Convergent evolution of plant prickles by repeated gene co-option over deep time

Abstract

An enduring question in evolutionary biology concerns the degree to which episodes of convergent trait evolution depend on the same genetic programs, particularly over long timescales. In this work, we genetically dissected repeated origins and losses of prickles-sharp epidermal projections-that convergently evolved in numerous plant lineages. Mutations in a cytokinin hormone biosynthetic gene caused at least 16 independent losses of prickles in eggplants and wild relatives in the genus Solanum. Homologs underlie prickle formation across angiosperms that collectively diverged more than 150 million years ago, including rice and roses. By developing new Solanum genetic systems, we leveraged this discovery to eliminate prickles in a wild species and an indigenously foraged berry. Our findings implicate a shared hormone activation genetic program underlying evolutionarily widespread and recurrent instances of plant morphological innovation.

Fig. 1.. Prickles evolved convergently across vascular plants and were lost repeatedly in the spiny Solanum lineage.

(A) Phylogeny, from (5), and corresponding images of representative vascular plants that independently evolved prickles. Number in parentheses indicates number of identified independent evolutionary origins of prickles (B) Phylogenetic tree [adapted from (10, 11)] of the spiny Solanum (subclades Wendlandii, Nemorense, and Leptostemonum) with species having lost prickles highlighted in red. Representative images of narrow and broad-based prickle morphologies are shown. (C and D) Images of Solanum taxa that have lost prickles captured from living (C) and herbarium (D) collections. Numbers correspond to species shown in (B).

Fig 2.. Losses of prickles in three domesticated Solanum species are caused by independent mutations in a LOG cytokinin biosynthetic gene.

(A) Fine-mapping of pl in a Brinjal eggplant (S. melongena) x wild progenitor species (S. insanum) mapping population. (B) Genome sequencing and chromosome-scale assemblies of two African eggplants, the Scarlet eggplant (S. aethiopicum) and the Gboma eggplant (S. macrocarpon) reveals synteny of the pl locus. Genome summary statistics are indicated. (C) Independent mutations in a LOG gene in the pl interval in all three prickleless crop species. (D) Mis-splicing of PL transcripts caused by the pl mutations in Bringal eggplant pl (Smelpl) and Gboma eggplant pl (Smacpl) confirmed by RT-PCR. SinsPL-IL denotes an introgression of S. insanum PL into the Brinjal eggplant genomic background. (E) QTL-Seq identifies two loci that independently cause the prickleless phenotype in Gboma eggplant. (F) Phenotypes resulting from CRISPR-Cas9 genome editing of SaetPL in a prickled S. aethiopicum accession. Arrowheads indicate prickles.

Fig. 3.. Mutations in PL are associated with prickle suppression across the spiny Solanum.

PL variants with strong probable deleterious effects on gene function identified in prickle-suppressed taxa but not in closely-related prickled sister taxa. Mutations are numbered and shown along with their corresponding species name and sample source in the table below. In the tables, bold text indicates cultivated species, (*) indicates that genotyping was performed on archival herbarium samples, (†,‡,§) indicate species pairs that share identical but not necessarily ancestral mutations.

Fig. 4.. Losses of convergently evolved prickles across angiosperms are associated with LOG mutations.

(A to D) Instances of prickle suppression in angiosperms associated with LOG mutations depicted in corresponding LOG gene diagrams. (A) Images of rice and barley WT inflorescences. Arrowheads indicate awns, which are shown for WT and mutant genotypes (rice, laba1; barley, rough awn1) by SEM. (B) Images of jujube trees, fruits, and stipular spines (arrowheads). Two less spiny cultivated varieties harbor two independent LOG mutations. (C) The ornamental giant spider flower (pictured) carries a mutated LOG gene in the sequenced ‘Purple Queen’ cultivar. Cultivated varieties bear fewer smaller prickles (arrowheads) than wild varieties, as reflected in herbarium samples. (D) (Left) Loss of prickles in rose maps to a ~2.5 Mb interval harboring a LOG gene with severely reduced expression in the prickleless cultivar relative to the prickled cultivar. Syntenic genes within the mapping interval of the prickled ‘Old Blush’ and prickleless ‘Basye’s Thornless’ parental lines are shown in black. Read pileups show average LOG expression in leaves of the parental genotypes (N = 3). (Right) VIGS targeting of the candidate LOG gene leads to suppression of prickles in an ornamental rose hybrid. (E) Protein-based phylogenetic tree of the Arabidopsis LOG1 orthogroup defined by Orthofinder, from the indicated asterid (red), rosid (black), and monocot (purple) species. LOGs encoded by genes with mutations in prickle-suppressed taxa are indicated by arrowheads.

Fig. 5.. The Solanum PL gene was co-opted from an ancestral gene duplication event enabling non-pleiotropic editing of PL for crop improvement.

(A) Whole-plant and fruit images of the prickled wild species Forest nightshade (S. prinophyllum, top) and its close foraged berry-producing relative Desert raisin (S. cleistogamum, bottom). Red-shaded region in map insets indicates approximate species ranges in Australia based on reported observations (http://www.flora.sa.gov.au/). (B) Genome sequencing and chromosome-scale assemblies of Forest nightshade and Desert raisin reveals that PL interval synteny is conserved in Brinjal eggplant and tomato (S. lycopersicum). Genome summary statistics are indicated. (C) Heatmap depicting the predictability of identifying cross-species co-expressed genes among cross-species pairs of LOG homologs based on their respective co-expression relationships in tomato and Arabidopsis. A higher Area Under the Receiver Operating Characteristic (AUROC) curve score indicates LOG homologs with increased conservation of their corresponding orthologous co-expressed genes. (D) Coding-sequence based maximum-likelihood phylogenetic tree of Solanum PL orthologs, their closely related paralog LOG1a, and AthaLOG1 in comparable tissue types. Heatmap shows expression in matched tissues. (E) CRISPR-Cas9 gene editing strategy and resulting mutant alleles generated in Forest nightshade, Desert raisin, and tomato. (F to H) Phenotypes of WT and gene edited pl null mutants in Forest nightshade (F), Desert raisin (G), and tomato (H). Prickles are nearly completely suppressed (Forest nightshade) and eliminated (Desert raisin) obvious pleiotropic consequences. In tomato where PL was not co-opted for prickle development, Slycpl^CR mutants resemble wild type. (I) Evolutionarily-informed trait analysis enables rapid and expedient removal of prickles for improved harvestability in Solanum crops.