Summary
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.
Introduction
Plants have evolved powerful defense mechanisms that can arrest attempted colonization by microbial pathogens. Timely defense activation requires perception of pathogen-derived molecules by cell-surface pattern-recognition receptors (PRRs) and intracellular Nucleotide-binding (NB), Leucine-rich repeat (LRR), or NLR, immune receptors (Jones and Dangl, 2006). Extensive NLR genetic diversity within plant populations is associated with robustness of NLR-mediated immunity (Baggs et al., 2017), and plant NLR sequences reveal diversifying selection on NLR genes compared to other genes (Kuang et al., 2004; Monteiro and Nishimura, 2018; Meyers et al., 1998). To investigate NLR diversity, next-generation sequencing technologies were combined with sequence capture to develop Resistance (R)-gene enrichment sequencing (RenSeq) (Jupe et al., 2013). This method has shed new light on NLR repertoires in several plant genomes including tomato, potato and wheat (Andolfo et al., 2014; Witek et al., 2016; Steuernagel et al., 2016). A comparison of 64 Arabidopsis thaliana (Arabidopsis) accessions using RenSeq documented NLR sequence diversity within a single species, revealing the Arabidopsis “pan-NLRome” (Van de Weyer et al., 2019). Each Arabidopsis accession contains 150-200 NLR-encoding genes. About 60% are found in clusters (within 200 kb from each other) that show copy number variation (Lee and Chae, 2020). From all the NLRs of the 64 accessions, 10% are singletons and the rest are distributed between 464 orthogroups. Each accession contains a unique subset comprising, on average, 25% of the orthogroups.
NLRs vary in their intramolecular architecture. Plant NLR proteins usually display either a “Toll, Interleukin-1, R-gene” (TIR), a “Coiled-Coil” (CC) or “Resistance to Powdery mildew 8” (RPW8) N-terminal domain, a central NB domain and a C-terminal LRR domain. Some NLRs also comprise a post-LRR (PL) domain. For example, RRS1 is an Arabidopsis TIR-NLR with PL WRKY domain required to detect the effectors AvrRps4 (from the bacterium Pseudomonas syringae) and PopP2 (from the bacterium Ralstonia solanacearum). The integrated WRKY is called a decoy as it mimics the authentic AvrRps4 and PopP2 effector targets (Sarris et al., 2015; Le Roux et al., 2015). Several other integrated decoy domains have been described (Cesari, 2017). RPP1 and Roq1, two TIR-NLRs from Arabidopsis and Nicotiana benthamiana respectively, form tetrameric resistosomes upon activation. In this structure, a PL C-terminal jelly-roll/Ig-like domain (C-JID) physically binds the cognate effector, along with the LRR domain (Ma et al., 2020; Martin et al., 2020). Analysis of NLR integrated domains can potentially reveal novel effector targets (Kroj et al., 2016).
A. candida causes white blister rust in Brassicaceae and serious annual yield losses in brassica crops such as oilseed mustard (Brassica juncea) in India (Gupta et al., 2018). It comprises several host-specific subclades, which includes race 2 from B. juncea, race 7 from Brassica rapa, race 9 from Brassica oleracea and race 4 from crop relatives (e.g., Capsella bursa-pastoris, Arabidopsis spp. and Camelina sativa) (Jouet et al., 2018; Pound and Williams, 1963). These have been proposed to evolve by rare recombination events that occurred between the races, followed by clonal propagation on susceptible hosts (McMullan et al., 2015). The Arabidopsis Columbia (Col-0) allele of WRR4A can confer resistance to isolates of all four races (Borhan et al., 2010; Borhan et al., 2008). The allele encodes a canonical TIR-NLR and belongs to an orthogroup of three genes in Col-0 at the same locus. The accession Ws-2 (susceptible to A. candida race 4) lacks WRR4A but contains the two other paralogs, illustrating intra-species copy number variation within clusters. Interestingly, one of these paralogs, WRR4B, also confers resistance to the Ac2V isolate of race 2 (Cevik et al., 2019). In addition, the CC-NLR-encoding BjuWRR1, which confers resistance to several candida isolates collected on B. juncea, was mapped and cloned from the European accession of B. juncea Donskaja-IV (Arora et al., 2019).
Several Col-0-virulent isolates of A. candida race 4 have been collected from naturally infected Arabidopsis plants. They were used subsequently to identify an alternative source of broad-spectrum white rust resistance. One was found in Arabidopsis accession Oy-0 that mapped to the WRR4 locus (Fairhead, 2016). Interestingly, one of these isolates (AcEx1), as well as the related white rust pathogen Albugo laibachii, were shown to suppress non-host resistance in Arabidopsis to the potato late blight pathogen Phytophthora infestans (Prince et al., 2017; Belhaj et al., 2017). Thus, we set out to clone the gene conferring AcEx1 resistance in Oy-0, and characterise the corresponding pathogen effector(s).
AcEx1 is also virulent in Camelina sativa, which is an emerging oilseed crop and has been engineered to provide an alternative source of fish-oil-derived long chain omega-3 polyunsaturated fatty acids (LC-PUFAs). Algal-derived genes were expressed in the seed to produce eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) as an improved LC-PUFA source for fish feed or as a human nutritional supplement (Ruiz-Lopez et al., 2014; Petrie et al., 2014). EPA and DHA, recognised for their health benefits, are mainly sourced from oily fish, and are acquired from feeding on algae-consuming plankton. In salmon farming, the major source for of omega-3 oil derives from oceanic fish. Industry collects 750,000 metric tons of fish oil every year, raising sustainability concerns (Napier et al., 2015). Transgenic camelina oil is equivalent to fish oil for salmon feeding and for human health benefits (Betancor et al., 2018; West et al., 2019). Despite challenges to distribute a product derived from a genetically modified crop (Napier et al., 2019), an increase in camelina cultivation can be expected in the near future. Fields of C. sativa will be exposed to A. candida and early identification of R-genes will enable crop protection.
In this study we identified two alleles of WRR4A conferring full resistance to AcEx1 from Arabidopsis accessions Oy-0 and HR-5. They both encode proteins with a C-terminal extension compared to the Col-0 WRR4A allele. This extension enables recognition of at least one effector from AcEx1. We propose that WRR4AOy-0 is the ancestral state, and that in the absence of AcEx1 selective pressure, an early stop codon in WRR4A generated the Col-0-like allele, enabling more robust recognition of other A. candida races while losing recognition of AcEx1. Finally, we successfully transferred WRR4AOy-0-mediated resistance to AcEx1 from Oy-0 into Camelina sativa.
Results
Resistance to AcEx1 is explained by WRR4A alleles of HR-5 and Oy-0
AcEx1 growth on Col-0 results in chlorosis that is not seen in the fully susceptible accession Ws-2 (Figure 1a). Since WRR4A confers resistance to all other A. candida races tested and Ws-2 lacks WRR4A, we tested if the chlorotic response could be explained by WRR4A, by testing a Col-0_wrr4a-6 mutant, and found that it shows green susceptibility to AcEx1. We also tested Ws-2 transgenic lines carrying WRR4A from Col-0 and observed chlorotic susceptibility (Figure 1a). Thus, WRR4A from Col-0 weakly recognises AcEx1 and provides partial resistance. However, AcEx1 is still able to complete its life cycle on Col-0.
In a search for more robust sources of AcEx1 resistance, we tested 283 Arabidopsis accessions (Table S2). We identified 57 (20.1%) fully resistant lines, including Oy-0 and HR-We phenotyped 278 Recombinant Inbred Lines (RILs) between Oy-0 (resistant) and Col-0 (susceptible) and conducted a quantitative trait locus (QTL) analysis that revealed one major QTL on chromosome 1 and two minor QTLs on chromosomes 3 and 5 (Figure S1a). All loci contribute to resistance, with a predominant contribution of the QTL on chromosome 1 (see Figure 3.7 of Fairhead, 2016). We did not investigate the minor QTL on chromosome 5. Fine mapping on chromosome 1 and 3 QTLs refined the QTL boundaries (Figure S2 and S3, see Experimental Procedures). Based on sequence identity between the QTL in Col-0 and in an Oy-0 RenSeq dataset (Van de Weyer et al., 2019), we identified four NLRs associated with the QTLs in Oy-0: three TIR-NLR paralogs on chromosome 1 (WRR4A, WRR4B and one absent in Col-0 that we called WRR4D) and a CC-NLR absent in Col-0 on chromosome 3 (that we called Candidate to be WRR11, CWR11) (Figure S1B-C).
We expressed these genes, with their own promoters and terminators, in the fully susceptible accession Ws-2. Only WRR4AOy-0 conferred full resistance (Figure 1b). CWR11, the only NLR from the WRR11 locus, does not confer AcEx1 resistance. The WRR11 locus is orthologous to the position 17.283-18.535 Mb on chromosome 3 of Col-0 (Figure S3). The RPW8 alleles of Col-0 are encoded between positions 18.722-18.734 Mb. Oy-0 carries 15 copies of RPW8 (Van de Weyer et al., 2019), so it could be that some copies recombined on the other side of the WRR11 locus. We cloned one homolog of RPW8.1 from Oy-0 and expressed it with its own promoter and terminator in Ws-2 (Figure 1). Most transgenic lines were susceptible to AcEx1 but four lines were resistant. These lines were also smaller than the other lines, suggesting an autoimmune phenotype. Since most transgenic lines are susceptible and ectopic expression of RPW8 is known to result in autoimmunity (Xiao et al., 2003), we did not further investigate the role of RPW8Oy-0 paralogs in AcEx1 resistance. The gene underlying WRR11 locus resistance remains unknown.
We conducted a bulk segregant analysis using an F2 population between HR-5 (resistant) and Ws-2 (susceptible). RenSeq on bulked F2 susceptible segregants revealed a single locus on chromosome 1, that maps to the same position as the chromosome 1 QTL in Oy-0 (Figure S4A). Since WRR4AOy-0 confers resistance to AcEx1, we expressed its HR-5 ortholog, in genomic context, in the fully susceptible accession Ws-2, and found that WRR4AHR-5 also confers full resistance to AcEx1 (Figure 1b).
In conclusion, WRR4A from Col-0 can weakly recognise AcEx1 but does not provide full resistance. We identified two WRR4A alleles, in Oy-0 and HR-5, that confer complete AcEx1 resistance.
WRR4ACol-0carries an early stop codon compared to WRR4AOy-0
To understand why the Oy-0 and HR-5 alleles of WRR4A confer full resistance to AcEx1, while the Col-0 allele does not, we compared the gene and protein sequences (Figure 2). First, we defined the cDNA sequence of WRR4AOy-0. The splicing sites are identical between the two alleles. There are 46 polymorphic amino acids between Col-0, HR-5 and Oy-0. Col-0 shares 96.03% identity with Oy-0 and 96.23% with HR-5, while Oy-0 and HR-5 share 97.15% identity. WRR4ACol-0 carries a 156-nucleotide insertion in the first intron compared to Oy-0 and HR-5. A striking polymorphism is a TGC->TGA mutation in WRR4ACol-0, resulting in an early stop codon compared to WRR4AOy-0 and WRR4AHR-5 (Figure 2), located 178 amino acids after the last LRR, resulting in an 89 amino acid extension in WRR4AOy-0 and WRR4AHR-5. The nucleotide sequence for this extension is almost identical between HR-5, Oy-0 and Col-0 (two polymorphic sites). Thus, by mutating TGA to TGC in Col-0, we could engineer an allele with the extension, that we called WRR4ACol-0_LONG (Figure 3a). By mutating TGC to TGA in Oy-0, we could engineer an Oy-0 allele without the extension, that we called WRR4AOy-0_SHORT. We expressed these alleles, as well as the WT Col-0 and Oy-0 alleles, with their genomic context, in the AcEx1-compatible accession Ws-2. None of the WRR4ACol-0_LONG and WRR4AOy-0_SHORT transgenic seeds germinated, suggesting autoactivity in Arabidopsis. We tried to generate Arabidopsis Col-0 lines with WRR4ACol-0-STOP using CRISPR adenine base editor (see Experimental Procedures). Out of 24 transformed plants, none displayed editing activity at all. Thus, we did not generate stable WRR4A stop codon mutants in Arabidopsis. However, we cloned these alleles under the control of the 35S promoter and the Ocs terminator for transient overexpression in N. tabacum.
Since many TIR-NLRs carry a PL C-JID, we conducted a Hidden Markov Model (HMM) search for C-JID and found one in WRR4A (www.ebi.ac.uk/Tools/hmmer/search/hmmsearch on Arabidopsis thaliana using C-JIC HMM previously reported (Ma et al., 2020), e-value = 5.7e-14). This C-JID is present in both Oy-0 and Col-0 alleles (Figure 2b). The C-terminal extension in WRR4AOy-0 relative to WRR4ACol-0 does not show homology with known protein domains.
Extension in WRR4A confers specific recognition of AcEx1 candidate effectors
We tried to identify AcEx1 effectors specifically recognised by WRR4AOy-0. We tested for hypersensitive response (HR), as typical marker of NLR activation, upon transient WRR4AOy-0 expression along with AcEx1 effector candidates in N. tabacum leaves. Secreted CxxCxxxxxG (or CCGs) proteins are expanded in the genomes of Albugo species and are effector candidates (Kemen et al., 2011, Furzer et al., 2021). We identified 55 CCGs in the AcEx1 genome (Jouet et al., 2018, Redkar et al., 2021), and PCR-amplified and cloned 21 of them, prioritizing those that showed allelic variation with other races. Eight CCGs (i.e. CCG9, CCG12, CCG17, CCG33, CCG35, CCG39, CCG66, CCG80) induced a WRR4AOy-0-dependent HR in N. tabacum in a preliminary screen with high Agrobacterium concentration (OD600 = 0.5) and were further investigated (Figure S5). We co-delivered four alleles of WRR4A with the eight candidates at lower Agrobacterium concentrations (OD600 = 0.4) to test the robustness of their recognition (Figure 3). At this concentration, only CCG39 is consistently recognised by WRR4AOy-0 and explains AcEx1 resistance in Oy-0. WRR4ACol- 0_LONG can also recognise CCG39, but WRR4AOy-0_SHORT cannot. Hence, the C-terminal extension fully explains the acquisition of recognition of CCG39. We also observed a weak recognition of CCG33 by WRR4ACol-0, as shown in complementary work (Redkar et al., 2021) and which could contribute to the WRR4A-dependent chlorotic response to AcEx1 in Col-0 (Figure 1). In addition, WRR4ACol-0_LONG recognises CCG9 and CCG35 (Figure 3b). Recognition of CCG9 and CCG35 is not explained solely by the C-terminal extension (as WRR4AOy-0 does not recognise them) or by the core region of the Col-0 allele (as WRR4ACol-0 does not recognise them).
WRR4ACol-0 can recognise eight CCG effectors from other races of A. candida (Redkar et al., 2021). We found that WRR4AOy-0 is able to recognise CCG40, CCG104 and CCG28, but not CCG67, CCG79, CCG33, CCG30 and CCG71 (Figure 3c). WRR4ACol-0_LONG recognises all the CCGs indistinguishably from WRR4ACol-0, indicating no influence of the C-terminal extension on their recognition.
In conclusion, we identified one AcEx1 effector specifically recognised by WRR4AOy-0. The C-terminal extension is required and sufficient for its recognition. We also found that WRR4AOy-0 does not recognise several of the Col-0-recognised CCG from other races.
WRR4A alleles carrying a C-terminal extension are associated with AcEx1 resistance
The NLR repertoire of 64 Arabidopsis accessions has been determined using resistance gene enrichment Sequencing (RenSeq) (Van de Weyer et al., 2019). We found 20 susceptible and 5 resistant genotypes that belong to the 64 accessions (Table S2). We retrieved WRR4A from these 25 accessions (http://ann-nblrrome.tuebingen.mpg.de/apollo/jbrowse/). The read coverage was insufficient to resolve WRR4A sequence in Bur-0 (susceptible) and Mt-0 (resistant). WRR4A is absent from the WRR4 cluster in Ws-2, Edi-0 and No-0. Consistently, these accessions are fully susceptible to AcEx1. From the DNA sequence of the 20 other accessions, we predicted the protein sequence, assuming that the splicing sites correspond to those in Col-0 and Oy-0 (Figure 4). There are two well-defined clades of WRR4A alleles, with a bootstrap value of 100. One clade includes Col-0; the other clade includes Oy-0. The Col-0-like and Oy-0-like clades are also discriminated in a phylogeny based on predicted protein sequences (Figure S6). All alleles from the Col-0 clade carry TGA (apart from ULL2-5, TGC, but WRR4A is pseudogenised in this accession), while all alleles from Oy-0 clade carry TGC, at the Col-0 stop codon position. Several alleles from both clades, including Bay-0, ULL2-5, Wil-2, Ler-0, Ws-0 and Yo-0, carry an early stop codon (i.e. upstream of the Col-0 stop codon position), so the resulting proteins are likely not functional. Consistently, all the accessions from the Col-0 clade and all the accessions carrying an early stop codon are susceptible to AcEx1. The only exception is Kn-0, that carries an Oy-0-like allele of WRR4A but is susceptible to AcEx1. Otherwise, the presence of an Oy-0-like C-terminal extension associates with resistance.
AcEx1 resistance can be transferred from Arabidopsis to Camelina
AcEx1 can grow on Camelina sativa (Figure 5). Like Arabidopsis, it can be transformed using the floral dip method (Liu et al., 2012). We generated a WRR4AOy-0-transgenic camelina line. Out of twelve individuals, eight showed complete resistance, three displayed a chlorotic response (likely WRR4A-mediated HR) and one enabled limited pustule formation. All twelve WT camelina control plants showed mild to severe white rust symptoms. This indicates that WRR4AOy-0 can confer resistance to AcEx1 in C. sativa.
Discussion
Col-0 and HR-5 WRR4A alleles recognise effectors from AcEx1
A screen for novel sources of resistance to AcEx1 identified accessions HR-5 and Oy-0 as worthy of further investigation. Positional cloning from Oy-0 and then allele mining in HR-5 showed that this immunity is mediated by alleles of WRR4A in HR-5 and Oy-0 with distinct recognition capacities compared to the Col-0 allele. In Oy-0, two additional dominant loci, WRR11 on chromosome 3 and WRR15 on chromosome 5 contribute resistance to AcEx1 but the molecular basis of these resistances was not defined. Further investigation on WRR11 was conducted but did not reveal the causal gene (Castel, 2019, Chapter 3).
WRR4AOy-0 recognizes at least one AcEx1 effector that is not recognized by WRR4ACol-0 (Figure 3). Conceivably, WRR4AOy-0 could be combined with WRR4ACol-0 and WRR4BCol-0 to expand the effector recognition spectrum of a stack of WRR genes that could be deployed in juncea or C. sativa.
WRR4A alleles fall into two clades that can or cannot confer AcEx1 resistance
Analysis of WRR4A allele diversity in Arabidopsis revealed WRR4AOy-0-like and WRR4ACol-0-like alleles. Since WRR4ACol-0-like alleles show near-identity in nucleotide sequence after the premature stop codon to WRR4AOy-0-like alleles, the latter are likely to be ancestral, and the WRR4ACol-0-like early stop codon occurred once, in the most recent common ancestor of Sf-2 and Col-0. Other early stop codons, resulting in loss-of-function proteins, occurred randomly in both Oy-0-and Col-0-containing clades. About a third of the investigated accessions contain an early stop codon resulting in a likely non-functional allele (Figure 4). The full-length Oy-0-like alleles are associated with resistance to AcEx1, while the Col-0-like alleles are associated with susceptibility (Figure 3). The only exception is Kn-0, which displays a full length Oy-0-like allele but is susceptible to AcEx1. Susceptibility in Kn-0 could be explained by SNPs, lack of expression or mis-splicing of WRR4A.
The Col-0 allele C-terminal truncation correlates with gain of recognition for some CCGs and loss of recognition for others, suggesting an evolutionary trade-off
We propose that, in the absence of AcEx1 selection pressure, the Col-0-like early stop codon occurred to provide a new function, along with the loss of AcEx1 effector recognition. This new function enables recognition of more CCGs from other A. candida races.
By combining the C-terminal extension on WRR4AOy-0 with the core region of WRR4A in Col-0 (Figure 3d), recognition of additional AcEx1 CCGs was enabled. Furthermore, Arabidopsis natural accessions carrying the core region of the Col-0-like allele also lack the C-terminal extension (Figure 4 and S6). This could be an example of intramolecular genetic suppression (Kondrashov et al., 2002; Schülein et al., 2001; Brasseur et al., 2001; Davis et al., 1999). The combination between the core region of the Col-0 allele with the C-terminal extension may form a hyper-active WRR4A allele with excessive fitness cost for the plant. The early stop codon may have occurred in Col-0 to compensate for hyper-activation of an ancestral WRR4A allele. Hyper-activation of the immune system is deleterious, as shown for example by hybrid incompatibility caused by immune receptors (Wan et al., 2021).
Many plant TIR-NLRs show homology in their post-LRR domains (Saucet et al., 2020). The recently published structure of Roq1, a TIR-NLR from N. benthamiana, reveals a clear-cut C-terminal jelly roll/Ig-like domain (C-JID) after the LRR (Martin et al., 2020). It interacts with the Roq1 cognate effector XopQ. Unlike the LRR, the C-JID does not play a role in auto-inhibition. Similarly, in RPP1 (TIR-NLR from Arabidopsis), a C-JID is observed after the LRR (Ma et al., 2020). It physically interacts with the cognate effector ATR1 in the active resistosome. We found that WRR4A also contains a PL C-JID. Both WRR4ACol-0- and WRR4AOy-0-carry the C-JID, so it does not explain the unique CCG recognition of each allele.
Arabidopsis WRR4A resistance to AcEx1 can be transferred to the crop camelina
Camelina sativa was recently engineered to produce LC-PUFAs, an essential component in the feed used in fish farming (Petrie et al., 2014). Currently, fish farming uses wild fish-derived fish oil. Fish oil-producing camelina offers a solution to reduce the need for wild fish harvesting, potentially reducing pressure on world marine fish stocks (Betancor et al., 2018). There are challenges in delivering products derived from transgenic crops but fish oil-producing crops could reduce the environmental impact of fish farming. White rust causes moderate symptoms on camelina. However, A. candida is capable of immunosuppression (Cooper et al., 2008). A. candida-infected fields constitute a risk for secondary infection of otherwise incompatible pathogens. To safeguard camelina fields against white rust, both chemical and genetic solutions are possible. Genetic resistance offers the advantage of a lower cost for farmers and reduces the need for fungicide release in the environment. Since the first report of white rust on camelina in France in 1945, no genetic resistance has been characterised. All the strains collected on camelina can grow equally irrespective of camelina cultivar (Séguin-Swartz et al., 2009). This absence of phenotypic diversity precludes discovery of resistance loci using classic genetic tools. We found that WRR4AOy-0 confers resistance to AcEx1 in camelina (Figure 5). Arabidopsis WRR4A resistance is functional in B. juncea and B. oleracea, suggesting that the mechanism of activation and the downstream signalling of WRR4A is conserved, at least in Brassicaceae.
In conclusion, we found a novel example of post LRR polymorphism within an NLR family, associated with diversified effector recognition spectra. By investigating the diversity of WRRA, we identified an allele that confers white rust resistance in the camelina crop.
Experimental procedures
Plant material and growth conditions
Arabidopsis thaliana (Arabidopsis) accessions used in this study are Øystese-0 (Oy-0, NASC: N1436), HR-5 (NASC: N76514), Wassilewskija-2 (Ws-2, NASC: N1601) and Columbia (Col-0, NASC: N1092). Col-0_wrr4a-6 mutant is published (Borhan et al., 2008). Seeds were sown directly on compost and plants were grown at 21°C, with 10 hours of light and 14 hours of dark, 75% humidity. For seed collection, 5-weeks old plants were transferred under long-day condition: 21°C, with 16 hours of light and 8 hours of dark, 75% humidity. For Nicotiana tabacum (cultivar Petit Gerard) and Camelina sativa (cultivar Celine), seeds were sown directly on compost and plants were grown at 21°C, with cycles of 16 hours of light and 8 hours of dark, at 55% humidity.
Albugo candida Infection assay
For propagation of Albugo candida, zoospores from the infected leaf inoculum were suspended in water (∼105 spores/ml) and incubated on ice for 30 min. The spore suspension was then sprayed on plants using a Humbrol® spray gun (∼700 μl/plant) and plants were incubated at 4°C in the dark overnight to promote spore germination. Infected plants were kept under 10-hour light (20 °C) and 14-hour dark (16°C) cycles. Phenotypes were monitored 14 to 21 days after inoculation.
QTL analysis
QTL mapping of the bipartite F8 Oy-0 x Col-0 population (470 Recombinant Inbreed Lines, RILs, publiclines.versailles.inra.fr/page/27) (Simon et al., 2008) was performed on a genetic map of 85 markers across the five linkage groups that accompanied the population using RQTL (Broman et al., 2003). Standard interval mapping using a maximum likelihood estimation under a mixture model (Lander and Botstein, 1989) was applied for interval mapping. Analysis revealed two major QTLs: on chromosome 1 and on chromosome 3.
Chromosome 1 QTL is located between 20,384 Mb and 22,181 Mb (Figure S2). It includes the TIR-NLR cluster WRR4 and the CC-NLR cluster RPP7. Six RILs (three resistant and three susceptible) recombine within the QTL and were used for fine mapping. We designed a Single Nucleotide Polymorphism (SNP, 21,195 Mb, Fw: TCAGATTGTAACTGATCTCGAAGG, Rv: CCATCAAGCACACTGTATTCC, amplicon contains two SNPs, Oy-0: A and G, Col-0: G and C) and an Amplified Fragment Length Polymorphism (AFLP, 21,691 Mb, Fw: AAGGCAATCAGATTAAGCAGAA, Rv: GCGGGTTTCCTCAGTTGAAG, Oy-0: 389 bp, Col-0: 399 bp) markers between WRR4 and RPP7. Four lines eliminate RPP7 from the QTL. The only NLR cluster in chromosome 1 QTL is WRR4.
Chromosome 3 QTL is located between 17,283 Mb and 19,628 Mb (Figure S3). It includes the atypical resistance-gene cluster RPW8, the CC-NLR ZAR1 and the paired TIR-NLRs At3g51560-At3g51570. Six RILs (three resistant and three susceptible) recombine within the QTL and were used for fine mapping. We designed an AFLP (18,016 Mb, Fw: gctacgccactgcatttagc, Rv: CCAATTCCGCAACAGCTTTA, Oy-0: 950 bp, Col-0: 1677 bp) and a Cleaved Amplified Polymorphic Sequence (CAPS, 18,535 Mb, Fw: TCAAGCCTGTTAAGAAGAAGAAGG, Rv: GCCCTCCACAAAGATTCTGAAGTA, enzyme: DdeI, Oy-0: uncleaved, Col-0: cleaved) markers between the QTL border and RPW8. We designed a CAPS marker (18,850 Mb, Fw: TCTCGGGGAAAATATGATTAGA, Rv: GGTTGATTTTTATTGTGGTAGTCGT, enzyme: SwaI, Oy-0: cleaved, Col-0: uncleaved) between RPW8 and ZAR1. We designed a SNP (18,937 Mb, Fw: CCACAAGGTCGGAATCTGTAGC, Rv: TGCACAGAAGTAACCCACCAAC, Oy-0: C, Col-0: T) and a CAPS (19,122 Mb, Fw: ACCACCACCTCGATGCATTTC, Rv: CCTTCCCTGCGAAAGACACTC, enzyme: BsrI, Oy-0: uncleaved, Col-0: cleaved) markers between ZAR1 and the TIR-NLR pair. Three recombinants eliminate the TIR-NLR pair, two eliminate ZAR1 and one eliminates RPW8. None of the Col-0 NLR clusters orthologs are present in the QTL. The gene underlying chromosome three resistance is located between the border of the QTL and RPW8.
Bulk segregant analysis and RenSeq
We generated an F2 population from a cross between HR-5 (resistant) and Ws-2 (susceptible). We phenol/chloroform extracted DNA from 200 bulked F2 lines fully susceptible to AcEx1. The bulked DNA sample was prepared as an Illumina library and enriched using the c. The sample was sequenced in a pooled MiSeq run (data available on request). Firstly, reads were aligned with BWA mem (Li and Durbin, 2009) to the and SNPs called with Samtools (Li et al., 2009). The genome was scanned for regions of high linkage with the next generation mapping tools at http://bar.utoronto.ca/ngm/ (Austin et al., 2011). Secondly, the reads were mapped using BWA to the RenSeq PacBio assembly generated for HR-5 (Van de Weyer et al., 2019). Highly linked regions were confirmed visually with the integrated genome viewer (Robinson et al., 2017).
Gene cloning
Genes were cloned with the USER method (NEB) following the manufacturer recommendations, with their natural 5’ and 3’ regulatory sequences into LBJJ233-OD (containing a FAST-Red selectable marker, pre-linearized with PacI and Nt. BbvcI restriction enzymes). For overexpression, genes were cloned into LBJJ234-OD (containing a FAST-Red selectable marker and a 35S / Ocs expression cassette, pre-linearized with PacI and Nt. BbvcI restriction enzymes). Primers, template and vectors are indicated in (Table S3). WRR4ACol-0 has been previously published (Cevik et al., 2019).
All the plasmids were prepared using a QIAPREP SPIN MINIPREP KIT on Escherichia coli DH10B thermo-competent cells selected with appropriate antibiotics. Positive clones (confirmed by size selection on electrophoresis gel and capillary sequencing) were transformed in Arabidopsis thaliana via Agrobacterium tumefaciens strain GV3101. Transgenic seeds were selected under fluorescent microscope for expression of the FAST-Red selectable marker (Shimada et al., 2010).
CRISPR adenine base editor
An sgRNAs targeting WRR4A stop codon in Col-0 (TTCTGAgaagcattcgaaag[nGA]) was assembled by PCR to a sgRNA backbone and 67 bp of the U6-26 terminator. It was then assembled with the AtU6-26 promoter in the Golden Gate compatible level 1 pICH47751. We designed a mutant allele of a plant codon optimized Cas9 with a potato intron (Addgene: 117515) with D10A (nickase mutant) and R1335V/L1111R/D1135V/G1218R/ E1219F/A1322R/T1337R, to change the PAM recognition from NGG to NG (Nishimasu et al., 2018). We assembled this Cas9 (golden gate compatible BpiI: GACA-GCTT) along with a barley codon optimized TadA module (golden gate compatible BpiI: AATG-GCTT) in a level 0 vector pICH41308. It was then assembled with the YAO promoter (Addgene: 117513) and the E9 terminator (Addgene: 117519) in a level 1 vector pICH47811 (with expression in reverse orientation compared to the other level 1 modules). It was then assembled with a FAST-Red selectable marker (Addgene: 117499) and the sgRNA level 1 cassette into a level 2 vector pICSL4723, using the end-linker pICH41766. Level 0 vector was cloned using BpiI enzyme and spectinomycin resistance. Level 1 vectors were cloned using BsaI enzyme and carbenicillin resistance. Level 2 vectors were cloned using BpiI enzyme and kanamycin resistance. It was expressed via Agrobacterium tumefaciens strain GV3101 in Arabidopsis Oy-0. In the first generation after transformation, we did not detect any mutant from 24 independent transformants. It indicates an absence of activity of the construct. It can be explained by the Cas9 mutations that were not tested before on this specific allele nor in combination with TadA.
Transient expression in N. tabacum leaves
A. tumefaciens strains were streaked on selective media and incubated at 28 °C for 24 hours. A single colony was transferred to liquid LB medium with appropriate antibiotic and incubated at 28 °C for 24 hours in a shaking incubator (200 rotations per minute). The resulting culture was centrifuged at 3000 rotations per minute for 5 minutes and resuspended in infiltration buffer (10 mM MgCl2, 10 mM MES, pH 5.6) at OD600 = 0.4 (2 ×108 cfu/ml). For co-expression, each bacterial suspension was adjusted to OD600 = 0.4 for infiltration. The abaxial surface of 4-weeks old N. tabacum were infiltrated with 1 ml needle-less syringe. Cell death was monitored three days after infiltration.
Resolution of WRR4AOy-0cDNA sequence
RNA was extracted from three Col-0_rpw8 and three Col-0 WT individual plants using the RNeasy Plant Mini Kit (QIAgen) and treated with RNase-Free DNase Set (QIAGEN). Reverse transcription was carried out using the SuperScript IV Reverse Transcriptase (ThermoFisher). PCR was conducted using Fw: TCTGATGTCCGCAACCAAAC (in the first exon) and Rv: GTCCTCTTCGGCCATATCTTC (in the last exon) with the Taq Polymerase enzyme (NEB) following the manufacturer protocol. The 2848 nt amplicon sequence, corresponding to the cDNA sequence (i.e. with already spliced introns) was resolved by capillary sequencing. It indicates that the splicing sites are identical between WRR4AOy-0 and the splicing sites reported in the database TAIR10 for WRR4ACol-0.
Data availability statement
All relevant data can be found within the manuscript and its supporting materials. The sequences of the genomic clones of WRR4AOy-0 and WRR4AHR-5 are deposited at NCBI GenBank as MW533532 and MW533533 respectively.
Author contributions
BC, SF, OF, AR, VC, EH and JJ designed research; BC, SF, OF, AR, SW, VC performed research; BC, SF, OF, AR, VC, EH and JJ analysed data; and BC and JJ wrote the paper. All authors read and approved the final manuscript.
Conflicts of interest statement
The authors declare that they have no conflict of interests
Supporting materials
Figure S1: Detailed map of candidate loci in Oy-0
Figure S2: Fine mapping of chromosome 1 QTL
Figure S3: Fine mapping of chromosome 3 QTL
Figure S4: Detailed map of candidate loci in HR-5
Figure S5: Some AcEx1 CCG candidate effectors are recognised by WRR4AOy-0 Figure S6: Phylogeny of WRR4A based on protein sequences
Table S1: Arabidopsis v1 RenSeq bait library (Arbor Bioscience, MI, USA) as described by (Jupe et al., 2013)
Table S2: Phenotype of 283 Arabidopsis accessions in response to AcEx1 infection Table S3: Primers used in this study
Acknowledgements
We thank the Gatsby Foundation (UK) for funding to the Jones lab. We thank Mark Youles in TSL Synbio for his excellent support with Golden Gate cloning and for providing modules. This research was supported in part by the NBI Computing infrastructure for Science (CiS) group and Dan MacLean’s group by providing computational infrastructure. B.C., S.F., O.F. and V.C. were supported by Biotechnology and Biological Sciences Research Council (BBSRC) grant BB/L011646/1. A.R. was supported by EMBO LTF (ALTF-842-2015). B.C, S.W and J.D.G.J. were supported in part by ERC Advanced Investigator grant to JDGJ ‘ImmunityByPairDesign’ Project ID 669926.