A feature of IDs that is not obvious from the initial publications is the proportion of overlap between IDs and their closest conspecific non-NLR homolog, i.e., what proportion of the presumed ID donor protein is integrated? Also, what is the level of amino acid sequence identity between ID and closest non-NLR homolog? In brief, BlastP searches indicate matches below 70 % amino acid identity and alignments of 25–60 % to the ‘full length’ closest conspecific protein using RRS1 and RGA5 IDs in searches.
The underlying proposition of the integrated decoy hypothesis, and I am adding some of my own embellishments here, is that, through evolution, integration of a domain from an effector target or non-integrated effector decoy (sensu Kamoun and van der Hoorn [6]) has occurred in some NLRs and has been selected as an effective detector of pathogen infection. How this ‘improves’ the effectiveness of NLRs and non-integrated effector targets is open to conjecture. The rather low level of relationship in terms of relative size and amino acid sequence between IDs and related conspecific proteins has several implications. First, only the domain of the host target directly involved in effector interaction (and not function of the domain donor protein) is commonly integrated. Second, the lack of exact sequence match may mean that the original target is no longer present in the host genome or, more likely, in the absence of sequence constraint for function apart from effector binding or effector modification, the ID may have evolved to maximise interaction strength with the corresponding effector and, in addition, the ability to interact with several different effectors that share a single or even closely related family of effector targets in the host. This is certainly the case with RGA5 and Pik-1 in rice where the two diverged ID sequences interact with three amino acid sequence-unrelated effectors from a single pathogen species (see [2]). Interestingly these three effectors share very similar protein structure [7]. The WRKY domain in RRS1 also interacts with two sequence-unrelated effectors from two bacterial species, although the structures of both proteins have not been determined yet [3]. Significantly, the directly interacting IDs in Pik-1 and Avr-Pik are represented by alleles in host and pathogen undergoing an ‘arms race’ [2].
One further interesting point among NLR-IDs with known resistance function is provided by the rice Pi5-2 protein, aka Pii, which carries a C-terminal ID related to “AvrRpt2 cleavage site” related domain (see [2]), also found in several other monocot and dicot NLR-IDs [5]. The corresponding effector is Avr-Pii [2], whose host target, necessary for Pi5/Pii-AvrPii recognition, is a non-ID Exo70 protein [8]. Interestingly, Exo70 occurs as an ID in several grass species but not rice [2, 4].