Machine learning based CRISPR gRNA design for therapeutic exon skipping

Restoring gene function by the induced skipping of deleterious exons has been shown to be effective for treating genetic disorders. However, many of the clinically successful therapies for exon skipping are transient oligonucleotide-based treatments that require frequent dosing. CRISPR-Cas9 based genome editing that causes exon skipping is a promising therapeutic modality that may offer permanent alleviation of genetic disease. We show that machine learning can select Cas9 guide RNAs that disrupt splice acceptors and cause the skipping of targeted exons. We experimentally measured the exon skipping frequencies of a diverse genome-integrated library of 791 splice sequences targeted by 1,063 guide RNAs in mouse embryonic stem cells. We found that our method, SkipGuide, is able to identify effective guide RNAs with a precision of 0.68 (50% threshold predicted exon skipping frequency) and 0.93 (70% threshold predicted exon skipping frequency). We anticipate that SkipGuide will be useful for selecting guide RNA candidates for evaluation of CRISPR-Cas9-mediated exon skipping therapy.     

Predicted changes in pre-mRNA secondary structure vary in their association with exon skipping for mutations in exons 2, 4, and 8 of the Hprt gene and exon 51 of the fibrillin gene

Exon skipping that accompanies exonic mutation might be caused by an effect of the mutation on pre-mRNA secondary structure. Previous attempts to associate predicted secondary structure of pre-mRNA with exon skipping have been hindered by either a small number of available mutations, sub-optimal structures, or weak effects on exon skipping. This report identifies more extensive sets of mutations from the human and hamster Hprt gene whose association with exon skipping is clear. Optimal secondary structures of the wild-type and mutant pre-mRNA surrounding each exon were predicted by energy minimization and were compared by energy dot plots. A significant association was found between the occurrence of exon skipping and the disruption of a stem containing the acceptor site consensus sequences of exon 8 of the human Hprt gene. However, no change in secondary structure was associated with skipping of exon 4 of the hamster Hprt gene. Using updated energy parameters we found a different structure than that previously reported for exon 2 of the hamster Hprt gene. In contrast to the previously reported structure, no significant association was found between predicted structural changes and skipping of exon 2. For all three Hprt exons studied, there was a significantly greater number of deoxythymidine substitutions among mutations accompanied by exon skipping than among mutations without exon skipping. For exon 8, deoxythymidine substitution was also associated with structural changes in the stem containing the acceptor site consensus sequences. For exon 51 of the human fibrillin gene, structural differences from wild type were predicted for all four mutations accompanied by exon skipping that were not were predicted for a single mutation without exon skipping. Our results suggest that both primary and secondary pre-mRNA structure contribute to definition of Hprt exons, which may involve exonic splicing enhancers.     

Multiple abnormal beta-hexosaminidase alpha chain mRNAs in a compound-heterozygous Ashkenazi Jewish patient with Tay-Sachs disease

Abnormal beta-hexosaminidase alpha chain cDNA clones were isolated from fibroblasts of an Ashkenazi Jewish patient with Tay-Sachs disease. Four abnormal cDNA clones were sequenced in their entirety. We showed previously that three of these mRNAs retained intron 12 with a mutation from G to C at the 5' donor site and that the patient was heterozygous with respect to this splicing defect (Ohno, K., and Suzuki, K., (1988) Biochem. Biophys. Res. Commun. 153, 463-469). One clone retained, in addition to intron 12, intron 13, which was truncated and polyadenylated due to a polyadenylation signal within intron 13. The fourth clone did not contain intron 12 and was missing exon 12. Some of these abnormal mRNAs were also missing one or more of upstream exons. The regions of exon 12-intron 12 and of upstream exons were evaluated in a total of 30 clones, including those completely sequenced, by restriction mapping and Southern analysis with appropriate probes. Of the 25 cDNA clones that included the exon 12-intron 12 region, 11 contained the exon 12-intron 12 sequence with the junctional transversion, and 11 were missing both exon 12 and intron 12. Among the 12 clones that included the region of exon 3-exon 9, 7 were missing one or more of upstream exons. Three clones gave results expected of normal cDNA in the region of exons 12 and 13. One of the three, furthermore, was 3.6-kilobases long and contained the completely normal beta-hexosaminidase alpha chain mRNA sequence on the 3' side and an abnormal 1.7-kilobase segment at the 5' end. These findings suggest that the splicing defect results in either retention of intron 12 or skipping of exon 12 in approximately equal proportions and that remote upstream exons are also frequently excised out. The three clones that were normal in the exon 12-intron 12 region could have derived from the other yet-to-be-characterized mutant allele. However, we were unable to obtain firm evidence that the abnormal upstream sequence is directly related to Tay-Sachs disease.     

Characterization of exon skipping mutants of the COP1 gene from Arabidopsis

The removal of introns from pre-mRNA requires accurate recognition and selection of the intron splice sites. Mutations which alter splice site selection and which lead to skipping of specific exons are indicative of intron/exon recognition mechanisms involving an exon definition process. In this paper, three independent mutants to the COP1 gene in Arabidopsis which show exon skipping were identified and the mutations which alter the normal splicing pattern were characterized. The mutation in cop1–1 was a G→A change 4 nt upstream from the 3′ splice site of intron 5, while the mutation in cop1–2 was a G→A at the first nucleotide of intron 6, abolishing the conserved G within the 5′ splice site consensus. The effect of these mutations was skipping of exon 6. The mutation in cop1–8 was G→A in the final nucleotide of intron 10 abolishing the conserved G within the 3′ splice site consensus and leading to skipping of exon 11. The splicing patterns surrounding exons 6 and 11 of COP1 in these three mutant lines of Arabidopsis provide evidence for exon definition mechanisms operating in plant splicing.     

Atypical 5' splice sites cause CFTR exon 9 to be vulnerable to skipping

The molecular basis of the skipping of constitutive exons in many messenger RNAs is not fully understood. A well-studied example is exon 9 of the human cystic fibrosis transmembrane conductance regulator gene (CFTR), in which an abbreviated polypyrimidine tract between the branch point A and the 3' splice site is associated with increased exon skipping and disease. However, many exons, both in CFTR and in other genes and have short polypyrimidine tracts in their 3' splice sites, yet they are not skipped. Inspection of the 5' splice sites immediately up- and downstream of exon 9 revealed deviations from consensus sequence, so we hypothesized that this exon may be inherently vulnerable to skipping. To test this idea, we constructed a CFTR minigene and replicated exon 9 skipping associated with the length of the polypyrimidine tract upstream of exon 9. We then mutated the flanking 5' splice sites and determined the effect on exon skipping. Conversion of the upstream 5' splice site to consensus by replacing a pyrimidine at position +3 with a purine resulted in increased exon skipping. In contrast, conversion of the downstream 5' splice site to consensus by insertion of an adenine at position +4 resulted in a substantial reduction in exon 9 skipping, regardless of whether the upstream 5' splice site was consensus or not. These results suggested that the native downstream 5' splice site plays an important role in CFTR exon 9 skipping, a hypothesis that was supported by data from sheep and mouse genomes. Although CFTR exon 9 in sheep is preceded by a long polypyrimidine tract (Y(14)), it skips exon 9 in vivo and has a nonconsensus downstream 5' splice site identical to that in humans. On the other hand, CFTR exon 9 in mice is preceded by a short polypyrimidine tract (Y(5)) but is not skipped in vivo. Its downstream 5' splice site differs from that in humans by a 2-nt insertion, which, when introduced into the human CFTR minigene, abolished exon 9 skipping. Taken together, these observations place renewed emphasis on deviations at 5' splice sites in nucleotides other than the invariant GT, particularly when such changes are found in conjunction with other altered splicing sequences, such as a shortened polypyrimidine tract. Thus, careful inspection of entire 5' splice sites may identify constitutive exons that are vulnerable to skipping.