The evolutionary history of LysM-RLKs (LYKs/LYRs) in wild tomatoes

Abstract

Background: The LysM receptor-like kinases (LysM-RLKs) are important to both plant defense and symbiosis. Previous studies described three clades of LysM-RLKs: LysM-I/LYKs (10+ exons per gene and containing conserved kinase residues), LysM-II/LYRs (1-5 exons per gene, lacking conserved kinase residues), and LysM-III (two exons per gene, with a kinase unlike other LysM-RLK kinases and restricted to legumes). LysM-II gene products are presumably not functional as conventional receptor kinases, but several are known to operate in complexes with other LysM-RLKs. One aim of our study was to take advantage of recently mapped wild tomato transcriptomes to evaluate the evolutionary history of LysM-RLKs within and between species. The second aim was to place these results into a broader phylogenetic context by integrating them into a sequence analysis of LysM-RLKs from other functionally well-characterized model plant species. Furthermore, we sought to assess whether the Group III LysM-RLKs were restricted to the legumes or found more broadly across Angiosperms.

Results: Purifying selection was found to be the prevailing form of natural selection within species at LysM-RLKs. No signatures of balancing selection were found in species-wide samples of two wild tomato species. Most genes showed a greater extent of purifying selection in their intracellular domains, with the exception of SlLYK3 which showed strong purifying selection in both the extracellular and intracellular domains in wild tomato species. The phylogenetic analysis did not reveal a clustering of microbe/functional specificity to groups of closely related proteins. We also discovered new putative LysM-III genes in a range of Rosid species, including Eucalyptus grandis.

Conclusions: The LysM-III genes likely originated before the divergence of E. grandis from other Rosids via a fusion of a Group II LysM triplet and a kinase from another RLK family. SlLYK3 emerges as an especially interesting candidate for further study due to the high protein sequence conservation within species, its position in a clade of LysM-RLKs with distinct LysM domains, and its close evolutionary relationship with LYK3 from Arabidopsis thaliana.

Keywords: Phylogenetics; Plant immunity; Population genetics; Solanum; Symbiosis.

Fig. 1

LysM-RLK phylogeny and functions. Maximum likelihood phylogeny and functions of canonical LysM-RLKs from Solanum lycopersicum (Sl), Arabidopsis thaliana (At), Lotus japonicus (Lj), Oryza sativa (Os), and Medicago truncatula (Mt). The phylogeny and 500 bootstrap replicates were inferred using RAxML under the WAG model with empirical frequencies and seed values of 100. The species of gene origin is given by the first two letters of the name given on the phylogeny. The phylogeny was rooted using the method of Minimal Ancestor Deviation [27]. The scale bar indicates amino acid substitutions per site. Gene functions are indicated: defense against fungi (F), bacteria (B), and oomycetes (O) and symbiosis with rhizobia (R) and mycorrhiza (M). LysM-RLKs form three clades. Red and black arcs indicate groups of proteins with distinct LysM domain sequences. Functions verified by mutation phenotypes are indicated by black circles. Functions inferred by differential expression are indicated by gray circles. Citations for sources of functional information are shown in brackets

Fig. 2

Phylogeny of canonical LysM-RLKs and CXC-motif-containing BlastP hits of LjLYS20. a Phylogeny of the LysM domain sequences only shows Group III and putative Group III sequences forming a clade with Group II. The phylogeny and 500 bootstrap replicates were inferred using RAxML under the WAG model with empirical frequencies and seed values of 100 and rooted using the method of Minimal Ancestor Deviation [27]. b Phylogeny of the kinase domains only shows Group I and Group II forming a clade together, while Group III and putative Group III sequences form another clade. The phylogeny and 500 bootstrap replicates were inferred using RAxML under the JTT model with empirical frequencies and seed values of 100 and rooted using the method of Minimal Ancestor Deviation [27]. The scale bar indicates amino acid substitutions per site

Fig. 3

Unrooted phylogeny of individual LysM-RLK domains. Phylogeny of amino acid sequences of each of the three LysM-RLK protein domains from each of the canonical LysM-RLKs of Solanum lycopersicum (Sl), Arabidopsis thaliana (At), Lotus japonicus (Lj), Oryza sativa (Os), and Medicago truncatula (Mt). Each gene is represented three times in the tree, once for each individual LysM domain (see emphasis on SlLYK3 in phylogeny), with color-coding by domain position. The phylogeny and 500 bootstrap replicates were inferred using RAxML under the WAG model with empirical frequencies and seed values of 100. The Log-likelihood was −13,081. The scale bar indicates amino acid substitutions per site. Sequences from the first, second, and third LysM domains generally cluster in clades with others of the corresponding LysM domain, but the first and third domains do not form separate clades. Especially long branches subtend the groups of the first and second domains of genes of interest, referred to as the red clade and the black clade. Domain sequences from the red clade are highlighted in red: AtLYK3, SlLYK3, LjLYS4, and LjLYS5. Those from the black clade are highlighted in black: OsLYK1, MtLYK10, SlLYK14, and LjLYS3. The third domain of OsLYK1 and first domain of LjLYS3 are separated from the corresponding domains of the second group

Fig. 4

Amino acid logos of LysM-RLK domains. Logos of LysM-RLK LysM domains, with those of the red and black clades (AtLYK3, SlLYK3, LjLYS4, LjLYS5, MtLYK10, OsLYK1, LjLYS3, and SlLYK14) computed separately. The third domains of both sets of sequences share conserved amino acids, while first and second domains of the two sequence sets share few conserved amino acids

Fig. 5

Phylogeny of wild tomato SlLYK8 and SlLYK9 orthologs with intact kinases. This phylogeny includes BLAST hits for Solanum lycopersicum LysM-RLKs SlLYK8 and SlLYK9 which extend past the point of SlLYK8 truncation. The phylogeny and 500 bootstrap replicates were inferred using RAxML under the GTRGAMMA model and seed values of 100 and rooted using the method of Minimal Ancestor Deviation [27]. The scale bar indicates nucleotide substitutions per site. Three sequences from Solanum peruvianum and Solanum chilense with intact kinase domains share more recent ancestry with SlLYK8 than with SlLYK9 and show the greatest percent identity with SlLYK8