PREreview of A biGWAS strategy reveals the genetic architecture of the interaction between wheat andBlumeria graminisf. sp.tritici

The following review was performed by members of the University of Illinois class IB465 Methods in Molecular Genetics and Genomics as part of their coursework looking at best practice in peer review and reporting of methods.

The manuscript “Joint GWAS reveals extensive R-Avr interactions shaping wheat-Bgt co- evolution” by Jingzhong Xie et al. presents the genomic analysis of a diverse wheat germplasm and its common pathogen Blumeria graminis f. sp. tritici (Bgt) isolates to further the genetic understanding of the host-disease interaction. Their study builds on previous work of the author in identifying R genes in wheat (Xie, J. et al, 2020), and they now aim to address the limitations of wheat host-pathogen interactions as imposed by the ‘one R for one Avr’ model. Inspired by the work of Wang, M. et al, 2018 and Zhang, F. et al, 2021, the authors conduct a joint genome- wide association analysis (GWAS), to map loci of plant resistance (R) genes and corresponding pathogenic virulence (Avr) genes. The joint GWAS relied on the sequencing of 245 Bgt isolates, the phenotyping of 120 genetically distinct isolates on a diverse panel of 581 wheat lines. The initial cross-species GWAS identified 251 R loci and 65 Avr loci with 212 suggested R-Avr interactions.

After identifying some potential R/Avr loci, the author selected four Avr loci and their most likely candidate genes for further study: Avr. locus18 (AvrN2), Avr. locus56 (AvrN16), Avr. locus22 (AvrN1) and Avr. locus11 (AvrN4). The author used the protoplast transfection assay and measured the fluorescence level first and found AvrN2, AvrN16, and AvrN1 showed significant reduction in fluorescence, indicating activation of cell death and validating that they were functional avirulence genes of their loci. And to validate the mapped R-Avr interactions, the author did the transient expression in Nicotiana benthamiana. The results showed AvrN2 was recognized by both Pm1a and Pm2a, thereby triggering cell death. AvrN2 is also recognized in wheat line W130, which lacks Pm1a and Pm2a, indicating that AvrN2 is recognized by at least one additional R gene, but the R gene is unknown yet. Overall, the molecular validation provided evidence for three new Avr genes and the new R-Avr interaction.

The authors’ work highlights the emerging complexity of R-Avr interaction networks, including complications when breeding R-genes into wheat lines with obstacles such as epistatic effects or direct R gene- R gene interaction and silencing. Strengths of this study include the large scale of the cross-species GWAS; while limitations include the study of novel Avr proteins through transient expression in Nicotiana, which may not fully represent the wheat-specific immune response. The study is novel as it is the first large- scale cross-species GWAS in a fungal-plant system. A key innovation of this study is the cross-species epistasis approach that they used to map 212 R-Avr interactions, which allowed them to gain a better understanding of the complex network. Being able to utilize biGWAS to study many different species is both a strength and a novel approach.

Statement of key claims

1.     Performing a biGWAS on the constructed diverse wheat panel and selecting 120 Bgt isolates identifies 251 R gene loci and 65 Avr gene loci, which include previously identified genes

2.     The biGWAS allows the mapping of R-Avr networks and reveals previously unknown interactions (unaddressed further)

3.     Three new functional Avr genes were identified, and new R-Avr interactions were verified with molecular assays

4.     Cross-species GWAS identifies R-Avr interactions with novel complexity.

5.     Pyramiding five major R loci has the potential to confer resistance to half of the Avr loci.

6.     There was little geographical preference for Avr alleles or their combinations, which was observed, suggesting that disease resistance breeding against this pathogen should be coordinated at the national level.

Overall we enjoyed reading this interesting and novel paper and found the claims to be justified by the data.

Comments

1.     The panel construction for wheat germplasm included a substantial proportion (more than 50% of lines) from the USDA germplasm collection, which represents a worldwide range of geographical origins for wheat lines. However, the Btg isolates assayed in this study were only sourced from China. Given the potential evolutionary differences between Btg isolates from Chinese populations vs Btg isolates from a broader geographic range, this not represented or recognized by the Chinese Btg isolates. It would be beneficial to mention the implications of this limitation in the discussion. It may strengthen the manuscript to discuss whether a lack of co-evolution between the globally diverse wheat panel and the China-derived Btg isolates may affect the detection power of the GWAS.

2.     The author states that all raw sequence data in this study were either generated (and made publicly available with the locations clearly stated) by the authors or downloaded from publicly available databases. However, in the methods if was unclear which data was publicly sourced and which was generated by the study, and whether downloaded data vs study study-generated data were processed in the same way. To increase transparency we recommend: clarify data source attribution for Bgt isolates and wheat lines, confirm the standardization of data processing for the variant calling pipeline and provide information on the preprocessed.

3.     The author briefly mentions whole-transcriptome mapping on 10 randomly selected lines to confirm SNP concordance between genotypic data and transcriptomic data. However, within the methods, I cannot easily find information on RNA-sequencing analysis or their pipeline, e.g., processing raw reads via trimming to mapping reads to SNP variant calling. To increase robustness and reproducibility and enable assessment of their results, we recommend including the RNA sequencing pipeline in full within the methods section.

4.     In Line 422, it would be beneficial for the readers if the authors could clarify what components are included in the biEpi map.

5.     For the transient expression in tobacco leaves, the leaf shown in Figure 7i has a mechanical wound near Pm5 b+AvrN2. Is the wound response affecting the result? If available we recommend replacing with another picture of this treatment.

6.     For Figure 7g, the color of the leaf makes the results (phenotype) difficult to identify and we recommend using another repeat picture of this treatment.

7.     For the Figure. 7b, d, e, please mark the significance in those figures.

8.     Within the introduction, the author is missing a citation to their own previous work (Xie, J. et al, 2020) describing their identification of a rare variant R gene Pm5e.

9.     In general, we thought the paper would benefit from a visual pipeline for the biGWAS, as the concept can be quite difficult to distinguish for unfamiliar readers. For example, this could help clarify early on how the phenotypic data and genotypic data are used to conduct two sequential/separate GWAS. Otherwise, readers might expect a parallel approach.

10.  Line 666 in the methods mentions ‘plants exhibiting high deviations were discarded’. Please clarify the values for ‘high deviation’ and how discarded plants were handled, e.g., were plants replaced? Or was the accession discarded altogether? Perhaps a supplementary table can be included to show which lines were discarded if that were the case.

11.  Line 668 ‘Two replicates per accession were planted in fields’, please specify if replicates were within the same field or distributed across different locations. Was this distribution consistent across the different years of field trials?

12.  Line 670 ‘Standard agronomic practices were followed.’ The author may wish to consider including details of the agronomic practices, as standards can vary, e.g., irrigation, pest control, and fertilizer use. Details such as fertilizer band and amount of usage should also be included, as this would aid reproducibility.

13.  Line 627, the author may wish to clarify if “average values” were for single-year data or across multiple years.

14.  Line 679 mentions repeating the process 6-10 times. Supplementary notes do not seem to mention if any control steps were taken to try and minimize the risk of contamination / minimize potential biases in isolated propagation. The author may wish to clarify the details of the process further.

15.  Line 682 discusses infection type scoring, which is well defined in the supplementary notes. However, the reasoning behind the scoring system does not seem to be included. It may be beneficial to include citations or references to support the reasoning of this system, e.g., to prove its validity if it is common in literature. This would increase the credibility of the protocol.

16.  Supplementary line 30 clearly lists the weekly schedule for phenotyping isolates, but the author may wish to consider including details on how selection of Bgt isolates occurred for each round. For example, if each 4-6 sets of phenotyping were included in one complete cycle, or if there was further grouping or considerations for potential batch effects.

17.  Lines 690-703 for the genome sequencing pipeline, the author lists the processing pipeline and mentions the use of quality control software such as fastp before aligning to reference genomes. Other steps are also listed with a confirmation that ‘default’ parameters were used for processing and variant discovery. For greater clarity for readers, the author may wish to consider listing the parameters in full and/or confirming the software versions, as parameters may change in software updates, which could impact the reproducibility of the methods in the future.

18.  Line 696- the paragraph seems to be mostly discussing Btg isolates, but the line regarding alignment mentions the wheat reference genome. If this is only Bgt isolates, the author may wish to clarify why Chinese Spring was used as a reference genome. If it is in relation to wheat sequencing data, the author may wish to amend the methods section to state this clearly, as currently the paragraph reads as only in relation to Bgt isolates.

19.  In Line 240, the author may wish to briefly discuss potential reasoning why the MLMM model would fail compared to other statistical models, and if this is something that could hold significance to the reader.

20.  Line 923 and the corresponding figure for 2b may benefit from density units being clarified for the y-axis to improve interpretability. The same advice can be applied to all figures that list ‘Density’ with no unit clarification.

21.  For the protoplast transfection assay, please list the plasmid construction map.

22.  For the luciferase activity assay, please provide the product code of the LUC substrate (luciferase).

23.  For transient expression, please provide more details about the bacterial cells' centrifugation conditions.

24.  For transient expression, please provide the process of Agrobacterium infiltration. For example, infiltrate the upper leaf surface or the lower leaf surface, and the infiltration area.

25.  For the Nicotiana benthamiana plant growth conditions, please provide the light intensity.

References

1.     Wang, M. et al. Two-way mixed-effects methods for joint association analysis using both host and pathogen genomes. Proc Natl Acad Sci U S A 115, E5440–E5449 (2018).

2.     Zhang, F. et al. Reciprocal adaptation of rice and Xanthomonas oryzae pv. oryzae: cross- species 2D GWAS reveals the underlying genetics. Plant Cell 33, 2538–2561 (2021).

3.     Xie, J. et al. A rare single nucleotide variant in Pm5e confers powdery mildew resistance in common wheat. New Phytol. 228, 1011–1026 (2020).

4.     Cesari, S. et al. The rice resistance protein pair RGA4/RGA5 recognizes the Magnaporthe oryzae effectors AVR-Pia and AVR1-CO39 by direct binding. Plant Cell 25, 1463-1481 (2013).

Competing interests

The authors declare that they have no competing interests.