The karrikin response system of Arabidopsis



Arabidopsis thaliana provides a powerful means to investigate the mode of action of karrikins, compounds produced during wildfires that stimulate germination of seeds of fire-following taxa. These studies have revealed close parallels between karrikin signalling and strigolactone signalling. The two perception systems employ similar mechanisms involving closely related α/β-fold hydrolases (KAI2 and AtD14) and a common F–box protein (MAX2). However, karrikins and strigolactones may be distinguished from each other and elicit different responses. The karrikin response requires a newly discovered protein (SMAX1), a homologue of rice protein D53 that is required for the strigolactone response. Mutants defective in the response to karrikins have seeds with increased dormancy, altered seedling photomorphogenesis and modified leaf shape. As the karrikin and strigolactone response mechanisms are so similar, it is speculated that the endogenous signalling compound for the KAI2 system may be a specific strigolactone. However, new results show that the proposed endogenous signalling compound is not produced by the known strigolactone biosynthesis pathway via carlactone. Structural studies of KAI2 protein and its interaction with karrikins and strigolactone analogues provide some insight into possible protein–ligand interactions, but are hampered by lack of knowledge of the endogenous ligand. The KAI2 system appears to be present throughout angiosperms, implying a fundamentally important function in plant biology.

What are Karrikins?

Karrikins are a small family of closely related compounds that are present in burnt or charred plant material, including smoke. They were discovered following the realization that charred plant material and smoke contain chemicals that stimulate the germination of seeds of the fire-following plant species that are characteristic of regions with Mediterranean-type climates, including Chile, South Africa, southern Australia and California (Keeley and Pizzorno, 1986; de Lange and Boucher, 1990; Baldwin et al., 1994; van Staden et al., 2004; Nelson et al., 2012). Purification of the parent compound and confirmation of its structure by chemical synthesis enabled studies of the origin and mode of action of karrikins (Flematti et al., 2004; Nelson et al., 2012). They may be generated by pyrolysis of carbohydrates, such as cellulose and simple sugars (Flematti et al., 2011), but have no other known natural origin. To date, the only known biological activities of karrikins are associated with seed germination and seedling growth. Structurally, karrikins consist of a five-membered butenolide ring fused to a six-membered pyran ring. Six compounds, differing with respect to methyl substitutions, have been identified in plant-derived smoke, and are annotated as KAR1 to KAR6 (Flematti et al., 2009) (Figure 1).

Figure 1.

Structures of various bioactive butenolides. KAR1 to KAR6, six described members of the karrikin family; strigol, a naturally occuring strigolactone with characteristic ABC–D ring structure; GR24, an artificial strigolactone analogue with the same ABC–D structure; carlactone, an endogenous strigolactone precursor; CN-debranone, an artificial strigolactone mimic that lacks the ABC and enol ether moieties. Of the four possible stereoisomers of GR24, only that which most closely resembles the naturally occurring strigol is shown. Debranones are synthesized as a racemic mixture of two stereoisomers.

An important development was the discovery that Arabidopsis thaliana responds effectively to karrikins (Nelson et al., 2009). This made it possible to perform studies on the mode of action of karrikins using the functional genomics resources and vast amount of fundamental knowledge available for this plant. This review focuses specifically on the karrikin response system of Arabidopsis thaliana, as nothing is yet known about the karrikin response system at the molecular level in any other taxa.

First, we summarize the known effects of exogenous karrikins on Arabidopsis, the discovery of genes required for karrikin responses, and the phenotypes of mutants. We then discuss the proposed mechanism by which karrikins are perceived by Arabidopsis, especially in the context of the putative karrikin receptor protein, KAI2. Finally, we discuss the characteristics of endogenous plant compounds that karrikins may mimic.

Response to Karrikins: Stimulation of Seed Germination

Karrikins alone do not appear to break physiological seed dormancy. Rather, it appears that they promote germination of seeds that have become conditioned or receptive to karrikin stimulation. In Arabidopsis, freshly harvested seeds are poorly responsive to karrikins, but seeds become karrikin-responsive during after-ripening. However, seeds then reach a point when they no longer show dormancy, and so karrikins have no stimulatory effect on germination rate or percentage germination. For experimental studies of germination, the Ler ecotype has been used successfully because, relative to other ecotypes of Arabidopsis, it exhibits a moderate degree of primary dormancy in freshly harvested seeds, which retain such dormancy when stored at −20 or −80°C. However, the stimulatory effect of karrikin is not observed on medium containing nitrate, because nitrate also overcomes such primary dormancy in Arabidopsis. For this reason, germination assays with karrikins are performed using water. Karrikins are more effective at stimulating seed germination compared with the gibberellic acid GA4, epi-brassinolide or 1–aminocyclopropane-1–carboxylic acid, an ethylene precursor (Nelson et al., 2009). The strong dormancy in ecotypes such as Cvi is not overcome by karrikins, and karrikins do not overcome the light requirement of Arabidopsis seeds (Nelson et al., 2009). Furthermore, karrikins do not overcome the requirement for synthesis or perception of GA, nor do the amounts of GA and abscisic acid in seeds change in response to karrikins during pre-germination (Nelson et al., 2009). Although KAR1 and KAR3 are the most active karrikins for stimulating seed germination of fire-following taxa (Flematti et al., 2007), Arabidopsis is unusual in responding more sensitively to KAR2 (Nelson et al., 2009). KAR1 demonstrates activity below 10−9 m in most smoke-responsive species, and has an EC50 (effective concentration to elicit 50% maximal germination) of 10−11 m for Lactuca sativa cv. Grand Rapids lettuce seed (Flematti et al., 2004). On the other hand, KAR2 shows an EC50 of 10−9 m for Grand Rapids lettuce seed, and at least 100-fold less activity for smoke-responsive taxa such as Solanum orbiculatum and Emmenathe pendulifora compared with KAR1 (Flematti et al., 2007); however, detailed comparisons have not been performed with other taxa. Arabidopsis shows an EC50 of approximately 10−8 m with KAR2 and 10−7 m with KAR1 (Nelson et al., 2009), and therefore the 3–methyl group appears to reduce activity in Arabidopsis. As KAR4 has almost no activity (Nelson et al., 2009), the 7–methyl group apparently prevents karrikin action in Arabidopsis.

When karrikins were discovered, their similarity to strigolactones was immediately noted (Flematti et al., 2004), as both contain a 3–methyl-butenolide ring (Figure 1). Germination of Arabidopsis seed is initiated by the synthetic strigolactone analogue GR24, although they respond more rapidly or are more sensitive to equivalent concentrations of karrikins (Nelson et al., 2009). This raised the possibility that exogenous karrikin may act in a similar manner to endogenous strigolactones to stimulate germination. However, seeds of known strigolactone biosynthesis mutants do not exhibit increased dormancy compared to wild-type seeds, suggesting that strigolactones do not play a prominent role in promoting germination of Arabidopsis seed (Nelson et al., 2011; Shen et al., 2012; Scaffidi et al., 2013). In addition to alleviating dormancy, karrikins also enhance the light-dependent germination of Arabidopsis seed, effectively increasing seed sensitivity to low fluence rates of light (Nelson et al., 2010). The conclusion from these physiological studies is that karrikins stimulate germination of seeds that are competent and primed for germination by an unknown mechanism.

Response to Karrikins: Seedling Development

Seedling photomorphogenesis is a precisely regulated process that involves a number of environmental signals. Most prominent among these cues is light, which initiates the transition to photoautotrophy through rapid inhibition of hypocotyl elongation, and the promotion of cotyledon expansion and greening. Similarly to the positive effect of karrikins on the sensitivity of seed germination to light, exogenously supplied karrikins also have a positive effect on these aspects of seedling photomorphogenesis. The effect is most pronounced under continuous red light of up to 20 μmol m−2 sec−1, because even weak blue light is a potent inhibitor of hypocotyl elongation (Gaba and Black, 1979; Young et al., 1992). Under red light, KAR1 and KAR2 inhibit hypocotyl length in a dose-dependent manner over a 100-fold concentration range; the hypocotyl length of Arabidopsis seedlings treated with 1 μm KAR2 is typically half that of untreated seedlings (Nelson et al., 2010; Waters and Smith, 2013). This response was also observed in seedlings of both lettuce (Lactuca sativa) and wild turnip (Brassica tournefourtii) (Nelson et al., 2010). Karrikins also positively influence the expansion of cotyledons and the accumulation of chlorophyll (Nelson et al., 2010). Surprisingly, while both karrikins and GR24 inhibit hypocotyl elongation, only karrikins promote cotyledon expansion, indicating that Arabidopsis is able to distinguish between karrikins and strigolactones (Nelson et al., 2010; Waters et al., 2012a). Overall, karrikins enhance the response of seedlings to a given intensity of light, although the effect diminishes with increasing light levels. We have suggested that exposure to karrikins may improve the establishment of seedlings after a fire event by enhancing germination and promoting seedling vigour (Nelson et al., 2010).

Karrikins do not affect the morphology of Arabidopsis seedlings in darkness, and, as expected, both phytochrome A and phytochrome B are necessary for karrikin responses under red light (Nelson et al., 2010), suggesting that karrikins cannot overcome the requirement for light in effecting photomorphogenesis. Downstream of the photoreceptors, relatively little is known regarding interactions between the light and karrikin signalling pathways. The transcription factor HY5, a central regulator of both blue- and red-light responses, is required for full morphological responses to karrikins, but not for many transcriptional effects (Nelson et al., 2010; Waters and Smith, 2013). In addition, HY5 appears to influence hypocotyl length in a genetically separable manner from known karrikin response pathway genes (Tsuchiya et al., 2010; Waters and Smith, 2013). However, the promoters of a large number of karrikin-responsive genes contain a G–box-binding element and are thus putative HY5 targets; these targets include KAI2/HTL, which encodes the probable karrikin receptor (see below). As such, while HY5 is not absolutely required for seedling responses to karrikins, it is possible that HY5- and karrikin-dependent signalling converge and act co-operatively during seedling development.

Genetic Dissection of the Karrikin Response

The robust response of Arabidopsis to karrikins facilitated the possibility of performing genetic analysis to discover the molecular mechanisms at work. Similarities in response and molecular structure suggested that karrikins and strigolactones may share common signalling elements. A forward genetic screen for karrikin-insensitive (kai) mutants led to identification of MAX2 as essential for the karrikin response (Nelson et al., 2011). MAX2 encodes an F–box protein that forms a complex with other components of E3 ubiquitin protein ligase complexes, including ASK1 and Cullin (Stirnberg et al., 2007). MAX2 – and its homologues in rice (Oryza sativa) and pea (Pisum sativum) – was previously identified as essential for the strigolactone response (Stirnberg et al., 2002; Gomez-Roldan et al., 2008; Umehara et al., 2008), but its role in mediating seed germination had not been previously reported (Nelson et al., 2011). This similarity between strigolactone and karrikin signalling triggered a search for karrikin-specific components on the assumption that they may be related to strigolactone signalling components. The rice D14 gene is also essential for the strigolactone response, which is typically revealed as a repression of secondary shoot outgrowth, and there are three genes with obvious similarity to D14 in the Arabidopsis genome. A combination of systematic reverse genetics and serendipitous analysis of another karrikin-insensitive mutant identified a role for one of these genes, KAI2, in mediating karrikin responses (Waters et al., 2012a). Another was identified as AtD14, the Arabidopsis orthologue of the rice D14, because it is required for suppression of secondary shoot growth (Waters et al., 2012a). KAI2 was previously described as HYPOSENSITIVE TO LIGHT (HTL) on the basis of its requirement for normal seedling photomorphogenesis, but its molecular function was not understood (Sun and Ni, 2011). D14 and KAI2 are classified as α/β–fold hydrolases, which typically are enzymes. Indeed, DAD2 (the D14 orthologue in Petunia hybrida) and rice D14 have been shown to hydrolyse the strigolactone analogue GR24 (Hamiaux et al., 2012; Jiang et al., 2013; Nakamura et al., 2013; Zhou et al., 2013). However, numerous lines of evidence indicate that KAI2 and D14 are also receptor proteins for karrikins and strigolactones, respectively.

In Arabidopsis, there is striking phenotypic overlap between kai2, Atd14 and max2 mutants. There is little in common between kai2 and Atd14 phenotypes, with the exception that both mutants exhibit abnormal leaf morphology (Waters et al., 2012a). Other than the abnormal leaf morphology, kai2 and Atd14 mutants display complementary subsets of the numerous max2 phenotypes: whereas max2 is insensitive to both strigolactones and karrikins, kai2 is only insensitive to karrikins, and Atd14 is only insensitive to strigolactones. While max2 and Atd14 have impaired control of shoot branching, similar to strigolactone biosynthesis mutants, this aspect of growth is normal in kai2. Notably, AtD14 is responsible for strict strigolactone responses, because Atd14 resembles strigolactone-deficient mutants except that it is strigolactone-insensitive. In contrast, max2 mutants exhibit additional phenotypes (increased seed dormancy and impaired seedling development) that strigolactone-deficient mutants do not share, but that are common to kai2. Thus, we infer that KAI2 is responsible for mediating the strigolactone-independent functions of MAX2. Consistent with this interpretation, double Atd14 kai2 mutants phenocopy max2 at the seedling stage (Waters et al., 2012a; Scaffidi et al., 2013). It has been suggested that KAI2 or D14 may act together or separately to regulate the activity of MAX2 in a ligand-dependent manner (Beveridge and Kyozuka, 2010; Hamiaux et al., 2012; Waters et al., 2012a). From an experimental viewpoint, the ability of karrikins to influence seedling morphogenesis has permitted development of a powerful assay system to examine the relative contribution of KAI2 and AtD14 in mediating responses to various butenolide compounds. For example, kai2 mutant hypocotyls do not respond to karrikins, but do respond to GR24; as Atd14 kai2 double mutants are unresponsive to both karrikins and GR24, we infer that AtD14 must mediate some responses to GR24 in kai2 seedlings (Waters et al., 2012a). Likewise, karrikins inhibit Atd14 mutant hypocotyls but not Atd14 kai2 hypocotyls, providing further evidence that KAI2 is required for karrikin perception. In addition, there are a number of transcripts, including the auxin-inducible AUX/IAA genes, that constitute additional responses to karrikins and GR24 (Waters et al., 2012a; Scaffidi et al., 2013).

The most recent genetic discovery relating to karrikin signalling emerged from a screen for suppressors of max2. By making use of the seed dormancy and seedling development phenotypes of max2, Stanga et al. (2013) isolated a new mutant in which several max2 phenotypes relating to seed germination and seedling development were restored. This led to isolation of SMAX1, and identification of an eight-membered SMAX1-LIKE (SMXL) gene family in Arabidopsis. SMXL genes have no previously reported functions, but share sequence similarity with the heat shock protein 101 class of molecular chaperones. In many respects, both smax1 single mutants and smax1 max2 double mutants resembled wild-type seedlings that had been treated with karrikins, suggesting that the karrikin signalling pathway was constitutively activated. Such an observation is consistent with the interpretation of SMAX1 as a repressor of karrikin signalling (Stanga et al., 2013). Such repressors are predicted by a model in which MAX2 directs the ubiquitin-mediated degradation of target proteins, with MAX2 activity being modulated by KAI2 or AtD14 (Figure 2). Strikingly, smax1 alleles do not suppress the max2 shoot-branching phenotype, perhaps implying that other members of the SMAX1 family mediate the strigolactone-related functions of MAX2 (Stanga et al., 2013). Consistent with this hypothesis, the function of D53, a SMAX1 homologue in rice, has recently been reported (Jiang et al., 2013; Zhou et al., 2013). A single dominant mutant d53 allele caused a dwarf high-tillering phenotype similar to those of other rice d mutants. As predicted by the model, D53 interacts with D3 (the rice orthologue of MAX2) and D14 in a GR24-dependent manner. Significantly, GR24 induces ubiquitination and degradation of D53 protein, but only in the presence of functional D3 and D14. The mutant d53 protein is resistant to degradation, blocking downstream strigolactone responses. It has thus been concluded that D53 acts as a key component of strigolactone signalling, specifically in processes relating to shoot architecture (Jiang et al., 2013; Zhou et al., 2013). Although D53, and by extension SMAX1, are implicated in the control of gene expression, the possibility that they mediate post-transcriptional responses should also be considered, as has been suggested for strigolactones (Shinohara et al., 2013). Thus, a picture is emerging in which various SMAX1/D53 family members may mediate different responses to karrikins and strigolactones at different stages of development.

Figure 2.

Overview of proposed karrikin signalling via KAI2 and MAX2. Association of KAI2 and MAX2 is assumed by analogy with DAD2 and PhMAX2a in petunia (Hamiaux et al., 2012) and D14 and D3 in rice (Jiang et al., 2013; Zhou et al., 2013). The ubiquitination of a repressor is hypothetical, as proposed by Waters et al. (2013), but also demonstrated for rice D53 (Jiang et al., 2013; Zhou et al., 2013). Genetic evidence suggests that SMAX1 acts downstream of MAX2, and so is a candidate for the hypothetical repressor, as are other members of the gene family SMAX1-LIKE (SMXL). Evidence for auxin signalling is indirect, based on expression of IAA genes and phenotypes of seedlings, and analogy with strigolactone signalling.

Structural and Molecular Aspects of Karrikin Perception

The structural similarity between karrikins and strigolactones, together with the fact that they require very closely related hydrolase-type proteins for activity, has led to the hypothesis that these two classes of plant growth regulators share common perception mechanisms. Indeed, the KAI2 protein required for the response to karrikins also mediates responses to synthetic strigolactone analogues (Waters et al., 2012a; Waters and Smith, 2013). The first molecular mechanism proposed for the chemical reactivity of strigolactones involved a Michael-type addition to the enol ether that bridges the ABC and D rings (Mangnus et al., 1992; Wigchert and Zwanenburg, 1999). This mechanism came into question as a result of observation of bioactivity of a number of strigolactone mimics lacking the enol ether functionality. An alternative mechanism involving a Michael addition to the C4 position of the butenolide ring was subsequently proposed (Zwanenburg and Mwakaboko, 2011). However, the report of active C4-substituted analogues cast doubt on this mechanism (Boyer et al., 2012). A similar mechanism for karrikins involving a Michael addition to the enol ether pyran carbons was also considered (Chiwocha et al., 2009; Zwanenburg et al., 2009). To evaluate this mechanism, various saturated derivatives lacking the Michael acceptor characteristic were prepared (Scaffidi et al., 2012). All analogues retained some level of germination activity for karrikin-responsive Solanum orbiculatum, and therefore doubt was also cast on the existence of a Michael addition mechanism for karrikins.

The identification of KAI2 and AtD14 led to the proposal of a new mechanism for karrikin and strigolactone activity, in which the butenolide ring is subject to nucleophilic attack (Figure 3) (Scaffidi et al., 2012; Zhao et al., 2013), consistent with other known α/β-fold hydrolases (Bains et al., 2011). KAI2 and D14 have a conserved catalytic triad (Ser–His–Asp) within a hydrophobic pocket, and D14 has been shown to bind (Kagiyama et al., 2013) and hydrolyse (Hamiaux et al., 2012) the synthetic strigolactone GR24. The amino acid residues of the catalytic triad of DAD2 have been shown to be necessary for strigolactone activity as well as for GR24-induced conformational change (Hamiaux et al., 2012). This mechanism is supported by crystallography, which revealed that the hydrolysed D ring of GR24 was covalently attached to the active-site Ser in the catalytic triad (Figure 3) (Zhao et al., 2013). In addition, this mechanism explains the activity of all compounds that display strigolactone-like activity. Whereas the previous model for strigolactone activity required a conjugated enol ether attached to a D ring, the new mechanism explains how the presence of a suitable leaving group attached through an ether bridge to a butenolide ring, such as CN-debranone (Figure 1), is sufficient to render a molecule active (Fukui et al., 2011; Boyer et al., 2012; Waters et al., 2012b).

Figure 3.

Proposed mechanism for interaction of butenolide ligands with receptor proteins. (a) The hydrolysis mechanism for D14 acting on GR24. Activation of Ser97 in the conserved catalytic triad and nucleophilic addition to the carbonyl group of the butenolide ring results in cleavage of the formyl ABC ring. The subsequent addition of two water molecules results in an enzyme intermediate that has been observed by X–ray crystallography (Zhao et al., 2013). Finally, cleavage of the covalent enzyme intermediate, followed by loss of water, returns the hydrolysed butenolide ring. (b) Proposed nucleophilic addition of the activated Ser95 of KAI2 on the butenolide ring of KAR1. Hydrolysis of the intermediate followed by ring closure re-generates the parent molecule.

An analogous mechanism is proposed to initiate hydrolysis of the karrikin butenolide ring in the active site of KAI2 (Figure 3b). There is no obvious leaving group for karrikins, unlike the ABC moiety of strigolactones that fulfils this role. Hydrolysis of the covalent intermediate associated with nucleophilic addition to the carbonyl group of karrikins is expected to return the parent molecule, rather than any new products (Figure 3b). Consistent with this mechanism, the catalytic Ser is essential for KAI2 function in Arabidopsis, in terms of complementing both the seedling morphogenesis phenotype of kai2–2 and seedling karrikin responses (Figure 4 and Methods S1). A mechanism that involves nucleophilic attack and hydrolysis of karrikins by KAI2 is consistent with that of strigolactone perception, in which it is the catalytic activity of D14 rather than the products of catalysis that are required for the response (Boyer et al., 2012; Hamiaux et al., 2012).

Figure 4.

The catalytic serine is essential for KAI2 function. The kai2–2 mutant was transformed with a chimeric construct consisting of the native KAI2 promoter and 5′ UTR fused to either the wild-type KAI2 coding sequence (KAI2:KAI2) or an otherwise identical coding sequence in which Ser95 was mutated to Ala (KAI2:KAI2 S95A). Seed homozygous for each transgene were sown on 0.5 x MS medium supplemented with karrikins as shown, and grown for 4 days under continuous red light. Further methodological details are provided in Methods S1. Scale bar = 4 mm.

Recently, X–ray crystallography studies by Guo et al. (2013) have provided some insights into the interaction between karrikins and KAI2. This study demonstrated that, through aromatic–aromatic interactions, KAR1 is positioned at the opening of the hydrophobic cavity containing the catalytic triad, but is located at an unfavourable distance to undergo nucleophilic attack by the activated Ser. In fact, no role was suggested for the conserved catalytic triad, with the data implying that the aromatic–aromatic interactions were sufficient to induce a conformational change in protein structure. Presumably, this change in structure provides the signalling activity, perhaps through protein interaction partners. The catalytic triad is conserved throughout all D14 and KAI2 proteins, consistent with an essential catalytic function. We speculate that the KAR1 molecule observed in KAI2 crystals by Guo et al. (2013) either requires repositioning for activity, or has already become repositioned away from the active-site Ser following its hydrolysis, which is the process that triggers the conformational change in the protein. Structural studies with mutated proteins may help to resolve questions regarding the interaction between the karrikin ligand and the KAI2 active site.

Genetic studies in Arabidopsis have demonstrated that karrikins only signal via KAI2, but GR24 signals via both AtD14 and KAI2 (Waters et al., 2012a; Waters and Smith, 2013). What is the basis for this substrate specificity? Comparison of the respective binding pockets shows surprisingly few differences between KAI2 and AtD14. Both proteins contain a hydrophobic pocket lined with highly conserved residues (Figure 5). Intriguingly, both KAI2 and AtD14 contain a conserved Phe residue that separates two distinct cavities. If the phenyl side chain is mobile, the second pocket may influence substrate specificity. Most of the few differences in the active-site pocket are apparently subtle but nevertheless contribute to noticeably different cavity geometries (Figure 5). Some of these changes are evolutionarily well conserved: for example, KAI2 sequences from several species contain Ile at position 193, whereas the corresponding residue in D14 sequences is typically Val. It is difficult to predict whether a given difference is significant with respect to ligand binding specificity, but ‘small’ changes should not be dismissed out of hand: apparently conservative mutations from Ile to Leu or Val in the pocket of the GA receptor GID1 (itself an α/β-hydrolase family member) change its binding affinity for different GAs (Shimada et al., 2008). Kagiyama et al. (2013) note that more bulky non-polar residues in the pocket of OsD14L (the rice orthologue of Arabidopsis KAI2) compared with OsD14 constrict the size of the pocket, and thus may potentially contribute to substrate selectivity. However, it is important to compare proteins for which the substrate specificities may be determined experimentally, and the Arabidopsis system is an invaluable tool in this respect. Searching for evidence of selective pressure at particular residues may also be a fruitful means to identify residues for site-directed mutagenesis and to exploit Arabidopsis transgenesis. Finally, it is also possible that the affinity of KAI2 and D14 for different substrates is modulated by their interacting protein partners in the SMXL/D53 family, rather like the TIR1–Aux/IAA co-receptor system for auxin (Calderon-Villalobos et al., 2012).

Figure 5.

Structural differences between the ligand binding pockets of KAI2 and D14 from Arabidopsis. The internal surface of the cavities is shown in grey. (a) KAI2 and its two pockets separated by residue Phe26. (b) D14 and its two pockets separated by residue Phe28. Among other common residues lining the pockets, KAI2 and D14 share the catalytic triad Ser95/97–His246/247–Asp217/218 at the bottom of pocket 1, and two Phe residues (134/136 and 194/195) near the entrance of pocket 1. Only a few residues lining the pockets differ between KAI2 and D14; these are highlighted in pink and blue, respectively. Conserved residues are shown in white. Images are derived from crystallography data published by Bythell-Douglas et al. (2013) and Zhao et al. (2013).

Clues to the Endogenous Substrate of KAI2

KAI2 is clearly important in the control of seed germination and seedling development, even in the absence of karrikins. If D14 is the strigolactone receptor, what does KAI2 perceive? A simplistic interpretation of the existence of two highly similar apparent receptor proteins is that each has affinity for different substrates, possibly different subsets of the same compound class, i.e. strigolactones. All known strigolactones are thought to be derived from the precursor carlactone (Figure 1), which is produced by the sequential action of the beta–carotene isomerase D27, and the carotenoid cleavage dioxygenases CCD7 and CCD8, on β–carotene (Alder et al., 2012; Seto et al., 2014), for which there is strong genetic evidence from multiple species (Brewer et al., 2013). Accordingly, if the KAI2 substrate is also derived from carlactone, then carlactone should have potent karrikin-like activity in seeds and seedlings. Recently, we demonstrated that carlactone is able to repress shoot branching and promote normal leaf morphogenesis, suggesting that it is converted into active strigolactones in planta (Scaffidi et al., 2013). Crucially, these effects in adult plants require the activity of both cytochrome P450 MAX1 and the putative strigolactone receptor D14. However, when supplied to primary dormant Arabidopsis seed, carlactone did not stimulate germination, but GR24 did. In addition, carlactone had very limited activity in seedlings, both in terms of inhibiting hypocotyl elongation and inducing changes in transcript abundance (Scaffidi et al., 2013). This weak activity did not depend on MAX1, suggesting that the activity did not result from conversion of carlactone to strigolactones. In addition, d14 mutant seedlings were essentially insensitive to carlactone, suggesting that KAI2 does not perceive carlactone or its derivatives. These findings imply that the endogenous substrate for KAI2 is derived from a carlactone-independent pathway. This conclusion is supported by the observation that seedlings of Arabidopsis strigolactone biosynthesis mutants (d27, max3, max4 and max1) are all morphologically indistinguishable from wild-type seedlings (Nelson et al., 2011; Shen et al., 2012; Flematti et al., 2013). Therefore, either another pathway exists in Arabidopsis for producing canonical strigolactones independently of carlactone, or there is another class of butenolide compounds of unknown biosynthetic origin. Testing whether KAI2-dependent processes are triggered by endogenous strigolactones would help to distinguish between these possibilities. The endogenous KAI2 substrate is expected to share structural features with stigolactones, debranones and karrikins. The butenolide is presumably essential, as is a hydrophobic substituent to anchor the molecule in the active site. This substituent must have electron-accepting properties, because a saturated pyran ring is inactive (Scaffidi et al., 2012). As is the case with strigolactones, we expect the compound to have extremely low abundance in plant tissues. However, the search for novel compounds with such characteristics will become increasingly feasible with ever-improving separation chemistry and genetic resources.

Concluding Remarks

Seed germination, seedling establishment and leaf morphogenesis represent critical processes during the life of a plant, and the karrikin response system of Arabidopsis is thus of importance in understanding these aspects of plant development more fully. Homologues of KAI2 and AtD14 are present throughout seed plants, and KAI2-type genes may be traced as far back as algae (Delaux et al., 2012; Waters et al., 2013). These observations suggest that KAI2 may have a key role in all plants, and may have been a target for selection in both plant breeding programmes and during adaptive evolution. Fire-following species may represent just one specialized example, but to date the KAI2 system has not yet been studied in taxa other than Arabidopsis. The comparative approach to study of karrikin and strigolactone signalling has provided crucial information for each area of investigation, with progress in one helping to inform the other. A comparative approach to studying KAI2 function in different plant species may help to achieve similar advances in our understanding of the karrikin response system.


We thank our colleagues Kingsley Dixon, Emilio Ghisalberti, Charlie Bond and Rohan Bythell-Douglas from the University of Western Australia for insightful discussions. We also thank the Australian Research Council for financial support grants LP0882775, DP1096717 and FT110100304.