Arabidopsis thaliana genes contributing to differences in the outcome of infection with generalist and specialist strains of Turnip mosaic virus identified by genome-wide association studies

Pathogens can be classified as generalists or specialists depending on their host breadth. While generalists are able to successfully infect a wide variety of host species, the host range of specialists is limited to a few related species. Even though generalists seem to gain an advantage due to their wide host range, they usually pay a cost in terms of fitness within each host species (i.e., the jack-of-all trades, master of none). On the contrary, specialists have high fitness within their own host. A highly relevant yet poorly explored question is whether generalist and specialist viruses differ in the way they interact with their host’s gene expression networks. To identify host genetic factors relevant for the infection of specialist or generalist viruses, we undertook a genome-wide association study (GWAS) approach. Four hundred fifty natural accessions of Arabidopsis thaliana were inoculated with turnip mosaic potyvirus strains that were either generalist (TuMV-G) or specialist (TuMV-S). Several disease-related traits have been associated with different sets of host genes for each TuMV strain. While most of the mapped loci were traitor strain-specific, one shared locus was mapped for both strains, a disease resistance TIR-NBS-LRR class protein. Likewise, only one locus was found involved in more than one of the disease-related traits evaluated, a putative cysteine-rich receptor-like protein kinase 20. To validate these results, the corresponding null mutant plants were inoculated with TuMV-G or -S and the outcome of infection was characterized. Author summary Generalist and specialist viruses are commonly found in nature, where they have potential for epidemics, and are classified depending on their host breath. In this study we used a genome-wide association study to characterize differences in the genetic basis of both infection strategies from a host perspective. Our experimental setup consisted of 450 accessions of A. thaliana and two strains of TuMV. We found differences in the number of associated genes and their functions in disease-related traits. Results were validated by characterization of viral infections in null mutant plants deficient for a set of the identified genes.


Introduction
6 variation is explained by genotypic variation or the genotyped SNPs relative to the contribution 105 of environmental factors [24]. Therefore, heritability can help in understanding the observed 106 phenotype and its underlying genetic complexity.

107
Identifying host factors responsible for resistance or permissiveness to infection is essential in 108 the study of pathogens as it may be crucial for disease management. To identify some of these

8
The two strains of TuMV used in this study, TuMV-G and TuMV-S were obtained by evolving,

162
The virus-infected plant tissue was frozen in liquid N2 and homogenized and prepared with 163 ten volumes of inoculation buffer (50 mM KH2PO4 pH 7.0, 3% polyethylene glycol 6000, 10% 164 Carborundum) right before the mechanical inoculation. After that, the two TuMV strains were

199
The normality of the data was checked with SPSS version 25 (IBM Corp., Armonk NY, USA).

200
Untransformed phenotypic data were used in the GWAS. Since phenotyping was done in two 201 blocks (150 accessions + 300 accessions), block effects were accounted for in the GWAS analysis.

202
Out of ~10 million SNPs [38], 1,815,154 SNPs had a minor allele frequency higher than 0.05 203 for all phenotypes. The multiple testing problem is rooted in the assumption that each test 204 performed is independent of others, which is often not true due to linkage disequilibrium between 205 genetic markers. This can lead to false negatives and important genes related to our phenotype 206 might not be discovered [21,23]. Therefore, the false discovery rate (FDR) was used to deal with

Characterization of infection traits in natural accessions. 252
The 450 A. thaliana accessions were phenotyped for the infection with generalist (TuMV-G) and

306
In summary, for the infection caused by the pathogens studied, the genetic architecture of 307 AUDPS and infectivity phenotypes is relatively simple. It involves few SNPs for the two viral 308 strains analyzed while also having a detectable large effect SNP that is being responsible for the 309 majority of the observed phenotype. Symptoms severity is genetically more complex and  14 Most of the genes were unique for (i) the TuMV-G or the TuMV-S strain and (ii) the infection 322 trait as seen in Fig. 4. There was one locus shared between the two viral strains, both at 14 and 323 21 dpi, for symptoms severity: the aforementioned AT2G14080. The NBS-LRR genes are the 324 most numerous class of the R (resistance) genes in A. thaliana. Their effector recognition LRR 325 domains recognize specific pathogens and can lead to a hypersensitive immune response (HR) or 326 to an extreme resistance against the virus infection. An HR restricts the pathogen at the primary 327 infection site causing cell death followed by SAR that increases SA accumulation and expression 328 of pathogenesis-related genes [46,47].

329
Comparing the results at 14 and 21 dpi for TuMV-G, genes at 14 dpi seem related to a more

353
CpLEPA is a highly conserved chloroplastic translation factor that could assist viral transcription

376
For the AUSIPS, if mutant values have higher intervals than the WT it will mean that the virus 377 is able to cause stronger symptoms in the absence of these host genes. This is the case for the

437
In the analysis of the ten selected KO null mutants significant differences can be detected with

460
[58]. None of the genes found by these authors was pinpointed in our study but this could simply 461 reflect three major experimental differences: (i) Rubio et al. grew their plants in a natural setting 462 where they were exposed to a changing environment. The highly complex natural setting can 463 lead to much more heterogeneous gene regulations, as opposed to a controlled environment that 464 minimizes external abiotic and biotic stressors. It was shown before that differences in 465 temperature, light and water availability influence the response of the plant to a virus [59,60].  (Table 1)