Contrasting evolutionary patterns of helper and sensor NRC NLRs in lettuce reflect functional divergence following subfunctionalization
Lettuce carries a small, tractable NRC network of two helpers and nine functional sensors, and the sensors that depend on the ancient helper NRC0 have diversified far less than those that depend on the newer one.
The NRC network is usually studied in Solanales, where it has run away with the genome: NRC-type receptors make up half the NLR repertoire. Asterales did the opposite. Across 29 Asterales genomes NRCs account for 6.6%, and lettuce has just 17 of them against 359 NLRs total, with TIR-NLRs taking the space instead. That smallness is the point. Nine lettuce sensors were shown to signal through two helpers in a partially redundant arrangement, and helper dependency falls out along phylogeny: group 1 sensors go through LsatNRC0 only, group 3 through Ast-LsatNRC1 only, group 2 through either. Selection analysis then splits the same groups. Helpers and group 2 show no positively selected sites at all, while group 3 sensors carry them on the concave face of the LRR.
Key findings (9)
The NRC network expanded far less in Asterales than in Solanales. From 21,232 non-redundant NLRs across 40 Solanales and 29 Asterales genomes, NRC-type NLRs make up 6.6% of the Asterales NLRome against 50.3% in Solanales, despite similar total NLR counts per genome (300 in Asterales, 338.6 in Solanales). Asterales expanded other classes instead, TIR-NLRs in particular.
Evidence de novo genome annotation with NLRtracker; phylogenetic extraction of the monophyletic NRC network (7,020 sequences) using RefPlantNLR references
Source Figures 1A and 1B; figures S1, S2; data S1-S3
Asterales NRC helpers split into two phylogroups: 27 sequences grouping with NRC0 and 61 forming a separate group with Solanales helpers such as NRC2 and NRC4, termed Ast-NRCs. All 504 Asterales sensors form one well-supported group beside the Solanales SD-type sensors, and none group with Rx-type sensors. The Asterales SD-type sensors lack the Solanaceous domain at their N-termini.
Evidence phylogenetic analysis of 592 Asterales NRC-H and NRC-S (88 helpers, 504 sensors)
Source Figure 1B; data S3
Lettuce has a compact network: 17 NRC sequences (2 helpers, 15 sensors) out of 359 total NLRs, roughly 5%. TIR-NLRs dominate at 206 sequences (over 57%), with CC-NLR, CCR-NLR and CCG10-NLR classes at 97, 9 and 42.
Evidence NLRome classification of the common lettuce genome
Source Figure S3
The two lettuce helpers differ in genomic context. Ast-LsatNRC1 sits in a physical cluster (under 100 kb) with group 2 and group 3 sensors on chromosome 8, within the previously reported MRC8c. LsatNRC0 on chromosome 3 is not clustered with any sensor and is the only NRC gene outside the reported major resistance clusters. Across other Asterales the pattern is species-specific: Cichorium intybus has both helper types clustered with sensors, Helianthus annuus has none.
Evidence genomic location analysis in Lactuca sativa, L. saligna, L. virosa and five further Asterales species
Source Figures 2A and 2B; figures S4-S7; data S4, S6
Nine lettuce sensors are functionally connected to the two helpers, in a pattern that follows phylogeny exactly. Three signal only through LsatNRC0 (all group 1), four only through Ast-LsatNRC1 (all group 3), and two through both (group 2). This is a partially redundant network.
Evidence co-agroinfiltration cell death assays in N. benthamiana with sensors as autoactive MHD D-to-V mutants and wild-type helpers; two replicates
Source Figures 3A and 3B; figure S8B; data S7
Both lettuce helpers carry a MADA motif; none of the tested sensors returned a hit against the MADA HMM profile. This matches the cell death data, where helpers but not sensors trigger cell death alone.
Evidence HMM profile search; cell death assays
Source Figure 3; data S8
AlphaFold 3 separates helpers from sensors structurally. Modelled as hexamers with 50 oleic acids as a plasma membrane proxy, five replicates each, both helpers gave resistosome-like oligomers (pTM and ipTM above 0.5 in every replicate, mean per-chain pTM 0.75) while no sensor from any of the three groups did (pTM and ipTM below 0.5, per-chain pTM below 0.7, high predicted aligned error). AlphaFold could not build a complete N-terminal funnel for LsatNRC0 in any replicate.
Evidence AlphaFold 3 modelling of 2 helpers and 16 sensors, 5 replicates each
Source Figure 4; figure S9; data S8
Selection pressure splits along the same lines as helper dependency. No positively selected sites were detected in the helpers or in group 2 sensors. Group 1 and group 3 sensors carry several, and in group 3 most localise to the LRR domain, concentrated on its concave face.
Evidence CodeML (PAML v4.10.7) M8 versus M8a likelihood ratio tests with Bayes Empirical Bayes site identification; sites mapped onto AlphaFold 3 models
Source Figures 5A-5C; data S10
The overall pattern is that sensors relying on the phylogenetically conserved helper NRC0 stay constrained, while sensors relying on the other helpers show higher rates of positive selection and gene duplication. The authors read this as functional divergence following subfunctionalization of the helper nodes.
Evidence synthesis of copy number, selection and functional dependency data
Source Figure 6; Discussion
NLRs
| Name | Class | Role in this paper |
|---|---|---|
| LsatNRC0 NRC0, Helixer_NC_056625.2_001598.1 | helper subject | the ancient helper; supports group 1 and group 2 sensors; unusually not clustered with any sensor in lettuce |
| Ast-LsatNRC1 Helixer_NC_056630.2_002654.1, RGC7 | helper subject | the Asterales-specific helper; supports group 2 and group 3 sensors; clustered with them on chromosome 8 |
| Lettuce NRC-S group 1 RGC26 | sensor subject | signal exclusively through LsatNRC0; high variability in CC and N-terminal LRR; carry positively selected sites |
| Lettuce NRC-S group 2 RGC27 | sensor subject | signal through both helpers; low variation and no detected positively selected sites |
| Lettuce NRC-S group 3 RGC9, Helixer_NC_056630.2_001736.1 (representative) | sensor subject | signal exclusively through Ast-LsatNRC1; most diversified group, with positively selected sites on the concave LRR face |
| Ast-LvirNRC1 Helixer_CAKMRJ010003334.1_001958.1 | helper subject | wild lettuce ortholog positioned near group 2 and 3 sensors on contig2395 |
| CiNRC0 and Ast-CiNRC1a/b Helixer_CM042016.1_001204.1, Helixer_CM042010.1_003195.1, Helixer_CM042010.1_003196.1 | helper subject | chicory helpers showing both NRC0 and Ast-NRC1 clustering with sensors, unlike lettuce |
| CcNRC0 and Ast-CcNRC1 Helixer_NC_037542.1_000084.1 | helper subject | further Asterales helper clustering comparison |
| NRC2 | helper comparator | Solanales reference helper that Ast-NRCs group with phylogenetically |
| NRC4 | helper comparator | Solanales reference helper |
| NRC6 | helper cited-only | cited as a case of helper-sensor genetic clustering |
| Rx Rx-type sensors | sensor cited-only | the Solanales sensor class with no counterpart found anywhere in Asterales |
| ADR1 | helper cited-only | cited as a low-variability helper, consistent with the NRC-H result |
| CcRPP1 | sensor comparator | outgroup used to root the NRC network tree |
Effectors
No effector is used. Sensors were activated genetically as autoactive MHD D-to-V mutants, so helper dependency is measured without needing a cognate pathogen ligand for any lettuce sensor. None is known.
Structures
| Entry | State | Description |
|---|---|---|
| AlphaFold3 predicted | hexamer (helpers only); sensors failed to oligomerise | Hexamer models of 2 lettuce NRC helpers and 16 NRC sensors, 5 replicates each, with 50 oleic acids as a plasma membrane proxy. Helpers gave resistosome-like rings (pTM and ipTM above 0.5, mean per-chain pTM 0.75); sensors did not (below 0.5 and 0.7 respectively). LsatNRC0 never produced a complete N-terminal funnel. |
| AlphaFold3 predicted | monomer | Monomer models of representative group 1 and group 3 sensors used to map positively selected sites, including Helixer_NC_056630.2_001736.1 |
Open questions
- No cognate effector or pathogen ligand is known for any lettuce NRC sensor, so what these receptors actually detect is unknown.
- Whether the helper dependency profile inferred for Lactuca virosa, L. saligna and other Asterales holds experimentally is untested; it was assumed from phylogenetic grouping.
- Whether the high variability of group 1 sensors reflects relaxed selection or coevolution with pathogens is unresolved.
- The origin of SD-type sensors and whether the SD integration was lost in ancestral Asterales, rather than never gained, needs wider superasterid sampling.
- Structural studies of NRC-S in resting and active states are needed to explain how sensors activate helpers if they cannot oligomerise themselves.
- Protein accumulation was not assessed for sensor-helper pairs that produced no cell death, so some negatives may be expression failures rather than genuine incompatibility.
Connections in this corpus
| Paper | Why |
|---|---|
toghani-2026-sensor-helper-interface | Direct dependency, and the clearest build-on relationship in the corpus. Toghani takes the lettuce network characterised here, with LsatNRC0 and Ast-LsatNRC1 as the two helpers and the group-specific dependency profile as the baseline, and shows two substitutions in a group 1 sensor let it signal through Ast-LsatNRC1 as well. Pai establishes that helper dependency tracks phylogeny; Toghani shows how few residues hold that boundary in place. Shared authors Toghani, Frijters, Contreras, Kamoun. |
toghani-2026-sni-undecamer | Both are clade-level analyses of the NRC family from large-scale phylogenomics, and both use AlphaFold 3 hexamer modelling with oleic acids as a membrane proxy to sort receptors by their capacity to oligomerise. Pai uses the helper-versus-sensor separation as a validated signal; Toghani builds the Structural Novelty Index on exactly that prior. Shared authors Toghani, Sugihara, Kourelis, Contreras, Kamoun. |
marques-2026-legume-tissue | The two phylogenomics papers of the corpus, and a natural pair. Both use NLRtracker and RefPlantNLR to annotate NLRomes across a plant order and ask how the repertoire is partitioned, Pai by helper dependency and selection, Marques by tissue of expression. Both find that NLR classes expand unevenly within a lineage. Shared authors Pai, Kamoun. |
seager-2026-avrcap1b-nrc3 | Seager finds that a conserved NB-domain motif determines effector susceptibility across NRC3 orthologs, with Capsicum behaving differently from tomato and potato; Pai maps how NRC helpers and their sensors diverge across lineages. Whether effector-binding surfaces track the evolutionary trajectories Pai describes is an open question that spans the two. |
guo-2026-ccg10-resistosome | Pai counts 42 CCG10-NLRs in the lettuce NLRome, a class Guo characterises structurally for the first time. Pai gives the clade its abundance in a campanulid genome; Guo gives it an architecture. Together they bear on whether CCG10-NLR expansion tracks with the octameric assembly mode. |
ibrahim-2026-nrg1-chloroplast | Both use large-scale NLR phylogenetics to place a helper clade in evolutionary context, Ibrahim across 56,280 CCR-NLRs in land plants and Pai across the NRC network in asterids. Pai counts only 9 CCR-NLRs in lettuce, a useful data point on how differently the two helper systems have expanded in the same genome. |
Provenance and caveats
The paper is internally inconsistent about how many lettuce NRC sensors were tested. Results says 16 tested and 15 assessed for MADA; the Figure 3 legend says 14 expressed as autoactive mutants; the Discussion says 15 tested. The count of nine functional sensors is consistent throughout. [UNVERIFIED - do not state a tested-sensor count without resolving against the figure.]
The lettuce NRC total is given as 17 (2 helpers + 15 sensors) in Results and as 18/359 in the Discussion. [UNVERIFIED - the 5% figure is stated in both places.]
Genome counts differ between sections: Results says 40 Solanales genomes, Methods says 39 Solanales plus one S. melongena assembly from a separate source, which reconciles to 40. Sequence counts of 7,020 (Results, NRC network) and 7,040 (Methods, network plus CcRPP1 outgroup) probably differ by the outgroup but this is not stated.
Protein accumulation was not assessed for sensor-helper pairs that gave no cell death, so negative results in Figure 3A are not controlled for expression. The authors state this.
Sources
- Full citation: Pai H, Sakai T, Posbeyikian A, Frijters R, Sugihara Y, Contreras MP, Kourelis J, Adachi H, Kamoun S, Toghani A (2026) Contrasting evolutionary patterns of helper and sensor NRC NLRs in lettuce reflect functional divergence following subfunctionalization. PLoS Genet 22(7):e1012245.
- DOI: 10.1371/journal.pgen.1012245
- PDF on file:
papers/journal.pgen.1012245.pdf - Extraction:
data/papers/pai-2026-lettuce-nrc.json(schema v1.0) - References: 79 works, retrieved from OpenAlex
- Licence: Open Access (PLOS)