Telomere-to-telomere haplotype-resolved reference genome reveals subgenome divergence and disease resistance in triploid Cavendish banana

Abstract

Banana is one of the most important crops of the world. Cavendish-type bananas, which have a monospecific Musa acuminata origin (AAA), account for around half of the global banana production, thereby are of great significance for human societies. However, until now, the high-quality haplotype-resolved reference genome was still undecoded for banana cultivars. Here, we reported the telomere-to-telomere (T2T) and haplotype-resolved reference genome of 'Baxijiao' (Cavendish) consisting of three haploid assemblies. The sizes of the three haploid assemblies were estimated to be 477.16 Mb, 477.18 Mb, and 469.57 Mb, respectively. Although with monospecific origins, the three haploid assemblies showed great differences with low levels of sequence collinearity. Several large reciprocal translocations were identified among chromosomes 1, 4, and 7. An expansion of gene families that might affect fruit quality and aroma was detected, such as those belonging to sucrose/disaccharide/oligosaccharide catabolic processes, sucrose metabolic process, starch metabolic process, and aromatic compound biosynthetic process. Besides, an expansion of gene families related to anther and pollen development was observed, which could be associated with parthenocarpy and sterility of the Cavendish cultivar. Finally, much fewer resistance genes were identified in 'Baxijiao' than in M. acuminata, particularly in the gene clusters in chromosomes 3 and 10, providing potential targets to explore for molecular analysis of disease resistance in banana. This T2T haplotype-resolved reference genome will thus be a valuable genetic resource for biological studies, molecular breeding, and genetic improvement of banana.

Figure 1

Overview of the Cavendish reference genome. A The circos diagram of the ‘Baxijiao’ genome. The tracks represent the following elements (from outer to inner): (a) schematic representation of the 33 chromosome sequences, (b) density of centromeric repeats, (c) GC contents, (d) density of transposable elements, (e) density of the Gypsy elements, (f) density of the Copia elements, (g) density of genes. The innermost is syntenic relationships. B Hi-C interaction heatmap for the ‘Baxijiao’ genome. C HiFi coverage of the 33 pseudo-chromosomes. D Mapping rates of HiFi, Illumina and RNAseq reads in BXJ1, BXJ2, and BXJ3. E BUSCO assessments in BXJ1, BXJ2, and BXJ3. F Sequence collinearity among BXJ1, BXJ2, and BXJ3. Gray lines represent the collinearity blocks with length of 15 000 bp, while red lines represent the potential inversions.

Figure 2

Whole genome comparisons within ‘Baxijiao’ and between ‘Baxijiao’ and other species in monocots. A Syntenic analysis at the gene level. Highlighted lines represent reciprocal translocations. The other lines represent syntenic blocks. B Inferred phylogenetic tree and the expanded (+) or contracted (−) gene families in ‘Baxijiao’ and other species in monocots. Number of expanded and contracted gene families at branch (share between species/BXJ haploid assemblies) and leaf nodes (share in specific species/BXJ haploid assemblies) has been denoted. C The top enriched biological processes for expanded and contracted genes in ‘Baxijiao’.

Figure 3

The distribution of NLR loci in MAv4 and ‘Baxijiao’ genomes. Lines on chromosomes suggest that the loci are located in coding regions and intergenic regions, respectively. The syntenic relationships of the gene clusters in chromosome 3 and 10 are presented in the right panel.

Figure 4

Homoeolog expression bias in syntenic homoeolog triads. A Ternary plot showing relative expression levels of 17 531 syntenic triads (52 593 genes) in ‘Baxijiao’. Each circle represents a gene triad with a BXJ1, BXJ2, and BXJ3 coordinate suggesting the relative contribution of each homoeolog to the overall triad expression (an example is shown on the top left). The gray circles represent the balanced triads, while the colored circles in vertices correspond to the triads from the homoeolog-dominant groups, and the colored circles close to edges and between vertices correspond to the triads from the homoeolog-suppressed groups. Box plots indicate the relative contribution of each homoeolog based on triad assignment to the seven groups, including BXJ1D, BXJ2D, BXJ3D representing BXJ1, BXJ2 and BXJ3 dominant respectively, and BXJ1S, BXJ2S and BXJ3S represent BXJ1, BXJ2, and BXJ3 suppressed. B Contribution of each group of homoeolog expression bias across the 4 tissues and in the combined analysis. C Box plot of absolute TPM expression abundance for each haploid genome from the seven groups. D Violin plots of the Ka/Ks values for the seven categories. The balance category shows slightly lower Ka/Ks values than the homoeolog-dominant and homoeolog-suppressed groups. E The distribution of homoeolog expression bias genes in all 33 chromosomes.