Pooled effector library screening in protoplasts rapidly identifies novel Avr genes

Abstract

Crop breeding for durable disease resistance is challenging due to the rapid evolution of pathogen virulence. While progress in resistance (R) gene cloning and stacking has accelerated in recent years^1-3, the identification of corresponding avirulence (Avr) genes in many pathogens is hampered by the lack of high-throughput screening options. To address this technology gap, we developed a platform for pooled library screening in plant protoplasts to allow rapid identification of interacting R-Avr pairs. We validated this platform by isolating known and novel Avr genes from wheat stem rust (Puccinia graminis f. sp. tritici) after screening a designed library of putative effectors against individual R genes. Rapid Avr gene identification provides molecular tools to understand and track pathogen virulence evolution via genotype surveillance, which in turn will lead to optimized R gene stacking and deployment strategies. This platform should be broadly applicable to many crop pathogens and could potentially be adapted for screening genes involved in other protoplast-selectable traits.

Fig. 1. Schematic of pooled effector library screening process to identify interacting R–Avr pairs.

Protoplast populations are co-transformed with a pooled library comprising hundreds of effector genes (E₁, E₂, E_n, Avr) from a pathogen along with either a known R gene or an empty vector (EV) for the negative control. The library MOT, defined as the number of plasmid molecules per cell (for each library construct), is chosen such that each library construct is expressed in an independent subpopulation of cells. Each cell individually receives a random but limited number of different constructs from the library together with either the empty vector or the R gene. In the presence of the R gene, protoplasts that express a matching Avr effector gene undergo cell death, while cells expressing the same Avr effector gene in the negative control remain alive. Living protoplasts are subsequently collected from both transformed populations and subjected to targeted (library-specific) RNA-seq. The expression of each effector in the library is then compared between the two samples (differential gene expression analysis). Avr gene candidates are identified by their decreased expression in the sample expressing the corresponding R gene.

Fig. 2. Optimization and validation of pooled effector library screening for rapid identification of interacting R–Avr pairs.

a, Wheat protoplasts were transformed with YFP and RFP reporter constructs at various MOTs (million plasmid molecules per cell) along with an empty vector whose amount was varied such that the combined MOT of the three constructs remained constant at 72 million plasmid molecules per cell. The percentage of cells showing fluorescence in the YFP, RFP or both wavelengths was determined by flow cytometry. Mean ± s.e.m. of 3 biological replicates. b, Wheat protoplasts were co-transformed with YFP, Sr50 and one of a series of mock libraries comprising AvrSr50 at various MOTs within a background of AvrSr35 (combined MOT of 36 million plasmid molecules per cell for the two constructs). Plot shows the percentage of YFP-positive living cells determined by flow cytometry. Mean ± s.e.m. of 3 biological replicates (dots), with relevant significant differences indicated (NS, not significant; two-tailed unpaired t-test assuming equal variances). c, Expression levels of AvrSr27-2 and AvrSr50 in a mock effector library screen. Wheat protoplasts were co-transformed with a mock library consisting of AvrSr50 (MOT of 0.14 million plasmid molecules per cell) and AvrSr27-2 (MOT of 0.14 million plasmid molecules per cell) within a background of AvrSr35 (MOT of 100 million plasmid molecules per cell) and either Sr50, Sr27 or an empty vector (MOT of 36 million plasmid molecules per cell). Relative expression levels of AvrSr27-2 and AvrSr50 were determined by targeted RNA-seq (shown in transcripts per million (TPM), normalized to AvrSr35 expression). Mean ± s.e.m. of 3 replicates, with significant differences indicated (two-tailed unpaired t-test assuming equal variances). d, Differential gene expression analysis of a pooled stem rust effector library comprising 696 predicted effectors co-transformed into wheat protoplasts with the R genes Sr50, Sr13c, Sr21, Sr22, Sr26 or Sr61 compared to the empty vector. Graphs show volcano plots of differential expression (x axis) versus P_adj (y axis) for each effector construct (dots). Effector gene constructs showing significantly reduced expression (red dots) within each treatment are labelled with their library ID number.

Fig. 3. Validation of AvrSr13 and AvrSr22 candidates.

a, Wheat protoplasts (cv. Fielder) were co-transformed with YFP, an Avr gene (AvrSr13, AvrSr22, AvrSr27-2 or AvrSr50) and an R gene (Sr13c, Sr22 or Sr27) or empty vector. b, YFP was co-transformed with AvrSr13 or AvrSr50 into protoplasts derived from wheat lines Kronos (KR, containing native Sr13a), Fielder (FL) or transgenic FL containing the Sr13c transgene (FL-Sr13c). c, The YFP reporter was co-transformed with AvrSr22 or AvrSr27-2 into protoplasts derived from wheat lines Schomburgk (SB, containing native Sr22), FL, transgenic FL containing the Sr22 transgene (FL-Sr22), or transgenic Robin containing a five-R-gene cassette including Sr22 (RB-Big5). Plots in a, b and c show the percentage of YFP-positive living cells determined by flow cytometry. Mean ± s.e.m. of 3 replicates, with significant differences indicated for relevant pairwise comparisons (two-tailed unpaired t-test assuming equal variances). All gene constructs were delivered at an MOT of 36 million plasmid molecules per cell. d, Agrobacterium-mediated transient co-expression of Sr27, Sr13c or Sr22 proteins (C-terminally fused to YFP) with AvrSr27-2, AvrSr13 or AvrSr22 (N-terminally fused to YFP) or YFP alone in N. tabacum leaves. Agrobacterial cultures were delivered at OD₆₀₀ of 0.4 (R gene constructs) or 0.7 (Avr gene constructs).