Genome assembly and characterization of a complex zfBED-NLR gene-containing disease resistance locus in Carolina Gold Select rice with Nanopore sequencing

Abstract

Long-read sequencing facilitates assembly of complex genomic regions. In plants, loci containing nucleotide-binding, leucine-rich repeat (NLR) disease resistance genes are an important example of such regions. NLR genes constitute one of the largest gene families in plants and are often clustered, evolving via duplication, contraction, and transposition. We recently mapped the Xo1 locus for resistance to bacterial blight and bacterial leaf streak, found in the American heirloom rice variety Carolina Gold Select, to a region that in the Nipponbare reference genome is NLR gene-rich. Here, toward identification of the Xo1 gene, we combined Nanopore and Illumina reads and generated a high-quality Carolina Gold Select genome assembly. We identified 529 complete or partial NLR genes and discovered, relative to Nipponbare, an expansion of NLR genes at the Xo1 locus. One of these has high sequence similarity to the cloned, functionally similar Xa1 gene. Both harbor an integrated zfBED domain, and the repeats within each protein are nearly perfect. Across diverse Oryzeae, we identified two sub-clades of NLR genes with these features, varying in the presence of the zfBED domain and the number of repeats. The Carolina Gold Select genome assembly also uncovered at the Xo1 locus a rice blast resistance gene and a gene encoding a polyphenol oxidase (PPO). PPO activity has been used as a marker for blast resistance at the locus in some varieties; however, the Carolina Gold Select sequence revealed a loss-of-function mutation in the PPO gene that breaks this association. Our results demonstrate that whole genome sequencing combining Nanopore and Illumina reads effectively resolves NLR gene loci. Our identification of an Xo1 candidate is an important step toward mechanistic characterization, including the role(s) of the zfBED domain. Finally, the Carolina Gold Select genome assembly will facilitate identification of other useful traits in this historically important variety.

Fig 1. NLR proteins encoded in Carolina Gold Select in relation to Nipponbare and selected R genes.

(a) Number and chromosomal distribution of all NLR-Annotator predicted NLR genes in Carolina Gold Select and Nipponbare assemblies. ‘Pseudo’, predicted NLR genes with stop codons in any domain. ‘Partial’, predicted NLR genes missing a canonical domain. All NLR gene types are included in order to provide a high-level comparison of NLR distribution in the two assemblies. (b) Maximum likelihood tree of encoded NB-ARC domains of NLR genes in Carolina Gold Select and Nipponbare, as predicted by NLR-Annotator. Incomplete NLR genes and genes with a stop codon in the NB-ARC domain are not included in the phylogeny. Sixteen cloned resistance genes are included for reference. Branches with bootstrap support greater than 80 percent are indicated with pink squares. Interactive tree available at http://itol.embl.de/shared/acr242. NB-ARC domain sequences available in S3 Table. (c) Examples of expansion (top), contraction (middle) and transposition (bottom) of NLR genes in Carolina Gold Select relative to Nipponbare. Bootstrap values greater than 80 percent are displayed. Further details available in S4 Table. In the example of expansion at the Xo1 locus, as described in the text, CGS chr4 nlr9 is CGS-Xo1₁, CGS chr4 nlr10 is CGS-Xo1₂, CGS chr4 nlr12 is CGS-Xo1₄, and Nb chr4 nlr16 is Nb-xo1₁.

Fig 2. Expansion at the Carolina Gold Select Xo1 locus and identification of an Xo1 candidate.

(a) Comparison of the Xo1 locus in Carolina Gold Select and in Nipponbare. Areas of darker color on the two cartoon loci connected by gray shading represent regions of high similarity. Triangles indicate positions of NLR genes predicted by NLR-Annotator, designated from left to right as CGS-Xo1₁ through CGS-Xo1₁₄ in Carolina Gold Select and Nb-xo1₁ through Nb-xo1₇ in Nipponbare. Filled triangles indicate NLR genes expressed in leaf tissue during infection (see text and S9 Table). (b) An excerpt of the phylogenetic tree from Fig 1A containing the NLR genes at the Xo1 locus and two known resistance genes, Xa1 and Pi63. NLR genes encoding an integrated zfBED domain fall into two clades, which we designate as Xo1 clades I and II. Branches with bootstrap support greater than 80 percent are indicated with pink squares. Interactive tree available at http://itol.embl.de/shared/acr242.

Fig 3. Structural comparison of the Xo1 candidate CGS-Xo1₁₁ with cloned R gene Xa1 and with Nb-xo1₅.

(a) All amino acid polymorphisms upstream of the LRR in the three predicted gene products. (b) Cartoon alignment of predicted products of CGS-Xo1₁₁, Xa1, and Nb-xo1₅ showing the zfBED domains, nuclear localization signals (NLS), coiled coil domains (CC), NB-ARC domains, tandem repeats, and final repeats. Lighter shade of color for the repeats of Nb-xo1₅ reflects their greater relative divergence. Synonymous and nonsynonymous nucleotide substitutions in relation to CGS-Xo1₁₁ are indicated by dashed and solid red lines respectively. (c) WebLogos showing amino acid conservation of the tandem repeats in each LRR (d) Heatmap of repeat unit nucleotide sequence percent identity within and among the three coding sequences. Nb-xo1₅ encodes an additional, cryptic repeat that does not align and is not included in (c) or (d).

Fig 4. zfBED-NLR proteins across the Oryzeae.

(a) Xo1 clades I and II from an NB-ARC domain-based maximum likelihood tree of 5,078 predicted NLR proteins from representative Oryzeae genomes. Clade I proteins are indicated by orange shading, clade II by purple, and presence of zfBED domain by dark purple. Numbers of tandem 279 bp C-terminal repeats, where present, are given. Additional detected, non-canonical NLR gene motifs are noted. Red branches correspond to NLR genes not on chromosome four. Predicted NLRs with stops in the NB-ARC domain are annotated with asterisks. Full Oryzeae tree in S3 Fig and interactive tree available at http://itol.embl.de/shared/acr242. Nb Chr8 nlr 18 was used as an outgroup and can be viewed in the interactive tree. (b) Maximum likelihood tree of the 36 predicted Oryzeae zfBED-NLR proteins based on the zfBED domain amino acid sequence (zfBED sequences and nucleotide tree in S7 Table and S4 Fig). In a) and b), branches with bootstrap support greater than 80 percent are indicated with pink squares. Interactive trees available at http://itol.embl.de/shared/acr242.