Results
Effect of gRNA sequence on RNA knock-down efficiency
Schematic diagram of the binary expression system of Cas13b(d) protein and crRNA was represented in Fig.1A. The expression of Cas13b(d) cDNA is driven by CMV promoter, and crRNA is expressed under U6 promoter. After the expression of Cas13b(d) and crRNA in cells, they form a complex and recognize the mRNA strand according to the sequence of the crRNA and cleavage the mRNA by Cas13b(d) protein[23]. In order to facilitate the detection, we first used the firefly luciferase as the target and renilla luciferase as an internal reference. The crRNAs complementary to the firefly luciferase mRNA were designed and cloned into the expression plasmid. Cas13b(d)-crRNA and luciferase expression plasmids were co-transfected into HEK293T. The result showed that as the amount of Cas13b(d) plasmid increased, the efficiency of firefly luciferase knock-down by both Cas13b and Cas13d were increased until a similar threshold reached. However, under the same amount of Cas13b(d) plasmid, Cas13d had a higher overall efficiency than that of Cas13b (Fig.1B).
To study whether the location site on target RNA strand and the sequence composition of crRNA has an impact on the effect of Cas13b(d) RNA cleaving, 33 crRNAs against luciferase were designed in three ways. The first batch of gRNA was designed based on CRISPR-Cas13 system targeting database (http://chopchop.cbu.uib.no). The second was based on Primer3 (https://primer3.ut.ee) to select sequences with high specificity. The third was selected randomly. All these 3 batches of gRNA showed a similar average efficiency and tendency. The results also indicated that the efficiency of Cas13b and Cas13d varied with different crRNA. Meanwhile, a site with high efficiency for Cas13b did not predict the same efficiency for Cas13d. Although no obvious relationship was found between the efficiency of Cas13b(d) and the base composition of crRNA (Fig.1E), it was worth to note that in our Case the crRNA selected against the position close to the middle of the target RNA generally had better degradation effect than those near the 5’ and 3’ marginal regions for both Cas13b and Cas13d (Fig.1C, 1D). However, the same outcome was not observed on d2GFP RNA as previous study (Fig.S1A-S1C)[32]. Compared with Luciferase, the d2GFP is much shorter, so whether the target RNA length and the secondary structure of RNA affect the efficiency of Cas13b(d) needs more research.
Effect of crRNA length on RNA knock-down efficiency
To explore whether length of gRNA could affect the efficiency of Cas13b(d), three pairs of crRNAs with 20 nt or 30 nt in length were chosen to target the 5’, middle or 3’of luciferase RNA respectively. Interestingly, Cas13b performed better than Cas13d when shorter crRNAs were used (Fig.2A-2D). To further confirm the result , crRNAs of different lengths were used to explore the effect of the length of crRNA on the efficiency of the two kinds of Cas13. To avoid the influence of protospacer flanking site (PFS) , the crRNAs with either a 5′ PFS of the original 30 nt crRNA or a 3 ′ PFS (Fig.2E) were used. As long as the 5’PFS or 3’ PFS remained unchanged, the cut efficiency were keep relatively stable for Cas13b but not for Cas13d (Fig.2F-2G). The most appropriate crRNA length for Cas13b was 22-25 nt. Notably, Cas13b still played its function even when crRNA was only 15 nt in length. On the contrary, Cas13d was dependent on longer crRNAs, when the length of crRNA was less than 22 nt, the effect of Cas13d was largely affected. The same tendency was obtained using another target site of luciferase RNA which Cas13b showed better efficiency than Cas13d with 30 nt gRNA (Fig.2H, 2I).
The crRNA precursor processing ability of Cas13b(d)
In order to test the processing ability of crRNA precursor of Cas13b(d), crRNA sequences located in the 5’, middle or 3’ region of luciferase RNA were selected respectively, and connected them together with DR repeat sequence to form crRNA precursor, and cloned into the expression plasmid (Fig.3A). The results showed that Cas13b with three crRNAs produced by one precursor expression vector were more efficient than that with the same crRNAs from three independent vectors (Fig.3B-3E). However, for Cas13d, there was no difference between the two strategies, even the opposite result was observed (Fig.3F-3I).
The result above suggested that Cas13b might be more capable of processing crRNA precursor than Cas13d.
Application of Cas13b and Cas13d in endogenous mRNA knock-down
To test the knock-down efficiency of Cas13 system to endogenous mRNAs, crRNAs for Zeb1 were used, both Cas13b and Cas13d could significantly reduce the expression of Zeb1 (Fig.4A, 4B). In order to further study the positional effect of crRNAs, we selected 20 crRNAs for each Zeb1 and Dnmt3a mRNA to test the knock-down efficiency. The results were generally consistent with the above findings. Namely, crRNAs located in the middle of mRNA had higher efficiency on average (Fig.4C, 4D, Fig.S2A, S2B).
The effect of crRNA length on the knock-down efficiency of Cas13b and Cas13d was also investigated using Zeb1 and Dnmt3a as targets. The results were in agreement with the above. Cas13b exhibited high compatibility for crRNAs with a wide range of length, while Cas13d required a relatively long crRNA (Fig.4E, 4F, Fig.S2C, S2D).
In order to know whether the precursor formed by concatenation of multiple crRNAs could affect the knock-down efficiency of endogenous genes, three crRNAs of Dnmt3a mRNA were concatenated with DR repeats into the precursor sequence (Fig.4G). Neither Cas13b nor Cas13d showed different efficiency compared to the mixed individual mature crRNAs (Fig.4H). Moreover, to test whether precursors composed of crRNAs targeting different genes could knock-down the genes simultaneously, crRNA precursors targeting Dnmt3a , PbxipI and Zeb1mRNA were designed (Fig.4I). The results showed that for the three genes, crRNA precursors had the same inhibitory efficiency as a mixture of each mature crRNAs (Fig.4J-4L).
Off target effects of Cas13b and Cas13d
To investigate the off-target effects of Cas13b and Cas13d, RNA-seq was performed. HEK293T cells were transfected with luciferase, crRNA and Cas13b/Cas13d expression plasmids. The same transfection system but without crRNA was served as the control groups. Both 20 nt and 30 nt crRNAs against luciferase RNA were combined with Cas13b and Cas13d respectively. The transfected cell samples were subjected to luciferase assay to confirm the knock-down efficiency of Cas13 system. RNA-seq results showed that the combination of Cas13d with 20nt crRNA had the least differential genes, and there was no significant difference among the other three groups (Fig. 5A-5D). The proportion of genes in different fold change clusters to the total differential genes was approximately the same in each group except the group of Cas13d with 20 nt crRNA (Fig.5F-5G). It is worth noting that 20nt crRNA is not optimal for Cas13d, so the fact that the lower number of differential genes in this group may be related to the insufficient knock-down of the system (Fig. 5E).
The overlap of differential genes between groups were very small (Fig.5H-5L). Go analysis showed that the differential genes were not enriched in any certain pathways but dispersed in different pathways (figures 5J-5M). These results indicated that most of the differential genes were from random changes which might be caused by the off-target effect of Cas13b or Cas13d.
Application of Cas13b and Cas13d in vivo
Further,we tried to explore the RNA knock-down ability of Cas13 system in vivo. The DNA fragment expressing Cas13b and a crRNA against luciferase were knocked into the Rosa26 locus of mouse genome. The obtained transgenic mice (designated as Cas13bKI) were crossed with mouse strain expressing luciferase (named as LucKI) from an expression cassette inserted in the Rosa26 site of mice to obtain the double transgenic mice (dTg) (Fig.6A). There was an obvious decrease in the RNA expression levels (Fig.6B). And in vivo optical imaging experiment showed that the luciferase expression indicated by luminance in dTg was significantly decreased compared to LucKI mice no matter from the belly or the back of animals, and in both male and female mice (Fig.6C, 6D).
A second transgenic mouse systemically expressing Cas13b or Cas13d and a gRNA against mCherry was also constructed by knock-in the expression vector into Rosa26 site. The mCherry expressing plasmid was delivered to the liver of the transgenic mice by hydrodynamic injection (Fig.6E). The results showed that compared to the control mice, both protein and RNA expression of mCherry was significantly reduced in transgenic mice at 72 hours after injection (Fig. 6F-6I).
Besides, we tested whether Cas13 RNA editing system could be used as treatment for acute hepatic failure. It is reported that increased expression of NF-kB and TNFa aggravates disease progression in acute hepatic failure[35, 36]. Here, the plasmid expressing Cas13b (or Cas13d) and a gRNA against NF-kB (orTNFa ) was injected via tail vein into mice by hydrodynamic injection. At 48 hours after injection, the acute hepatic failure model was induced by LPS/D-GalN administration (Fig.6J). The results showed that mice with NF-kB or TNFa knock-down had a higher survival rate. None of the mice with NF-kB knock-down by Cas13b system died; Only one mouse died in Cas13b-TNFa group, Cas13d-NF-kB group and Cas13d-TNFa group respectively. However, half of mice died in the non-treatment group. Hematoxylin eosin staining showed that the liver injury in the non-treatment group was more serious in terms of pathological indicators such as interstitial hyperemia, hepatocyte separation, liver histolysis and diffuse hepatocyte necrosis (Fig.6K). Meanwhile, the RNA expression levels of NF-kB and TNFa in knock-down groups were much lower compared to the non-treatment group (Fig.6L-6M).