-mediated fused protein degradation and partially recover the expression of the RESCUE-mCherry protein. After TMP was removed for 24 and 48 h, the inhibitory effect of ecDHFR DD was markedly restored and the inhibitory effect was remarkably increased with the extension of TMP removal time (Figure 2d and S2). To confirm that ecDHFR DD regulated protein levels at the post-translational level without affecting mRNA expression, the mRNA expression of dRanCas13b, ADAR2, and fused mCherry in both the RESCUE and ecRESCUE RNA base editors was evaluated by performing reverse transcriptase-polymerase chain reaction (RT-PCR). As expected, the expression of dRanCas13b, ADAR2, and fused mCherry in the presence or absence of TMP was similar in the two RNA base editor systems (Figure 2c). Taken together, these data demonstrate that C-terminal ecDHFR DD fusion enables remarkable, rapid, and reversible control of the expression of the RESCUE-related functional proteins, thereby making ecRESCUE a suitable tool for the generation of inducible RNA editing systems.
ecRESCUE mediates A-to-I and C-to-U base editing of endogenous mRNA
Furthermore, the endogenous RNA editing efficiency of ecRESCUE was evaluated in 293T cells. To detect A-to-I and C-to-U editing efficiency, 13 endogenous sites were selected. Specifically, we considered KRAS, NRAS, NFKB1, AHI1, APC, COL3A1, DMD, MSH6, PRKN, SCN9A, SH3TC2, TARDBP, and UBE3A for performing A-to-I editing, and KRAS, NRAS, NFKB1, AHI1, ALDOB, DMD, IL2RG, MSH6, PRKN, SCN9A, SH3TC2, TARDBP, and UBE3A for performing C-to-U editing. For each endogenous site, one single guide RNA (sgRNA, Table S1) fused with GFP was designed to guide the precise endogenous RNA editing. To compare the RNA editing efficiency between the RESCUE and ecRESCUE systems, two experiment groups were set up, including RESCUE/ecRESCUE group and RESCUE/ecRESCUE+TMP group, in which 2 ng/μL TMP was added at 24 h post-transfection. At 48 h post-transfection, 2 × 104 GFP-positive cells were collected by fluorescence-activated cell sorting, and total RNA from these cells was collected. The target sequences of the endogenous sites were amplified and subjected to Sanger sequencing. The mRNA base editing efficiency of these targets were calculated by using the Edit R web tool (https://moriaritylab.shinyapps.io/editr_v10/). The results showed relatively less RNA editing efficiency with the ecRESCUE system in the absence of TMP. Moreover, no significant differences were observed in both A-to-I (Figure 3a) and C-to-U (Figure 3b) RNA editing efficiency between the RESCUE and ecRESCUE RNA editors with TMP treatment, demonstrating that the ecRESCUE RNA editor was a versatile RNA base editor.
Transcriptome-wide A-to-I and C-to-U RNA off-targets were significantly decreased in ecRESCUE system
We evaluated whether the new RNA base editor could decrease the off-target effects. To this end, three endogenous sites (KRAS, NFKB1, and NRAS) were examined for mRNA A-to-I base editing by using the RESCUE and ecRESCUE RNA editors (Figure 4a). In a first approach, transcriptome-wide off-targets were assessed by performing RNA-sequencing of all mRNA samples. The results showed that there were 2,116 and 2,176 A-to-I off-targets in KRAS targeting cells obtained by using RESCUE, and 136 and 148 off-targets were obtained by using ecRESCUE with 50× coverage; 2,726 and 2,700 A-to-I off-targets in NFKB1 targeting cells were obtained by using RESCUE, and 139 and 127 off-targets were obtained by using ecRESCUE with 50× coverage; and 1,507 and 1,568 A-to-I off-targets in NRAS targeting cells were obtained by using RESCUE, and 150 and 124 off-targets were obtained by using ecRESCUE with 50× coverage (Figure 4b). These results showed that the ecRESCUE RNA editor led to the formation of markedly less A-to-I off-targets than the standard RESCUE system. Furthermore, the RNA-sequencing results also showed that there were 1,526 and 1,586 C-to-U off-targets in KRAS targeting cells obtained by using RESCUE, and 54 and 52 off-targets were obtained by using ecRESCUE with 50× coverage; 1,974 and 1,965 C-to-U off-targets in NFKB1 targeting cells were obtained by using RESCUE, and 60 and 50 off-targets were obtained by using ecRESCUE with 50× coverage; and 1,098 and 1,142 C-to-U off-targets in NFKB1 targeting cells were obtained by using RESCUE, and 84 and 80 off-targets were obtained by using ecRESCUE with 50× coverage (Figure 4c); demonstrating that the ecRESCUE RNA editor led to the formation of markedly less C-to-U off-targets. Taken together, these results demonstrated that the ecRESCUE editor led to the formation of remarkably less off-targets, thus considerably improving the RNA editor system specificity.