Specifying self-recognition: peptides lead the way


The extraordinary evolutionary success of flowering plant species is attributed, at least in part, to their ability to maintain genetic fitness by the promotion of out-crossing. In order to accomplish this, many flowering plants have evolved self-incompatibility (SI) systems which utilize receptor–ligand-type interactions to perceive, recognize and reject self-pollen (incompatible pollen), whilst compatible pollen from genetically unrelated individuals of the same species is discriminated from self-pollen and is allowed to grow and fertilize the plant. Thus, the outcome of this interaction is absolutely crucial for determining mate choice. At the most fundamental level, all SI systems depend on the cognate interaction of proteins expressed in the pistil with proteins expressed in the pollen to achieve this highly specific cell–cell recognition. In this issue of New Phytologist, Kemp and Doughty (pp. 619–629) have investigated the nature of this interaction in Brassica, using a novel peptide array to establish specific regions involved in receptor–ligand interaction.

‘Evidence suggests that several self-incompatibility systems have apparently evolved independently.’

In order to facilitate self-recognition, the genes encoding both pollen and pistil components must exhibit tight physical linkage at the multiallelic S locus. The polymorphic nature of the S locus and the quest to establish the nature of this tight specificity have made SI a subject of considerable interest at a theoretical level. The potential for application of such knowledge to plant breeding has given this added edge. Research over the last ∼20 yr has focused on elucidating the components that comprise the S locus and the mechanisms that control pollen rejection at a molecular level. Evidence suggests that several SI systems have apparently evolved independently. Three of the SI systems have been well characterized at a molecular and cellular level. These comprise: (1) the Brassicaceae, which have the S receptor located in the stigmatic papillae to interact with a pollen S ligand; (2) a large group that utilize an S-RNase SI system (in the Solanaceae, Rosaceae and Scrophulariaceae); and (3) the Papaveraceae, where plants secrete a ligand from the stigma which interacts with the pollen receptor (see Takayama & Isogai, 2005; McClure & Franklin-Tong, 2006 for recent reviews).

It was work on Brassica spp. that led to the first identification of both female and male determinants of SI, comprising a receptor and its ligand. The pistil component, S-receptor kinase (SRK), is a receptor that transduces signals perceived at the surface of stigmatic papillar cells (Stein et al., 1991). The pollen component, S-locus cysteine-rich (SCR; also known as SP11), is a small cysteine-rich protein located within the pollen coat (Schopfer et al., 1999; Shiba, 2001; Takayama et al., 2001). It has been demonstrated that SRK and SCR/SP11 interact in an S-haplotype-specific manner and that this stimulates SRK autophosphorylation (Cabrillac et al., 2001; Kachroo et al., 2001; Takayama et al., 2001; Shimosato et al., 2007). However, progress in identifying regions involved in the haplotype-specific interaction has been difficult and until now there has been no information on the regions of SRK that are important for receptor–ligand binding and activation.

In this issue of New Phytologist, Kemp and Doughty present important data that move the story significantly forward. In their paper they shed light on the nature of the regions of SRK that are involved in haplotype-specific recognition. First, they identified three subdomains (PAN, lectin and hypervariable) of the extracellular component of SRK (eSRK). Using a binding assay, the authors aimed to identify which of these domains play a key role in eSRK–SCR interaction. Although these domains had previously been identified, no one had previously tested their ability to bind SCR/SP11. Kemp and Doughty convincingly demonstrate that it is the hypervariable subdomain of eSRK that strongly binds the pollen component, SCR, and not the PAN or lectin subdomain.

Probably the most novel aspect of this paper is the authors’ strategy of using a peptide array (Fig. 1) to carry out a fine mapping analysis of the SRK hypervariable subdomain. Although this is commonly used to study receptor–ligand interactions in animal-based systems, this has not been used in plant systems. Using the hypervariable subdomain from three SRK haplotypes, they designed an overlapping series of 20-mer peptides, which were arrayed and assayed for binding to SCR from the same three haplotypes. This approach had the advantage that one can begin to detect S specificity in interactions with SCR, whereas use of an intact, larger domain was unable to identify such specificity. Furthermore, this in vitro approach provides a relatively quick and informative way to generate information about the receptor–ligand interactions involved in Brassica SI.

Figure 1.

Outline of the peptide array experiment. Peptides were designed to each of three S-receptor kinase (SRK) hypervariable subdomains (SRK5, SRK29 and SRK63) plus the hypervariable domain of SLG29 (SLG, S-locus glycoprotein). In each case, the peptides were designed with a 10 amino acid overlap between adjacent peptides. The peptides were synthesized and spotted onto Immobilon P membrane. The peptide arrays were then incubated with radio-iodinated S-locus cysteine-rich (SCR) (SCR5, SCR29 and SCR63). The diagram shows peptide spots with signal above background (dark grey) indicating positive interaction between these peptides and SCR.

The authors found that, with each SRK, at least one peptide bound to its cognate SCR and in some cases haplotype-specific binding was observed. Crucially, this approach allowed the identification of specific sequences of amino acids that are essential in the interaction of SRK with SCR. They define a small region (in peptide 6 for the S5 haplotype) that is probably crucial for defining haplotype S specificity. Thus, this represents a key step forward in establishing specific defined regions that are involved in binding. This is an important finding, as it begins to uncover how SI interactions in Brassica are initiated. Furthermore, it provides a valuable approach to analysing the biology of this system. The ability to locate small peptides that are important for SCR binding within the hypervariable domain of SRK leads to the possibility of conducting domain-swapping experiments. Indeed, using peptide arrays it should be possible to analyse the effect of changing single amino acids within SRK, given these data. By pinpointing which amino acid residues confer S-haplotype specificity, use of this type of analysis should also provide important data on how SRK alleles evolved in parallel to their cognate SCR alleles. The authors hope to confirm these data by using domain swaps in transgenic plants in future experiments.

We are now beginning to build a picture of the nature of Brassica SI interactions, which to date have been only patchily understood. The important data in this paper fit well with those of another key study (Shimosato et al., 2007), which shows that high-affinity SCR–SRK interactions only occur when SRK is membrane-bound and dimerized. The authors propose a new model for SRK–SCR/SP11 binding, based on these data. Kemp and Doughty propose a ‘two-stage’ recognition process between SRK and SCR. They suggest that an initial low-affinity generic binding of all SCRs to eSRK, which occurs regardless of haplotype (perhaps to recruit the SCR proteins), is followed by a high-affinity, S-haplotype-specific interaction that is probably responsible for SRK activation. This high-affinity binding probably involves dimerization; this might act to reveal a high-affinity pocket for specific interactions to proceed, although this brings us into the realms of speculation as this remains to be investigated. We attempt to conceptualize the basis of this model in Fig. 2.

Figure 2.

A proposed model for S-locus cysteine-rich (SCR) binding to S-receptor kinase (SRK). Kemp and Doughty suggest a model for Brassica self-incompatibility (SI), involving both low- and high-affinity binding between SCR and SRK. In the left panel, SCR3 (green circles) is able to interact with low affinity with all SRK and extracellular SRK (eSRK) molecules in a non-haplotype-specific manner. However, it is unable to bind dimerized SRK1 (red) or SRK2 (blue) with high affinity. No autophosphorylation of SRK follows and therefore the self-incompatibility signal cascade is not initiated. In the right panel, generic non-haplotype-specific binding of SCR2 (blue circles) to eSRK1, eSRK2, SRK1 and monomeric SRK2 occurs. However, on interaction with transmembrane-bound SRK2 dimers, high-affinity binding occurs (red stars). This results in autophosphorylation of SRK2 leading to signal transduction and eventually self-incompatibility.

The significance of this work is potentially broader than providing a better understanding of SI receptor–ligand interactions, as Arabidopsis has several hundred orphan receptor proteins, together with several hundred putative secreted protein ligands. As the Brassica SRK–SCR interaction is currently the ‘gold standard’ for studies of plant receptor–ligand binding, new data about the nature of this interaction are of considerable interest to all those studying plant cell signalling.