CRISPR/Cas9-Mediated Enrichment Coupled to Nanopore Sequencing Provides a Valuable Tool for the Precise Reconstruction of Large Genomic Target Regions

Abstract

Complete and accurate identification of genetic variants associated with specific phenotypes can be challenging when there is a high level of genomic divergence between individuals in a study and the corresponding reference genome. We have applied the Cas9-mediated enrichment coupled to nanopore sequencing to perform a targeted de novo assembly and accurately reconstruct a genomic region of interest. This approach was used to reconstruct a 250-kbp target region on chromosome 5 of the common bean genome (Phaseolus vulgaris) associated with the shattering phenotype. Comparing a non-shattering cultivar (Midas) with the reference genome revealed many single-nucleotide variants and structural variants in this region. We cut five 50-kbp tiled sub-regions of Midas genomic DNA using Cas9, followed by sequencing on a MinION device and de novo assembly, generating a single contig spanning the whole 250-kbp region. This assembly increased the number of Illumina reads mapping to genes in the region, improving their genotypability for downstream analysis. The Cas9 tiling approach for target enrichment and sequencing is a valuable alternative to whole-genome sequencing for the assembly of ultra-long regions of interest, improving the accuracy of downstream genotype-phenotype association analysis.

Keywords: Cas9-tiling enrichment; de novo assembly; nanopore sequencing; pod-shattering; variant calling.

Figure 1

Sequencing data show high divergence between the Midas cultivar and the P. vulgaris reference genome within the pod-shattering region. (A) Integrative Genome Browser Visualization (IGV) of Midas Illumina whole-genome sequencing reads mapped to the pod-shattering region (Chr05:38,489,481–38,723,757). Alignments to the regions highlighted with black squares are magnified in panels (B,C) and represent the regions with the highest divergence. Each IGV shows the associated genotypability in the same region (% PASS at DP ≥ 5).

Figure 2

Sequencing of the pod-shattering region by combining Cas9 tiling with ONT sequencing. (A) Integrative Genome Browser Visualization (IGV) of ONT data mapping to the pod-shattering region (Chr05:38,489,481–38,723,757) after Cas9 tiling and ONT sequencing of Midas DNA. (B) Fraction of ONT reads and average coverage on each sub-ROI, the whole ROI, and whole genome (WG) after Cas9 tiling and ONT sequencing.

Figure 3

The pod-shattering region features multiple variations and a large deletion of 3 kbp overlapping a gene, when comparing the Midas cultivar and P. vulgaris reference genome. (A) The contig assembly based on Midas Cas9 tiling data (y-axis) was aligned to the pod-shattering region of the P. vulgaris reference genome (x-axis) using NUCmer. Alignments longer than 1 kbp were subsequently filtered and visualized with Dot viewer. Aligned dots highlight regions where the two sequences diverge, and the arrow shows a ~3-kbp deletion in Midas (magnified in panels (B,C)). (B) Integrative Genome Browser Visualization (IGV) of Cas9 assembly and Midas Illumina WGS reads aligned to the P. vulgaris reference genome (Chr05:38,557,292–38,570,137) revealing the detailed tracks of SNVs and SVs identified in the Cas9 assembly compared to the reference genome and the annotated gene overlapping the deletion marked in panel (A). (C) IGV of Midas Illumina WGS reads aligned to the Cas9 assembly (Chr05:38,195,825–38,203,228) in the region including the orthologous gene shown in panel B. Internal boxes show the genotypability (% PASS at ≥5 reads) of the two orthologous genes.

Figure 4

Comparison of sequences and mapped reads of orthologous genes in the P. vulgaris reference genome and Cas9 assemblies. (A) Percentage sequence identity of the orthologous genes and corresponding proteins derived from the P. vulgaris reference genome and Cas9 tiling assembly (excluding untranslated regions, which were not annotated de novo in the Cas9 assembly). (B) Number of Midas WGS Illumina reads mapping onto the orthologous genes from the P. vulgaris reference genome and Cas9 tiling assembly, after normalizing by each gene length. (C) Proportion of bases that can be genotyped (% PASS at ≥5 reads) in the orthologous genes from the P. vulgaris reference genome and Cas9 tiling assembly. *** p-value < 0.0001, paired t-test.

Figure 5

Comparison of de novo assemblies from Cas9 tiling and nanopore WGS data in the pod-shattering region. (A) Dot plot viewer of contig–contig alignment showing the nanopore WGS assembly on the x-axis and the Cas9 assembly on the y-axis. Divergent regions are highlighted with arrows. (B,C) Integrative Genome Viewer (IGV) visualization of the aligned Cas9-derived and nanopore WGS-derived contigs at the divergent regions. Blue boxes indicate repetitive regions annotated by RepeatMasker.