Fern genomes elucidate land plant evolution and cyanobacterial symbioses

Abstract

Ferns are the closest sister group to all seed plants, yet little is known about their genomes other than that they are generally colossal. Here, we report on the genomes of Azolla filiculoides and Salvinia cucullata (Salviniales) and present evidence for episodic whole-genome duplication in ferns-one at the base of 'core leptosporangiates' and one specific to Azolla. One fern-specific gene that we identified, recently shown to confer high insect resistance, seems to have been derived from bacteria through horizontal gene transfer. Azolla coexists in a unique symbiosis with N₂-fixing cyanobacteria, and we demonstrate a clear pattern of cospeciation between the two partners. Furthermore, the Azolla genome lacks genes that are common to arbuscular mycorrhizal and root nodule symbioses, and we identify several putative transporter genes specific to Azolla-cyanobacterial symbiosis. These genomic resources will help in exploring the biotechnological potential of Azolla and address fundamental questions in the evolution of plant life.

Fig. 1. Genome size evolution in Salviniales.

a, Members of Salviniales have smaller genome sizes than other ferns (averaging 1C = 12 Gb). Two whole-genome duplication (WGD) events identified in this study were mapped onto the phylogeny, with divergence time estimates obtained from Testo and Sundue. b,c, Whole genomes were assembled from A. filiculoides (b) and S. cucullata (c). d,e, The genome of S. cucullata has substantially reduced levels of RNA (d) and DNA (e) transposons compared to A. filiculoides. Image in panel c courtesy of P.-F. Lu.

Fig. 2. Evolution of ethylene biosynthesis.

The ethylene-forming pathway involves the Yang cycle, where ACC is synthesized from S-adenosyl-methionine (SAM; also known as AdoMet) by ACC synthase. ACC oxidase then catalyses the conversion of ACC to ethylene. We found that ACC oxidase is unique to seed plants (green) and its origin probably drove the expansion of the ACC synthase gene family (orange; Supplementary Fig. 6) to create a regulated ethylene biosynthetic mechanism.

Fig. 3. The history of WGD in Azolla and Salvinia.

a, MAPS analysis identified two WGD events: one specific to Azolla (orange circle) and one predating the core leptosporangiates (green circle). The blue line illustrates the percentage of subtrees indicative of a gene duplication shared by the descendants at each node. The grey lines display the gene birth–death simulation results without WGD. The species divergence dates are from Testo and Sundue. b, Density plots from fitting Gaussian mixture models to K_s distributions estimated from pairs of syntenic paralogues within the Azolla and Salvinia genomes, as well as of syntenic orthologues between Azolla and Salvinia. c, Examples of synteny between Azolla and Salvinia genomic regions. The left and right panels display a 2:1 and 2:2 syntenic relationship between Azolla and Salvinia regions, respectively. Each subpanel represents a genomic region in Azolla or Salvinia, with gene models on both strands shown above and below the dashed line. High-scoring sequence pairs (HSPs) in protein-coding sequences are marked by short vertical bars above the gene models. Selected HSP links between genomic regions are depicted as coloured lines crossing the subpanels, whereas others (for example, the HSP links between the two Azolla genomic regions in the left panel) are left out for clarity. Collinear series of HSPs across genomic regions indicates a syntenic relationship between the regions concerned. Genomic regions conserved in duplicate after the WGD that occurred prior to the divergence between Azolla and Salvinia should show a 2:2 syntenic relationship, whereas regions conserved in duplicate after the Azolla-specific WGD should show a 2:1 syntenic relationship with Salvinia regions. The left and right panels can be regenerated at https://genomevolution.org/r/ujll and https://genomevolution.org/r/ukys, respectively.

Fig. 4. Origin of a fern insecticidal protein.

Phylogenetic analysis of the chitin-binding domain Pfam PF03067 shows that the fern Tma12 insecticidal protein was probably derived from bacteria through an ancient HGT event. The numbers above the branches are bootstrap (BS) support values (BS = 100 is omitted), and the thickened branches indicate BS > 70. The tree is rooted based on the result from a broader phylogenetic analysis of PF03067 and PF08329 (Supplementary Data). The pink star denotes the sequence from the S. cucullata genome.

Fig. 5. Azolla–cyanobacterial symbiosis.

a, The cyanobiont phylogeny largely mirrors the host species phylogeny, indicating a convincing cospeciation pattern between the two partners. All nodes received a maximum likelihood bootstrap support of 100%, and for the host phylogeny, all nodes also received a local posterior probability of 1.0 from the ASTRAL analysis. Both the nuclear and the plastome data sets gave the same topology for the host, and the branch lengths shown here were from the plastome tree. Scale bars represent 0.01 substitutions per site. b, The CSP genes were lost in the Azolla and Salvinia genomes (empty boxes), whereas orthologues can be found in other fern transcriptomes (red boxes). *Arabidopsis lacks the CSP genes and does not have AM symbiosis. c, Cyanobionts have a large effect on the Azolla transcriptome. d, The Azolla transcriptome responds to nitrogen starvation more significantly when cyanobionts are absent than when they are present. PC, principal component. e, Candidate genes involved in nutrient transport and communication with cyanobionts.