Resistosome architecture and oligomeric state

Nine of the ten papers here bear on what an activated NLR builds. Between them they take the count of known plant resistosome stoichiometries from one, for the CC-type receptors, to four, and add two states that are not endpoints at all.

The counts

An autoactive wheat CCG10-NLR resolves as an octamer with C8 symmetry, roughly 200 A across and 135 A high, at 3.95 A. Two further states appear in the same dataset: a dual octamer fused at the funnel ends, and one joined back-to-back through the LRR domains, which had not been seen in any plant resistosome (guo-2026-ccg10-resistosome). Arabidopsis RPS2, from the same clade but the other side of the monocot-dicot split, also forms octamers, so this is a clade property rather than a wheat one.

The NRC7 clade goes further. Purified autoactive NRC7 from potato gives ring particles about 250 A in diameter, against the roughly 150 A a hexamer would give, and counting peripheral densities in the 2D class averages returns eleven protomers. Orthologs from tomato and N. benthamiana, on separate branches of the clade, do the same (toghani-2026-sni-undecamer).

What that count rests on is worth stating: negative-stain 2D class averages support a protomer count and an overall diameter. They do not support a structure. There is no high-resolution reconstruction of the undecamer (toghani-2026-sni-undecamer).

The states in between

Assembly is not a switch. An effector that binds activated tomato NRC3 stalls it at three protomers, and the stalled species superimposes on one half of the complete hexamer with an RMSD of 0.793 A. Truncating the effector's N-terminal domains lets a fourth protomer in. Resistosomes therefore build by progressive addition of activated protomers, and it is possible to catch them doing it (seager-2026-avrcap1b-nrc3).

At the other end, one receptor never assembles at all unless its mRNA is spliced first. The intron-retained isoform of a potato late blight receptor barely oligomerises on blue native gels; the spliced form does (gao-2026-vnt1-splicing).

What determines the number

Two proposals appear here, neither validated beyond its own case. One measures the HD1-WHD interdomain angle: 80 degrees for the octamer, 88 for the NRC2 hexamer, 92 for the ZAR1 pentamer, with larger angles giving fewer protomers, and AlphaFold monomer models reproducing the experimental values (guo-2026-ccg10-resistosome). The other scores eleven parameters across predicted hexamers and flags clades that do not fit (toghani-2026-sni-undecamer).

The second is the more interesting failure. AlphaFold could not model NRC7 at any stoichiometry from 5-mer to 11-mer, including the correct one. It located the anomaly without describing it (toghani-2026-sni-undecamer).

What is missing from every structure here

The N-terminal alpha1 helix — the part that enters the membrane and forms the pore — is unresolved in every reconstruction in this corpus (guo-2026-ccg10-resistosome, seager-2026-avrcap1b-nrc3). Pore geometry is therefore computed from what remains: a minimum radius of 12.7 A at Lys140 for the octamer against 4.3 A at Gln126 for the NRC4 hexamer.

Resting states are scarcer still. Only ZAR1 and NRC2 have been resolved at rest, which is why the mechanism by which a single LRR substitution renders a receptor autoactive remains unexplained (guo-2026-ccg10-resistosome).

Four stoichiometries, two of them found in the same year, and the sampling is still confined to two plant families.