Evolutionary trade-offs at the Arabidopsis WRR4A resistance locus underpin alternate Albugo candida recognition specificities

The oomycete Albugo candida causes white rust of Brassicaceae, including vegetable and oilseed crops, and wild relatives such as Arabidopsis thaliana. Novel White Rust Resistance (WRR)-genes from Arabidopsis enable new insights into plant/parasite co-evolution. WRR4A from Arabidopsis accession Col-0 provides resistance to many but not all white rust races, and encodes a nucleotide-binding (NB), leucine-rich repeat (LRR) (NLR) immune receptor protein. Col-0 WRR4A resistance is broken by a Col-0-virulent isolate of A. candida race 4 (AcEx1). We identified an allele of WRR4A in Arabidopsis accession Oy-0 and other accessions that confers full resistance to AcEx1. WRR4AOy-0 carries a C-terminal extension required for recognition of AcEx1, but reduces recognition of several effectors recognized by the WRR4ACol-0 allele. WRR4AOy-0 confers full resistance to AcEx1 when expressed as a transgene in the oilseed crop Camelina sativa. Significance A C-terminal extension in an allele of the Arabidopsis resistance-protein WRR4A changes effector recognition specificity, enabling the WRR4AOy-0 allele to confer immunity to Albugo candida races that overcome the WRR4ACol-0 allele. This resistance can be transferred to the oil-producing crop Camelina sativa. Graphical abstract

selection on NLR genes compared to other genes (Kuang et al., 2004;Monteiro and 48 Nishimura, 2018; Meyers et al., 1998). To investigate NLR diversity, next-generation 49 sequencing technologies were combined with sequence capture to develop Resistance (R)-50 gene enrichment sequencing (RenSeq) (Jupe et al., 2013). This method has shed new light comprises several host-specific subclades, which includes race 2 from B. juncea, race 7 from 76 Brassica rapa, race 9 from Brassica oleracea and race 4 from crop relatives (e.g., Capsella 77 bursa-pastoris, Arabidopsis spp. and Camelina sativa) (Jouet et al., 2018;Pound and 78 Williams, 1963). These have been proposed to evolve by rare recombination events that 79 occurred between the races, followed by clonal propagation on susceptible hosts ( AcEx1 is also virulent in Camelina sativa, which is an emerging oilseed crop and has been 98 engineered to provide an alternative source of fish-oil-derived long chain omega-3 99 polyunsaturated fatty acids (LC-PUFAs). Algal-derived genes were expressed in the seed to from oily fish, and are acquired from feeding on algae-consuming plankton. In salmon farming, 104 the major source for of omega-3 oil derives from oceanic fish. Industry collects 750,000 metric 105 tons of fish oil every year, raising sustainability concerns (Napier et al., 2015). Transgenic 106 camelina oil is equivalent to fish oil for salmon feeding and for human health benefits (Betancor 107 et al., 2018;West et al., 2019). Despite challenges to distribute a product derived from a 108 genetically modified crop (Napier et al., 2019), an increase in camelina cultivation can be 109 expected in the near future. Fields of C. sativa will be exposed to A. candida and early 110 identification of R-genes will enable crop protection. 111 In this study we identified two alleles of WRR4A conferring full resistance to AcEx1 from 112 Arabidopsis accessions Oy-0 and HR-5. They both encode proteins with a C-terminal 113 extension compared to the Col-0 WRR4A allele. This extension enables recognition of at least 114 one effector from AcEx1. We propose that WRR4A Oy-0 is the ancestral state, and that in the 115 absence of AcEx1 selective pressure, an early stop codon in WRR4A generated the Col-0-116 like allele, enabling more robust recognition of other A. candida races while losing recognition of AcEx1. Finally, we successfully transferred WRR4A Oy-0 -mediated resistance to AcEx1 from 118 Oy-0 into Camelina sativa. 119 120 Results:

121
Resistance to AcEx1 is explained by WRR4A alleles of HR-5 and Oy-0 122 AcEx1 growth on Col-0 results in chlorosis that is not seen in the fully susceptible accession 123 Ws-2 (Figure 1a). Since WRR4A confers resistance to all other A. candida races tested and 124 Ws-2 lacks WRR4A, we tested if the chlorotic response could be explained by WRR4A, by 125 testing a Col-0_wrr4a-6 mutant, and found that it shows green susceptibility to AcEx1. We 126 also tested Ws-2 transgenic lines carrying WRR4A from Col-0 and observed chlorotic 127 susceptibility (Figure 1a). Thus, WRR4A from Col-0 weakly recognises AcEx1 and provides 128 partial resistance. However, AcEx1 is still able to complete its life cycle on Col-0.

138
In a search for more robust sources of AcEx1 resistance, we tested 283 Arabidopsis 139 accessions (Table S2). We identified 57 (20.1%) fully resistant lines, including Oy-0 and HR-140 5. We phenotyped 278 Recombinant Inbred Lines (RILs) between Oy-0 (resistant) and Col-0 141 (susceptible) and conducted a quantitative trait locus (QTL) analysis that revealed one major 142 QTL on chromosome 1 and two minor QTLs on chromosomes 3 and 5 ( Figure S1a). All loci 143 contribute to resistance, with a predominant contribution of the QTL on chromosome 1 (see 144  We conducted a bulk segregant analysis using an F2 population between HR-5 (resistant) and 165 Ws-2 (susceptible). RenSeq on bulked F2 susceptible segregants revealed a single locus on 166 chromosome 1, that maps to the same position as the chromosome 1 QTL in Oy-0 ( Figure  167 S4A). Since WRR4A Oy-0 confers resistance to AcEx1, we expressed its HR-5 ortholog, in genomic context, in the fully susceptible accession Ws-2, and found that WRR4A HR-5 also 169 confers full resistance to AcEx1 (Figure 1b). 170 In conclusion, WRR4A from Col-0 can weakly recognise AcEx1 but does not provide full 171 resistance. We identified two WRR4A alleles, in Oy-0 and HR-5, that confer complete AcEx1 172 resistance. nucleotide sequence for this extension is almost identical between HR-5, Oy-0 and Col-0 (two 185 polymorphic sites). Thus, by mutating TGA to TGC in Col-0, we could engineer an allele with 186 the extension, that we called WRR4A Col-0_LONG (Figure 3a). By mutating TGC to TGA in Oy-0, 187 we could engineer an Oy-0 allele without the extension, that we called WRR4A Oy-0_SHORT . We and which could contribute to the WRR4A-dependent chlorotic response to AcEx1 in Col-0 242 (Figure 1). In addition, WRR4A Col-0_LONG recognises CCG9 and CCG35 (Figure 3b). 243 Recognition of CCG9 and CCG35 is not explained solely by the C-terminal extension (as 244 WRR4A Oy-0 does not recognise them) or by the core region of the Col-0 allele (as WRR4A Col-0 245 does not recognise them). 246 WRR4A Col-0 can recognise eight CCG effectors from other races of A. candida (Redkar et al., 247 2021). We found that WRR4A Oy-0 is able to recognise CCG40, CCG104 and CCG28, but not 248 CCG67, CCG79, CCG33, CCG30 and CCG71 (Figure 3c). WRR4A Col-0_LONG recognises all 249 the CCGs indistinguishably from WRR4A Col-0 , indicating no influence of the C-terminal 250 extension on their recognition. 251 In conclusion, we identified one AcEx1 effector specifically recognised by WRR4A Oy-0 . The C-252 terminal extension is required and sufficient for its recognition. We also found that WRR4A Oy-253 0 does not recognise several of the Col-0-recognised CCG from other races.   an early stop codon resulting in a likely non-functional allele (Figure 4). The full-length Oy-0-326 like alleles are associated with resistance to AcEx1, while the Col-0-like alleles are associated 327 with susceptibility (Figure 3). The only exception is Kn-0, which displays a full length Oy-0-328 like allele but is susceptible to AcEx1. Susceptibility in Kn-0 could be explained by SNPs, lack 329 of expression or mis-splicing of WRR4A. 330

331
The Col-0 allele C-terminal truncation correlates with gain of recognition for some 332

CCGs and loss of recognition for others, suggesting an evolutionary trade-off 333
We propose that, in the absence of AcEx1 selection pressure, the Col-0-like early stop codon 334 occurred to provide a new function, along with the loss of AcEx1 effector recognition. This new 335 function enables recognition of more CCGs from other A. candida races. 336 By combining the C-terminal extension on WRR4A Oy-0 with the core region of WRR4A in Col-337 0 (Figure 3d), recognition of additional AcEx1 CCGs was enabled. Furthermore, Arabidopsis 338 natural accessions carrying the core region of the Col-0-like allele also lack the C-terminal 339 extension (Figure 4 and S6). This could be an example of intramolecular genetic suppression 340 suggesting that the mechanism of activation and the downstream signalling of WRR4A is 373 conserved, at least in Brassicaceae. 374 In conclusion, we found a novel example of post LRR polymorphism within an NLR family, 375 associated with diversified effector recognition spectra. By investigating the diversity of 376 WRRA, we identified an allele that confers white rust resistance in the camelina crop.