Engineering receptors

Four papers here modify a receptor and get a working result. In every case the modification is small, guided by a structure, and aimed at an interface rather than a domain.

Moving a recognition surface between receptors

The most complete case transfers effector recognition from one cereal receptor to another. The barley receptor MLA3 binds the blast effector Pwl2 through the C terminus of its LRR, using a surface that mimics the effector's real host target. That surface was transplanted into SR50, a rye receptor 77% identical to MLA3 whose own specificity for AvrSr50 maps elsewhere on the LRR, to position 870 (gomezdelacruz-2026-mla3-mimicry).

Two chimeras were made: one carrying the whole C terminus, one carrying only the binding interface. Both gained Pwl2 recognition, and the narrower one retained AvrSr50 recognition. Transgenic barley expressing it resisted a stem rust isolate carrying AvrSr50, comparable to wild-type SR50, and blast carrying Pwl2 (gomezdelacruz-2026-mla3-mimicry).

One LRR domain, two independent binding interfaces, two pathogens. What makes it work is that the two specificities sit in different places on the same solenoid (gomezdelacruz-2026-mla3-mimicry).

Reprogramming which helper a sensor can use

The second case changes network wiring rather than recognition. In lettuce, a sensor that natively signals only through LsNRC0 gained the ability to activate Ast-LsNRC1 after either of two single substitutions in its NB-domain interface, with the double mutant giving a stronger response than either alone. LsNRC0 signalling was retained, so this is an expansion of compatibility rather than a switch (toghani-2026-sensor-helper-interface).

The residues were chosen from the predicted sensor-helper interface, and both are predicted to contact the LRR domain of both helpers. The boundary they cross is the one that phylogeny and selection analysis independently place between lettuce sensor groups (pai-2026-lettuce-nrc).

What makes these tractable

Both cases depend on the same two conditions. The interface has to be known at residue resolution, and the receptor has to tolerate substitution there without becoming autoactive (pai-2026-lettuce-nrc).

The second condition is not free. Interface mutants elsewhere in this corpus frequently become trigger-happy, firing with a sensor alone (seager-2026-avrcap1b-nrc3), and a receptor variant locked in its assembly-competent state was counterselected in transgenic potato, giving only quantitative resistance (gao-2026-vnt1-splicing). Gain of function and loss of restraint are neighbours.

Where the target list comes from

Bioengineering needs candidates, and two papers here are effectively surveys of where to look. The lettuce network is small enough to characterise completely, which is why it became the substrate (pai-2026-lettuce-nrc). And legume NLR expression data, covering 113 phylogenetic subclasses across 28 genomes, is offered explicitly as a resource for precision breeding, on the argument that which tissue a resistance gene is expressed in matters as much as which pathogen it detects (marques-2026-legume-tissue).

Neither engineered receptor here has been tested for durability, or for cost to the plant. Both were shown to work in a season.