NLR immune receptors can exhibit tissue-specific expression patterns across legume species
Legume NLRs are more than twice as tissue-specific as genes generally, and they lean towards roots while the rest of the transcriptome leans towards leaves.
Where an immune receptor is expressed is as much a part of its function as what it detects, and legumes make the point sharply. Across seven species, 57% of expressed NLRs differ between leaf and root, against 21% of genes overall, and the direction reverses: genes generally favour leaves 1.68 to 1, NLRs favour roots 1.45 to 1. The bias is not uniform across classes. TNPs are almost entirely root-specific, CCG10-NLRs lean root, CCR-NLRs lean leaf. Most striking, TIR-NLRs are mostly root-specific while the CCR-NLR helpers they signal through sit in leaves, which is hard to reconcile with a simple sensor-helper pairing. The dataset covers 28 legume genomes and 113 phylogenetic subclasses, and is offered as a resource for precision breeding.
Key findings (11)
Legume NLR repertoires vary widely, from 88 in Lupinus albus to 970 in Arachis hypogaea, correlating only moderately with proteome size (R = 0.59, p < 0.001). Three species (M. truncatula, A. hypogaea, P. alba) carry up to three times as many NLRs as their proteome size predicts, 645 to 931 each.
Evidence NLR identification across 28 legume genomes and 4 outgroups
Source Figures 1A and 1B
Legume NLRs fall into five monophyletic clades: CC-NLR, TIR-NLR, CCG10-NLR, CCR-NLR and TNP (TIR-NB-ARC-like with beta-propeller WD40 or tetratricopeptide repeats). CC-NLRs predominate, followed by TIR-NLRs, then CCG10-NLRs, with CCR-NLRs and TNPs least abundant. Monocot outgroups lack TIR-NLRs entirely. V. radiata and the early-diverging legumes P. cineraria, P. alba and C. canadensis are enriched for CCG10-NLRs.
Evidence NB-ARC domain alignment against RefPlantNLR; phylogenetic classification
Source Figures 1B; figure S2
Tissue identity accounts for more expression variation than species identity across the seven transcriptome species, which is what makes the cross-species comparison interpretable. Top differentially expressed genes are recognised tissue markers (leghemoglobin, nodulin-25, chitinase in roots), supporting data quality.
Evidence RNA-seq of leaf and root from seedlings at the three-leaf stage, three biological replicates, Illumina NovaSeq 6000, 150 bp paired-end, 40M reads per sample
Source Figure 2A; Results
NLRs are far more tissue-specific than genes in general and the bias runs the other way. Of 138,748 expressed genes, 21.47% were differentially expressed between tissues, favouring leaves 1.68 to 1. Of 1,336 NLR transcripts with enough expression for analysis, 57.04% were differentially expressed, favouring roots 1.45 to 1. That is a 2.66-fold enrichment for tissue-dependent expression.
Evidence differential expression analysis across seven legume species
Source Figure 2B; table S8
On average 21% of expressed NLRs are root-specific and 11.5% leaf-specific, with most (50-83%) detected in both tissues. M. truncatula is highest for root-specificity (30%) and L. sativus lowest (8%); P. sativum is highest for leaf-specificity (16%) and C. arietinum lowest (2%).
Evidence per-species tissue specificity quantification
Source Figure 3B; figure S4
Grass pea (Lathyrus sativus) and pea (Pisum sativum), two closely related species, break the pattern and show more leaf-specific than root-specific NLR expression.
Evidence per-species differential expression
Source Figure 2B; figure S4
NLR classes are deployed to different tissues. TNPs are strongly root-biased (86.7%, 13/15 root-specific). CCG10-NLRs favour roots (45.2% root-specific, 57/126, against 21.4% leaf-specific). CCR-NLRs favour leaves (40%, 27/67, against 19% root). CC-NLRs and TIR-NLRs sit in between, at roughly 25-26% leaf-specific and 32-35% root-specific.
Evidence class-resolved differential expression analysis
Source Figures 2B, 3C, 3D; tables S7, S8
TIR-NLRs and the CCR-NLR helpers they signal through show opposing tissue predominance: most TIR-NLRs are root-specific while their NRG1- and ADR1-subfamily helpers are predominantly leaf-expressed. The paper calls this surprising and does not resolve it.
Evidence class-resolved differential expression analysis
Source Results; table S8
NLRome size tracks with how much of it goes unused. C. arietinum, with the smallest NLRome, has the lowest proportion of non-expressed NLRs (11%) and tissue-specific NLRs (17%); M. truncatula, with the largest, has the highest of both (40% and 50%).
Evidence comparison across the expression dataset
Source Figure S4E
Tissue expression is phylogenetically structured. Across 113 NLR subclasses, clustering of leaf-specific and root-specific NLRs is driven by NLR class and species relatedness, with Hologalegina species grouping apart from Phaseoloids. Root-specific NLRs are expressed at markedly higher levels than leaf-specific ones. Root-specific CCR-NLRs all belong to the NRG1 subfamily, found in four species; TNPs are root-specific in five species, with M. truncatula the only species carrying a leaf-specific TNP.
Evidence phylogenetic analysis of tissue-specific NLRs with TMM expression mapping; iTOL v6 visualisation
Source Figures 4, 5A and 5B; tables S9, S10
Genome resource quality varies enough to matter. Fifteen of 28 legume genomes were highly contiguous (contig N50 above 1 Mbp) and 16 proteomes complete (under 5% missing BUSCOs), but only 11 of 32 proteomes met both criteria. Duplicated BUSCO scores ranged from 1.6% in C. arietinum to 95.2% in A. hypogaea. All proteomes were retained regardless.
Evidence assembly contiguity and BUSCO completeness assessment
Source Figures S1B and S1C; tables S1, S2
NLRs
| Name | Class | Role in this paper |
|---|---|---|
| CC-NLR class coiled-coil NLR | singleton subject | most abundant legume NLR class; ~26% leaf-specific and ~32% root-specific |
| TIR-NLR class Toll/interleukin-1 receptor NLR | sensor subject | second most abundant; mostly root-specific, in contrast to their leaf-expressed CCR-NLR helpers |
| CCG10-NLR class G10-subclade CC-NLR | singleton subject | root-favouring class (45.2% root-specific); overrepresented in V. radiata and early-diverging legumes |
| CCR-NLR class RPW8-like CC-NLR, RNL, NRG1 subfamily, ADR1 subfamily | helper subject | least abundant alongside TNPs; leaf-biased (40% leaf-specific); all root-specific members belong to the NRG1 subfamily |
| TNP class TIR-NB-ARC-like beta-propeller WD40/tetratricopeptide repeat | singleton subject | most tissue-restricted class, 86.7% root-specific; only M. truncatula has a leaf-specific TNP |
Effectors
No effector is studied. The work is comparative phylogenomics and transcriptomics; no pathogen challenge was performed.
Open questions
- Why TIR-NLRs are predominantly root-expressed while the CCR-NLR helpers they require are predominantly leaf-expressed is unexplained, and sits awkwardly with a straightforward sensor-helper dependency.
- Whether the root bias reflects adaptation to soil-borne pathogens, the demands of rhizobial symbiosis, or something else is proposed but not tested.
- Why grass pea and pea reverse the pattern towards leaf-specificity, unlike their close relatives, is not accounted for.
- Expression was measured in uninfected seedlings at one developmental stage, so nothing here addresses how these patterns change on infection or with age.
- Only 11 of 32 proteomes met both contiguity and completeness criteria, yet all were retained. How much assembly quality drives the observed NLR count variation is not separated out.
- Missing species representatives in many phylogenetic clusters could reflect genuine lineage-specific expansion or loss, or simply incomplete genomic resources; the paper says it cannot distinguish these.
Connections in this corpus
| Paper | Why |
|---|---|
pai-2026-lettuce-nrc | The two phylogenomics papers of the corpus and a natural pair. Both use NLRtracker and RefPlantNLR to annotate NLRomes across a lineage and ask how the repertoire is partitioned, Pai by helper dependency and selection pressure, Marques by tissue of expression. Both find NLR classes expanding unevenly within a lineage, and both quantify a class that the structural papers characterise mechanistically. Shared authors Pai, Kamoun. |
ibrahim-2026-nrg1-chloroplast | Both are large-scale surveys of the CCR-NLR clade reaching opposite kinds of conclusion. Ibrahim places NRG1 across 56,280 CCR-NLRs and 360 million years of land-plant divergence and shows what it does subcellularly; Marques shows where in the plant CCR-NLRs are expressed, finding them leaf-biased overall with all root-specific members belonging to the NRG1 subfamily. Compartmentalisation at two different scales, organelle and organ. |
guo-2026-ccg10-resistosome | Marques finds CCG10-NLRs are the third most abundant legume class and root-favouring, overrepresented in early-diverging legumes; Guo gives that clade its first resistosome structure. One supplies the abundance and expression context, the other the architecture. |
toghani-2026-sni-undecamer | Both scale NLR analysis across many genomes to find structure in a large family, Marques across 28 legume genomes by expression and class, Toghani across 346 Solanaceae genomes by predicted assembly. Both depend on consistent NLR annotation pipelines to make cross-genome comparison meaningful. |
gao-2026-vnt1-splicing | Both concern regulation of NLR availability rather than recognition specificity. Gao shows an NLR held inactive by an mRNA isoform until surface signalling removes it; Marques shows NLRs restricted to the tissue where they are needed. Transcript-level and organ-level control of the same underlying problem, keeping immune receptors off until and where they are wanted. |
Provenance and caveats
This paper reports classes and subclasses, not named receptors. It contributes no named NLR entities to the graph. Stage 2 alias reconciliation must keep class-level entries (CCR-NLR class, TNP class) separate from protein-level entries (NRG1, ADR1) even though the names overlap, or the entity graph will conflate a gene with its clade.
The abstract states 57% of expressed NLRs show predominant expression in one tissue with 34% root (451/1336) and 23% leaf (311/1336). The Results text gives per-species averages of 21% root-specific and 11.5% leaf-specific. These are different measures, predominance versus strict specificity, and must not be quoted interchangeably.
Only 11 of 32 predicted proteomes met both contiguity and completeness criteria and all were retained. NLR count comparisons across species therefore carry an unquantified assembly-quality confound, which the authors note but do not correct for.
The seedlings had spontaneously formed root nodules at harvest across all species. Root expression profiles therefore include nodulated tissue, which is not separated from root tissue proper.
This is a v1 preprint, not peer reviewed.
Sources
- Full citation: Marques RM, Santos C, Pai H, Vaz Patto MC, Kamoun S, Kourelis J (2026) NLR immune receptors can exhibit tissue-specific expression patterns across legume species. bioRxiv 2026.01.25.701577.
- DOI: 10.64898/2026.01.25.701577
- PDF on file:
papers/2026.01.25.701577v1.full.pdf - Extraction:
data/papers/marques-2026-legume-tissue.json(schema v1.0) - References: 85 works, retrieved from OpenAlex
- Licence: CC BY 4.0