Long-read sequencing uncovers the adaptive topography of a carnivorous plant genome

Abstract

Utricularia gibba, the humped bladderwort, is a carnivorous plant that retains a tiny nuclear genome despite at least two rounds of whole genome duplication (WGD) since common ancestry with grapevine and other species. We used a third-generation genome assembly with several complete chromosomes to reconstruct the two most recent lineage-specific ancestral genomes that led to the modern U. gibba genome structure. Patterns of subgenome dominance in the most recent WGD, both architectural and transcriptional, are suggestive of allopolyploidization, which may have generated genomic novelty and led to instantaneous speciation. Syntenic duplicates retained in polyploid blocks are enriched for transcription factor functions, whereas gene copies derived from ongoing tandem duplication events are enriched in metabolic functions potentially important for a carnivorous plant. Among these are tandem arrays of cysteine protease genes with trap-specific expression that evolved within a protein family known to be useful in the digestion of animal prey. Further enriched functions among tandem duplicates (also with trap-enhanced expression) include peptide transport (intercellular movement of broken-down prey proteins), ATPase activities (bladder-trap acidification and transmembrane nutrient transport), hydrolase and chitinase activities (breakdown of prey polysaccharides), and cell-wall dynamic components possibly associated with active bladder movements. Whereas independently polyploid Arabidopsis syntenic gene duplicates are similarly enriched for transcriptional regulatory activities, Arabidopsis tandems are distinct from those of U. gibba, while still metabolic and likely reflecting unique adaptations of that species. Taken together, these findings highlight the special importance of tandem duplications in the adaptive landscapes of a carnivorous plant genome.

Keywords: Utricularia; carnivorous plant; evolution; plant genomics; polyploidy.

Fig. 1.

A chromosome-scale view of the architecture of the U. gibba genome. (A) Gene density, TE density tracks, telomeres, and the locations of CRM centromeric retrotransposon sequences are shown for all U. gibba contigs >1 Mb in size. Four complete chromosomal contigs are shown in blue, and partial chromosomes that have at least one end with telomere sequence are shown in orange. Putative centromeric regions are visible as peaks of increased TE density and decreased gene density. Most CRMs are localized at putative centromeric regions. (B) MUMmer (82) pairwise dot-plot alignment of contigs 0 and 22, which represent complete chromosomes. Blue and purple dots indicate hits on each DNA strand, respectively. Putative centromeric regions of strong sequence similarity are apparent as a densely hit square.

Fig. 2.

Syntenic relationships among V. vinifera, S. lycopersicum, and U. gibba regions containing tandemly duplicated cysteine protease genes. Some parts of these tandem arrays clearly preexisted in U. gibba’s prepolyploid ancestral genomes, with further tandem duplications having occurred since those events, together increasing functional potential for U. gibba’s carnivory. A typical ancestral region in Vitis can be traced to up to three regions in Solanum (through the latter's genome triplication) and up to eight regions in U. gibba (where as many as three WGDs are possible). Red connecting lines highlight matching cysteine proteases in the selected regions; genes otherwise syntenic are shown in gray.

Fig. 3.

Molecular and structural evolutionary analysis of U. gibba cysteine proteases suggests adaptive protein evolution accompanying WGD and tandem duplication events. (A) Best-scoring tree from maximum-likelihood based searches, with bootstrap support (BS) values ≥50 indicated at branches. Symbols on branches indicate significant evidence for positive selection (orange stars), divergent selection (green circles), or asymmetrical sequence evolution (purple hexagons) as determined using PAML (83) (SI Appendix, Dataset S10). The heatmap above the phylogeny shows trap-dominant expression of particular homologs in U. gibba, based on trap, shoot, and inflorescence transcriptome data (47) (SI Appendix, Dataset S2). Note that two tandem duplicates (g1 and g2) were repredicted at locus utg699.g19345. (B) The protein homology surface model for the catalytic domain of utg699.g19348 (encoded by the gene annotated by an arrow in A; based on the Venus flytrap [D. muscipula] enzyme structure (77)) shows that some residues under positive selection lie within or near the substrate-binding cleft. The cleft is depicted in yellow, and amino acid sites identified as under positive selection are indicated in red or cyan. Three (E24, V69, and S160) amino acid sites under positive selection (BEB confidence >0.82, Bonferroni-corrected P < 0.0015) are within five amino acids of known D. muscipula functional residues, where they line the substrate-binding cleft (red). (C) Plot of utg699.g19348 amino acid sites under positive selection, with colors corresponding to specific sites in the surface model (SI Appendix, Fig. S4B).