Convergent evolution of strigolactone perception enabled host detection in parasitic plants

See allHide authors and affiliations

Science  31 Jul 2015:
Vol. 349, Issue 6247, pp. 540-543
DOI: 10.1126/science.aab1140

How plant parasites evolved to find hosts

The seeds of parasitic plants need to be able to sense their host's presence to germinate at the correct time and in the correct place. This is done through the detection of plant hormones, strigolactones. However, the origin of this sensory system is unknown. Conn et al. investigated the diversity of strigolactone receptors in multiple lineages of parasitic plants and their close relatives. They found a greater copy number and accelerated evolution in parasitic plants as compared with nonparasitic relatives. Functional analyses of parasitic plant strigolactone receptors in transgenic Arabidopsis suggested that convergent evolution has occurred to allow the parasitic plants to detect their hosts.

Science, this issue p. 540


Obligate parasitic plants in the Orobanchaceae germinate after sensing plant hormones, strigolactones, exuded from host roots. In Arabidopsis thaliana, the α/β-hydrolase D14 acts as a strigolactone receptor that controls shoot branching, whereas its ancestral paralog, KAI2, mediates karrikin-specific germination responses. We observed that KAI2, but not D14, is present at higher copy numbers in parasitic species than in nonparasitic relatives. KAI2 paralogs in parasites are distributed into three phylogenetic clades. The fastest-evolving clade, KAI2d, contains the majority of KAI2 paralogs. Homology models predict that the ligand-binding pockets of KAI2d resemble D14. KAI2d transgenes confer strigolactone-specific germination responses to Arabidopsis thaliana. Thus, the KAI2 paralogs D14 and KAI2d underwent convergent evolution of strigolactone recognition, respectively enabling developmental responses to strigolactones in angiosperms and host detection in parasites.

The Orobanchaceae plant family comprises thousands of species that parasitize other plants through a haustorial connection to a host root. Several obligate parasites in this family are noxious weeds that can cause complete crop loss, affecting millions of smallholder farmers and reducing yields by billions of U.S. dollars each year (1, 2). Their seeds can lie dormant in soil for many years, until the detection of strigolactones (SLs) exuded by a nearby host triggers germination. This essential adaptation may be exploited to combat parasite infestations by inducing suicidal germination or modifying SL output in crops. To aid the development of these strategies, we set out to understand how parasites sense SLs.

SLs are plant hormones that control shoot branching, root architecture, cambial growth, and senescence, but also serve as extraorganismal signals in soil that recruit symbioses with arbuscular mycorrhizal fungi (36). SL responses in plants require the F-box protein MORE AXILLARY GROWTH2 (MAX2) and the α/β-hydrolase DWARF14 (D14) (3). MAX2 also mediates responses to karrikins (KARs), a family of butenolide compounds found in smoke that trigger the germination of many species after fire but do not promote parasite germination (7, 8) (fig. S1). KAR responses in Arabidopsis thaliana require KARRIKIN-INSENSITIVE2 (KAI2), a paralog of D14. AtKAI2 is required for normal seed germination and seedling growth, but D14 and SL biosynthesis genes are not (9).

D14 and KAI2 are likely receptors that require conformational changes induced by enzymatic activity for signal transduction. D14 binds and hydrolyzes SL, and KAI2 binds KAR1; catalytic residue mutations abolish their function (1015). The crystal structures of D14 and KAI2 show nearly identical topologies, but differences are found in the ligand-binding pocket shapes (1116). Only KAI2 occurs in basal plant lineages, and the appearance of D14 in higher plant genomes coincides with that of MAX2 targets that control branching (3, 9). Therefore, we infer that SL recognition may not have been the ancestral role of KAI2 but probably evolved after duplication of KAI2 in the paralog now called D14.

Host recognition might have arisen in the parasitic Orobanchaceae through adaptation of the MAX2-dependent germination control mechanism. MAX2 itself is well conserved in parasite genomes, and a MAX2 ortholog from the parasite Striga hermonthica can rescue an A. thaliana max2 mutant (17). We hypothesized that in parasites, KAI2 evolved ligand specificity for SLs and/or that the SL receptor D14 gained a role in germination.

We investigated D14 and KAI2 evolution in 10 species that represent the full range of parasitism in the Orobanchaceae. We used next-generation shotgun genome sequencing and de novo genome assembly algorithms to identify D14 and KAI2 genes in four obligate holoparasites (Orobanche minor, O. cernua, O. cumana, and Conopholis americana) and two facultative hemiparasites (Phtheirospermum japonicum and Triphysaria versicolor) (18). We also searched an expressed sequence tag database of S. hermonthica and de novo transcriptome assemblies of five parasites and a basal nonparasite. Gene fragments matching D14 and KAI2 in the assemblies were used to generate full-length coding sequences (data files S1 and S2). We verified the predicted sequences of 29 genes from six parasite species by Sanger sequencing (18). To make comparisons to nonparasitic genomes, we also mined available dicot genome assemblies and de novo transcriptome assemblies of 18 species in the Orobanchaceae-containing order Lamiales. We restricted further analyses to a conservatively defined data set of 53 D14 and 144 KAI2 sequences from 55 species (tables S1 and S2).

We identified at most one D14 copy in parasitic Orobanchaceae and their nonparasitic relatives in the Lamiales, similar to the number in nonparasitic dicots (fig. S2 and table S3). The D14 genes in the Lamiales formed a monophyletic clade (Fig. 1A). In contrast, most parasite genomes had more copies of KAI2 than did nonparasite Lamiales species (5.6 ± 1.2 versus 1.8 ± 0.19; Wilcoxon rank sum test, Embedded Image 9.4, P = 0.0022); the nonweedy parasites C. americana and O. fasciculata were exceptions. The KAI2 genes in the Lamiids (Lamiales, Gentianales, and Solanales) formed a monophyletic clade that we divided into conserved (KAI2c), intermediate (KAI2i), and divergent (KAI2d) subclades (Fig. 1B). Most Lamiids [28 out of 33 species (28/33 spp.)] had a single basal KAI2c sequence. When present, KAI2i paralogs were also usually limited to a single copy (15/17 spp.), but KAI2i paralogs were not detected in all Lamiids or in any of the six obligate holoparasites. The KAI2d clade, however, was parasite-specific and contained the majority of KAI2 paralogs (fig. S4).

Fig. 1 D14 and KAI2 evolution in dicots.

(A) D14 Bayesian phylogeny. Clades are designated P (parasitic Orobanchaceae) and N (nonparasitic Lamiales). (B) KAI2 Bayesian phylogeny. Clades are designated conserved (C, KAI2c), intermediate (I, KAI2i), and divergent (D, KAI2d). Physcomitrella patens KAI2 was used as an outgroup for both analyses. Uncollapsed phylogenies are shown in figs. S3 and S4.

We tested for evidence of relaxed purifying selection or positive selection that may have enabled subfunctionalization or neofunctionalization of D14 and KAI2 in parasites. We analyzed codon substitution patterns across the phylogeny by allowing the rate of protein evolution (ω = dN/dS) to vary across clades (tables S4 to S7). D14 was under strong purifying selection in dicots that was relaxed in the parasitic Orobanchaceae (ω0 = 0.07, ωP = 0.15). Different strengths of purifying selection were supported for the three KAI2 subclades in Lamiids. Purifying selection was strongest for KAI2cC = 0.07), comparable to other dicots for KAI2i0 = 0.11, ωI = 0.10), and weakest for KAI2dD = 0.27). Although ωD was not indicative of recurrent positive selection (i.e., ω > 1), the elevated ωD value could result if positive selection acted only on a subset of KAI2d codons or for a short period of time after the duplication event.

We next investigated whether KAI2 proteins in parasites had evolved structural changes that suggest altered ligand specificities. Because of high sequence identity (54 to 80%), we were able to generate homology models of KAI2 from parasites and related nonparasitic species using a KAR1-bound AtKAI2 structure as the template (15). Parasite KAI2 models were compared to AtKAI2-KAR1 and rice D14 (OsD14) bound to a byproduct of SL hydrolysis (12) (Fig. 2, A and B; fig. S5; table S8). Models of KAI2c proteins had similar substrate-binding cavities to AtKAI2 in terms of volume and shape. KAI2i substrate-binding cavities were predicted to have an intermediate morphology to AtKAI2 and OsD14, in which two adjacent pockets are joined. In contrast, KAI2d models had substrate-binding cavities that were larger than those of KAI2c and KAI2i (Fig. 2C; P < 0.01, Student’s t test), and appeared more superficially similar to those of OsD14 than AtKAI2. This structural convergence suggests that the parasite-specific KAI2d clade may have evolved the ability to recognize SL.

Fig. 2 Homology models of parasite KAI2.

(A) Ligand-binding pockets of AtKAI2 [left, Protein Data Bank (PDB) entry 4JYM] and OsD14 (right, PDB entry 3WIO) (12, 15). KAR1 and a hydroxy D-ring (D-OH) product of GR24 hydrolysis are indicated. OsD14 positions are offset by a 50–amino acid leader sequence that is absent in AtD14. (B) Homology models of KAI2 protein sequences in S. hermonthica (Sh) and P. aegyptiaca (Pa). Cavities within the protein models are shown as a semitransparent surface. Catalytic residues (S95, D217, and H246) and residues highlighted in (D) are shown in stick representation. (C) Box plots of ligand-binding cavity volumes in AtKAI2, OsD14, and KAI2 models here and in fig S5. (D) Amino acid frequency plots at equivalent positions to AtKAI2 residues, in the three KAI2 clades in Lamiids and in D14 in dicots.

The stratification of parasite KAI2 substrate-binding pockets into AtKAI2-like, OsD14-like, and somewhere in between is largely due to differences at a few highly conserved amino acids. Most prominently, the hydroxyl group of Y124 in the AtKAI2 crystal structure occludes a pocket adjacent to the substrate-binding cavity. Smaller amino acid substitutions at this position prevent the bisection of these two pockets, providing a larger substrate-binding cavity (Fig. 2A). Y124 is conserved in all 34 KAI2c proteins in Lamiids, and in 78% of KAI2 proteins in other dicots. The majority of KAI2i proteins in Lamiids (16/19) have a Y124F substitution, as seen in D14. Substitution of this residue with even smaller hydrophobic amino acids is typically observed in KAI2d (Fig. 2D). Other conserved residues that contribute to the morphology or chemical characteristics of the substrate-binding pocket are highly variable in KAI2d, supporting the hypothesis of altered ligand specificity in this clade (Fig. 2D).

We tested the ligand specificity of four KAI2 paralogs from S. hermonthica and three from P. aegyptiaca by functional complementation of an A. thaliana kai2 mutant, which has increased seed dormancy and no germination response to KARs or the synthetic SL rac-GR24 (9). Two transgenes, ShKAI2c and ShKAI2d2, had no effect on germination and were presumed nonfunctional in A. thaliana. The remaining transgenes gave three types of germination responses (Fig. 3B). PaKAI2c rescued the kai2 dormancy phenotype but did not confer responses to either KARs or rac-GR24, despite its predicted structural similarity to AtKAI2. The kai2 dormancy phenotype may result from the inability to detect an unknown, endogenous germination-promoting signal (10). If so, PaKAI2c may be specific for this signal. The intermediate-clade gene ShKAI2i conferred germination responses to KARs but not to rac-GR24. Although this activity was consistent with the KAR receptor role of AtKAI2, it was unexpected as S. hermonthica is insensitive to karrikins (fig. S1). Most notably, three KAI2d genes conferred a strong germination response to rac-GR24 alone (Fig. 3B).

Fig. 3 Cross-species complementation assays of parasite KAI2.

(A) Chemical structures of KAR1, KAR2, and four GR24 stereoisomers. (B) Seed germination after 6 days in 16 hours light, 21°C, with 1 μM KAR1, KAR2, and rac-GR24 treatments. Transgenes were expressed in the null kai2-2 mutant background (Ler ecotype) under control of an AtKAI2 promoter. (C) Seed germination after 5 days in 16 hours light, 21°C, with 1 μM GR24 stereoisomer treatments. Mean germination ± SE is shown (n = 4 independent seed batches per genotype, 50 seeds tested per seed batch). *P < 0.01, Tukey-Kramer HSD test, comparison to control treatment for each genotype.

Although rac-GR24 is commonly used as a synthetic SL, it is a racemate of natural (2′R) GR245DS and unnatural (2′S) GR24ent-5DS stereoisomers that respectively activate D14 and KAI2 signaling in A. thaliana seed and seedlings (19). Positive responses to rac-GR24 can be misinterpreted as SL responses when GR24ent-5DS is actually the active signal. We therefore tested germination responses to four enantiomers of GR24 (Fig. 3, A and C). Two of the three rac-GR24–responsive KAI2d conferred stronger responses to the natural SL stereoisomers GR245DS and GR244DO than to their unnatural enantiomers GR24ent-5DS and GR24ent-4DO. ShKAI2d1 responded to all four stereoisomers.

Thus, KAI2d paralogs have gained structural similarity to D14 and the ability to respond to SL. This example of convergent molecular evolution probably arose in parasites via a duplication of KAI2 before the divergence of the Lamiales and Solanales. During the transition to parasitism in the Orobanchaceae, further KAI2 duplication and relaxed selection and/or bursts of positive selection on these paralogs formed the SL-responsive KAI2d clade (Fig. 4). The presence of KAI2d in facultative hemiparasites, which do not require a host, suggests that the capacity for SL perception preceded obligate parasitism.

Fig. 4 Model of KAI2 and D14 evolution.

KAI2 homologs are found in charophyte algae and other basal lineages, but their functions and ligands are unknown. D14 probably arose from a duplication of KAI2 before the evolution of seed plants (spermatophytes). AtKAI2 recognizes KAR and may recognize an endogenous KAI2 ligand (KL). Duplication of KAI2 after the evolution of Lamiids produced KAI2c and KAI2i paralogs in the Lamiales and Solanales. KAI2c may recognize KL, and KAI2i may recognize KAR. Further duplication events in the parasitic Orobanchaceae led to a fast-evolving clade of KAI2d that recognize SL.

AtD14 did not rescue the kai2 mutant. This suggests that changes in D14 expression patterns would not be sufficient to evolve SL-specific germination in A. thaliana, perhaps because KAI2 and D14 interact with different downstream signaling partners in the SMAX1/D53 family (10). The evolution of SL specificity in KAI2 may be a simpler path to host recognition than neofunctionalization of D14.

The antagonistic coevolution of hosts and parasites may drive diversification of SL molecules in hosts and SL detection mechanisms in parasites. Supporting this idea, (i) at least 20 SLs and other host-derived germination stimulants with butenolide or lactone moieties have been identified (20, 21), (ii) SL profiles vary within and across host species (2123), and (iii) parasite species also vary considerably in their germination responses to different host root exudates and SLs (24, 25). We hypothesize that multiple KAI2d genes in parasite genomes may enable diversified stimulant recognition, which in turn influences host range and buffers against changes in host availability. Because we only observed one or two KAI2 genes in C. americana and O. fasciculata, which are parasites of oaks and Artemisia spp., respectively, it may be that KAI2 diversity is less critical for parasites of perennial hosts.

We have identified 29 KAI2d genes from five weedy parasites that are likely to have roles in host recognition. This discovery will enable high-throughput in silico and in vitro screening of chemical libraries for KAI2d agonists to control parasite infestations. It will also aid investigations of how parasites rapidly evolve new host specificities and become virulent agricultural weeds.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Tables S1 to S10

References (2644)

Data Files S1 to S4


  1. Information on materials and methods is available on Science Online.
  2. ACKNOWLEDGMENTS: Genome sequencing data are deposited in Sequence Read Archives at the National Center for Biotechnology Information (accession no. SRP056321) and the DNA Data Bank of Japan (accession no. DRA003500). D14 and KAI2 sequences and alignments are in supplementary material data files S1 to S4. Supported by the University of Georgia (UGA) Research Foundation, NSF grant IOS-1350561 (D.C.N.); an NSF Graduate Research Fellowship (C.E.C.); NSF grant DEB-1149350 (K.A.D.); National Institute of Food and Agriculture grant no. 135997, NSF grant DBI-0701748, and NSF grant IOS-1238057 (J.H.W.); and Ministry of Education, Culture, Sports, Science and Technology (MEXT) KAKENHI grant nos. 221S0002, 25711019, 25128716, and 24228008 (S.Y., K.S.). We thank G. Ka-Shu Wong (U. of Alberta), C. dePamphilis (Penn State), J. Leebens-Mack (UGA), and the OneKP consortium ( for Lamiales sequences. OneKP is funded by the Alberta Ministry of Innovation and Advanced Education, Alberta Innovates Technology Futures Innovates Centres of Research Excellence, Musea Ventures, and BGI–Shenzhen. B. Schmitz (UGA), M. French (Callaway Gardens), and A. Scaffidi and M. Waters (U. of Western Australia) provided materials.
View Abstract

Navigate This Article