Genome Re-Sequencing and Functional Analysis Places the Phytophthora sojae Avirulence Genes Avr1c and Avr1a in a Tandem Repeat at a Single Locus

The aim of this work was to map and identify the Phytophthora sojae Avr1c gene. Progeny from a cross of P. sojae strains ACR10×P7076 were tested for virulence on plants carrying Rps1c. Results indicate that avirulence segregates as a dominant trait. We mapped the Avr1c locus by performing whole genome re-sequencing of composite libraries created from pooled samples. Sequence reads from avirulent (Pool1) and virulent (Pool2) samples were aligned to the reference genome and single nucleotide polymorphisms (SNP) were identified for each pool. High quality SNPs were filtered to select for positions where SNP frequency was close to expected values for each pool. Only three SNP positions fit all requirements, and these occurred in close proximity. Additional DNA markers were developed and scored in the F2 progeny, producing a fine genetic map that places Avr1c within the Avr1a gene cluster. Transient expression of Avr1c or Avr1a triggers cell death on Rps1c plants, but Avr1c does not trigger cell death on Rps1a plants. Sequence comparisons show that the RXLR effector genes Avr1c and Avr1a are closely related paralogs. Gain of virulence on Rps1c in P. sojae strain P7076 is achieved by gene deletion, but in most other strains this is accomplished by gene silencing. This work provides practical tools for crop breeding and diagnostics, as the Rps1c gene is widely deployed in commercial soybean cultivars.


Introduction
Plants are challenged by a diversity of pathogens and rely on pre-existing and elaborated defense mechanisms including innate immunity.Pathogens have developed effector arsenals to successfully colonize plants and overcome plant defenses.Host plants detect microbe-associated molecular patterns (MAMPs) and pathogen effector molecules by use of immune receptors positioned on the outer plasma membrane or inside the cell cytoplasm [1].The activation of pattern-triggered immunity (PTI) or effector-triggered immunity (ETI) leads to defense activation and pathogen containment.In many cases host immune receptors and pathogen effector proteins underlie resistance (R) and avirulence (Avr) gene interactions described by plant pathologists.
Soybean resistance to the root rot pathogen Phytophthora sojae is determined in part by the presence of Rps genes.P. sojae effectors that are recognized by Rps proteins and cause ETI are Avr factors.
As key elements that determine disease outcome on soybean, the Avr factors of P. sojae have been a focus of study and targeted for identification [2].Several P. sojae Avr genes have been identified to date including Avr1a [3], Avr1b [4], Avr1d [5], Avr1k [6], Avr3a/5 [3,7], Avr3b [8], Avr3c [9], and Avr4/6 [10].Each Avr gene is predicted to encode a small secreted protein containing a signal peptide followed by an RXLR (Arg-any amino acid-Leu-Arg) sequence.This conserved motif is proposed to have a role in the delivery of the effector protein to the host cell cytoplasm [11,12].Additional conserved features in the C-terminal or effector domains of the proteins are suggested to be important functional elements [13,14].
The identification of P. sojae Avr genes has relied on map-based approaches.Segregating F 2 populations arising from outcrosses of parental strains that differ in virulence on particular Rps genes have provided the basis for investigations into Avr determinants.This core strategy has been assisted by strain-specific associative mapping, whole genome sequence information, transcript profiling, and the annotation of some 400 predicted P. sojae RXLR effectors known as Avh genes [15][16][17].Selecting candidate Avh genes a priori can accelerate mapping and identification but this tactic is not fail-safe.Bulked segregant analysis (BSA) or pooled sampling is a powerful method that aids de novo mapping by providing genetic markers linked to gene targets [18,19].Recently, pooling has been combined with deep sequencing technologies to rapidly map and identify target regions in a variety of organisms [20][21][22].
The objective of the present study was to identify the P. sojae Avr1c locus.A genetic cross of P. sojae strains ACR106P7076, which differ in virulence on Rps1c, was employed to track the segregation of Avr1c in F 1 and F 2 progeny.Candidate Avh genes were selected for mapping but this was not successful, as none of the candidates co-segregated with Avr1c.We resorted to mapping by performing BSA and single nucleotide polymorphism (SNP) analysis of whole genome re-sequencing data.This led to the finding that Avr1c is part of the Avr1a gene cluster.The Avr1c gene encodes a predicted RXLR effector protein of 126 amino acids.The Avr1a and Avr1c genes are close paralogs that are nearly identical in sequence at their N-terminal regions but that have diverged substantially in their C-terminal effector domains.

Results
Avirulence towards Rps1c is a dominant trait in P. sojae cross ACR106P7076 To determine how virulence towards Rps1c is inherited in P. sojae, we analyzed progeny from outcrosses between strains.The P. sojae strains ACR10 and P7076 differ in virulence towards soybean plants carrying the Rps1c gene; ACR10 is avirulent while P7076 is virulent.Both strains are virulent towards control plants that lack Rps1c, such as cv Williams (no known Rps genes) or Harosoy (Rps7).Outcrosses of ACR106 P7076 were performed and the resulting F 1 and F 2 progeny were tested for virulence towards Rps1c.From a total of 25 F 1 progeny, an avirulent:virulent phenotypic ratio of 23:2 was observed.Testing of 40 F 2 progeny derived from a single F 1 individual resulted in consistent readouts for 28 of the progeny, whereas 12 F 2 progeny could not be reliably scored for virulence on Rps1c, or had lost virulence toward control plants lacking the Rps1c gene.For the 28 F 2 progeny with reliable phenotypes and that retained virulence towards the control plants, the avirulent:virulent ratio towards Rps1c was 22:6 (Chi-squared probability = 0.66 for 3:1).Overall, the results suggest that avirulence towards Rps1c segregates as a dominant trait in outcross between P. sojae strains ACR106P7076.

Selected candidate Avh genes do not co-segregate with Avr1c
To accelerate the identification of Avr genes, pre-selecting candidate genes has proven a successful tactic in many cases [5,8,9,[23][24][25].Briefly, candidates are chosen based upon sequence characteristics, expression and polymorphism data and tested for co-segregation or association with the virulence trait, or directly screened for functional interaction with selected R-genes.We selected a total of 24 Avh genes as candidates for Avr1c, based upon comparison of Avh gene sequences in the reference P. sojae strain P6497 and re-sequenced strains P7064, P7074 and P7076, as shown in Supplementary Table S1.Strains P6497, P7064, and P7074 are avirulent, while strain P7076 is virulent on Rps1c.Cleaved amplified polymorphic (CAP) markers were designed to score each of the 24 candidate Avh genes in the ACR106P7076 F 2 progeny.Unfortunately, none of the 24 candidate Avh genes tested co-segregated with Avr1c, so this approach was not successful.

Deep sequencing of progeny pools identifies SNPs linked to Avr1c
The de novo mapping of Avr1c was accomplished using nextgeneration sequencing and BSA applied to two pools of F 2 progeny [18,20,21].Equal amounts of DNA from six avirulent (Pool1) or six virulent (Pool2) F 2 individuals were combined to construct two composite genomic libraries for deep sequencing and SNP analysis.In excess of 150 million sequence reads were obtained from each of the pools, as detailed in Table 1 and Figure 1.Reads were aligned to the reference genome P6497 and SNPs called from the resulting alignment map.Mapping SNPs linked to Avr1c proceeded under the following assumptions.Based on the dominant phenotype observed for Avr1c, the avirulent pool (Pool1) was expected to contain both alleles of Avr1c and therefore consist of a mix of homozygous dominant and heterozygous genotypes in a ratio of 1:2.The expected non-reference base allele frequency for a causal SNP in this pool should then be the average of one homozygous dominant allele (0) and two heterozygous alleles (0.5) for an expected frequency of 0.33.However, the virulent pool (Pool2) should consist of exclusively homozygous recessive genotypes with only the virulent allele of Avr1c represented and thus possess a nonreference base allele frequency of 1.0 for the causal SNP.These numbers may vary slightly due to effects in sampling, sequencing errors or mapping accuracy.However, both the causal SNP for virulence as well as nearby SNPs genetically linked to the loci should exhibit allele frequencies very close to these expectations.Conversely, for SNPs linked to the avirulent allele, the reciprocal values of 0.67 (Pool1) and 0.0 (Pool2) would be expected.Therefore, our causal SNP and those linked to it should be identifiable by selecting for SNPs that exhibit the expected allele frequency in each respective pool.
With this is mind, a sequence alignment map (SAM) was generated by aligning reads from each sequenced pool to the P6497 reference.The SNPs were then called using Samtools and filtered for quality (Phred quality .30).A total of 1,107 (Pool1) and 434 (Pool2) high quality SNPs were obtained and filtered using a window of acceptable allele frequencies based on the expected values for each pool.A window of acceptable nonreference allele frequency values ranging from 0.18 to 0.48 was applied to Pool1 SNPs, and a window of 0.9 to 1.0 was used to select for Pool2 SNPs.The reciprocal values were also used in order to capture SNPs linked to either allele.By this criteria, 547 (Pool1) and 204 (Pool2) SNP positions were found, but only three SNPs exhibited allele frequencies that fit both pool expectations, as shown in Figure 1A.Remarkably, as would be expected from genetic linkage, all three of the identified SNPs occur in close physical proximity within a 92 kb segment on Scaffold_7 in the P. sojae reference genome assembly (v5.0) (Figure 1B).

Avr1c is located in the Avr1a gene cluster
The SNPs identified in our mapping occurred in close proximity in a region rich in Avh genes, including the previously identified Avr1a locus, as shown in Figure 2A.In order to verify that Avr1c occurs in this genome region and to more accurately map its position, a set of DNA markers was developed based upon polymorphic sites identified by comparison of the P. sojae reference genome and sequence reads from each of the pools.Useful markers were also available from previous mapping work on Avr1a [3,26].Each marker was scored in a mapping population of 28 F 2 progeny.The results indicate that Avr1c maps precisely to the Avr1a locus (Figure 2B).

The predicted RXLR effector gene Avh275c corresponds to Avr1c
Previous work on Avr1a indicated that this locus is highly polymorphic and displays copy number variation among P. sojae strains [3].The reference strain P6497 contains a tandem array of four related 5.2 kb segments, as shown in Figure 2C.Two of these segments include copies of Avr1a while adjacent segments include paralogous genes named Avh275c and Avh72.The Avh72 open reading frame includes a frame-shift and no transcripts can be detected for this gene, indicating that Avh72 is a pseudogene.However, the Avh275c open reading frame is intact and transcripts can be detected for this gene in several P. sojae strains.The Avh275c gene encodes a predicted RXLR effector protein of 126 amino acids.Results described below indicate that Avh275c corresponds to Avr1c, thus for simplicity, we will hereafter refer to the Avh275c gene as Avr1c.
To investigate the structure of the Avr1c locus among P. sojae strains, genomic DNA was digested with the restriction enzyme PmlI, blotted, and hybridized with a probe that detects Avh72, Avr1c and Avr1a.The PmlI digestion is known to resolve each of the four copies in the tandem array, including Avh72, Avr1c, and the two copies of Avr1a, as we have previously demonstrated [3].The results shown in Figure 3 indicate that P. sojae strain P7076 lacks a copy of the Avr1c gene, but that all other strains tested possess a copy of Avr1c.Additionally, data from re-sequencing of the parental strains ACR10 and P7076 was used to measure read coverage across the Avr1c segment, as constructed for the reference strain P6497 [3].As is shown in Figure 4, there is no coverage in the Avr1c region in virulent strain P7076, while this region is covered by sequence reads in the avirulent strain ACR10.Thus, the DNA blot and the sequence coverage analyses indicate that Avr1c is deleted from P. sojae strain P7076 but that copies of Avh72 and Avr1a remain in this tandem gene cluster.
Sequence analysis revealed three different Avr1c alleles within our P. sojae strain collection (Figure S1).These include the Avr1c-1 allele (represented by strain P6497; also present in strain ACR10), the Avr1c-2 allele (represented by strain P7064) and the Avr1c-3 allele (strain P7074).The Avr1a DNA sequence encoding the open reading frame for the protein is not polymorphic among known strains of P. sojae that carry copies of the gene.Comparison of the predicted amino acid sequences of Avr1a and each of the Avr1c alleles is shown in Figure 5.The N-terminal portions of the Avr1a and Avr1c predicted proteins are nearly identical, while the Cterminal regions have diverged substantially.Among the Avr1c alleles, Avr1c P7064 displays four amino acid changes (S113G, K114R, I115L,V120G) and Avr1c P7074 displays five changes (L69Q, S82P, S113G, K114R, I115L), compared to the reference allele Avr1c P6497 .
To test for the presence of Avr1c mRNA transcripts among P. sojae strains, we performed reverse transcriptase polymerase chain reaction (RT-PCR).The Avr1a transcript was also measured for comparison since these two genes occur adjacent to each other.Results show that transcripts of Avr1c and Avr1a are present in strains that are avirulent towards Rps1c and Rps1a, respectively, but not in the corresponding virulent strains.One exception is strain ACR9, which expresses the Avr1c transcript but nonetheless is virulent on Rps1c (Figure 6).

Transient expression of Avr1c triggers cell death in Rps1c plants
Transient expression of an Avr effector protein in a soybean plant containing a matching Rps gene triggers cell death.In order

Discussion
The soybean Rps1c gene is extensively deployed for the management of root and stem rot disease caused by P. sojae.Based on recent variety trial data shown in Table 2, Rps1c is the most prevalent source of resistance to P. sojae in the province of Ontario, where over 1 million hectares of land is annually dedicated to soybean production.The Rps1c gene is also widely deployed in the USA and other major soybean growing regions.Thus, identification of the Avr1c gene in P. sojae will have practical uses in soybean breeding and pathogen diagnostics.
Genomic resources and annotation of RXLR effector repertoires for oomycete plant pathogens have facilitated Avr gene identification in several species.Large scale effector screens have proven to be successful for identifying or predicting Avr   determinants in Phytophthora infestans [24,28] and Bremia lactucae [29], but this method is less useful in P. sojae because of the difficulty in performing Agrobacterium mediated infiltration and transient expression in soybean.However, being a homothallic organism P. sojae offers certain advantages over predominately heterothallic species such as P. infestans and B. lactucae.For example, creating F 2 populations for genetic mapping is generally possible for homothallic but not heterothallic species.In fact, identification of all known P. sojae Avr genes has relied on mapping in segregating F 2 populations.Most recently, this strategy has been accelerated by pre-selection of candidate RXLR effectors [5,8,9].The failure of the candidate gene approach in the present study was primarily a result of an incomplete RXLR inventory.To select candidates based on strain-specific polymorphisms, we relied on previous work that described a comparative genome-wide analysis of Avh genes among P. sojae strains [15].Unfortunately, Avh275c was not included in the gene set because of poor annotation of tandem arrays in the reference genome.In addition to Avh275c, we also should have included known P. sojae Avr genes (or closely linked markers) in our initial screen, as this would have revealed co-segregation of Avr1c and Avr1a.
Of course, it is easier to select candidates in hindsight, and any approach of choosing genes a priori will always carry a risk because it involves uncertainties.In contrast, de novo mapping of genes represents an unbiased and more robust method for target identification but can involve a greater research investment.Deep sequencing technologies combined with bioinformatics offer powerful tools for gene mapping.Previously, BSA has been combined with amplified fragment length polymorphism (AFLP) [26], and microarray analyses [3] for mapping Avr genes in P. sojae.Our present scheme of combining BSA with whole genome resequencing is a natural step forward, especially since this technique has proven successful in other organisms [20,22].With a high quality reference genome assembly, the path from finding linked markers to identifying target genes can proceed rapidly, as we have shown in this study.
Our conclusion that avirulence towards Rps1c is inherited as a dominant trait in crosses between P. sojae strains ACR106P7076 is consistent with the segregation results, despite the deviations that were observed.Previous inheritance studies have suggested that Avr1c is dominant [30] or recessive [31,32], but the strains used in past studies were different from those used in the present analysis.We show that the Avr1c gene is deleted from the Rps1c-virulent strain, P7076, which is consistent with avirulence being dominant in the cross ACR106P7076.Our finding that many P. sojae strains carry a copy of Avr1c, yet do not express the gene, indicates that gain of virulence on Rps1c can also result from gene silencing.This could also explain past inheritance results that showed virulence towards Rps1c to be dominant in F 1 hybrids [32], since it is known that gene silencing in P. sojae can cause transgenerational epigenetic inheritance that results in gain of virulence [33,34].In fact, comparison of results of RT-PCR analyses suggests transcriptional switching of gene expression/silencing of Avr1c and Avr1a in clonally propagated P. sojae cultures, because we noted some differences in expression between a previous study [3] and the present one, for strains ACR6, ACR9, ACR16, and ACR20.Variation in virulence of clonal P. sojae and P. infestans cultures is well known [35,36].The alternate loss and recovery of virulence in  successive clonal generations has puzzled investigators [35], but our results indicate that epiallelic switching of effector gene expression is a mechanism that could account for this phenomenon.
Besides gene deletion and gene silencing, our results suggest that P. sojae can escape Rps1c-mediated immunity by other means.The transient expression assay indicates that Avr1c P6497 , Avr1c P7064 , and Avr1a are recognized by Rps1c plants but that Avr1c P7074 is not.There are five amino acid polymorphisms between Avr1c P6497 and Avr1c P7074 , but three of these (S113G, K114R, I115L) are shared with Avr1c P7064 , and one (L69Q) with Avr1a.Thus, we hypothesize that the critical change to Avr1c P7074 that enables it to escape Rps1c is the S82P mutation.Introduction of a proline residue represents a severe change with potential to alter the secondary structure in a region that is predicted to be alphahelical, providing a rationale for the importance of the S82P mutation.However, P. sojae strain P7074, which carries and expresses the Avr1c P7074 allele, is avirulent towards Rps1c soybean plants.We propose that expression of Avr1a by strain P7074 triggers immunity on Rps1c, because our results show that Rps1c can detect Avr1a in addition to Avr1c P6497 and Avr1c P7064 .We predict that a P. sojae strain expressing the Avr1c P7074 allele but lacking Avr1a transcripts will be virulent towards Rps1c, but this remains to be demonstrated.
Yet another mechanism to evade Rps1c immunity must be invoked to account for the finding that strain ACR9 is virulent on Rps1c plants.Results shown here indicate that ACR9 expresses transcripts corresponding to the Avr1c P6497 allele, which should trigger immunity on Rps1c based on the transient expression results, yet ACR9 is virulent on Rps1c plants.Each of these tests, mRNA analysis and phenotypic scoring, was replicated several times during the course of the present study with consistent results.Additionally, plants inoculated with strain ACR9 tested positive for Avr1c expression by RT-PCR (not shown).Given the results, how do we explain the virulence of ACR9 on Rps1c plants?A possible explanation is that there are additional effectors in strain ACR9 that suppress ETI caused by the Avr1c-Rps1c interaction.It is known that particular RXLR effectors can suppress immune responses triggered by other RXLR effectors, and that potential Avr-R gene interactions can be masked by these apparent epistatic effects [15,[37][38][39][40][41].
Epistatic interactions could also cause unusual inheritance behaviour, especially if combined with gene-conversion or mitotic crossing over events that commonly occur in hybrids from outcrosses of P. sojae [42] and other species of Phytophthora [43].Thus, it is possible that the phenotypic penetrance or inheritance of Avr1c in particular hybrid progeny may be influenced by epistatic loci, gene conversion, and mitotic crossing over.This is in addition to the epigenetic affects discussed earlier.
Our past work on Avr1a demonstrated that gain of virulence on Rps1a is caused by gene deletion or gene silencing [3].In contrast to Avr1c, no sequence polymorphisms within the Avr1a open reading frame were detected among the P. sojae strains that carry copies of Avr1a.It is also evident that Avr1c and Avr1a are individually and collectively dispensable, as are most known Avr genes in P. sojae.
The transient expression assays suggest that Rps1c can recognize Avr1c and Avr1a.This conclusion will require further work to verify, such as by performing transformation of P. sojae to demonstrate that avirulence to Rps1a and Rps1c can be acquired by ectopic expression of Avr1a.Another example of an apparent dual specificity of a soybean Rps gene was reported recently, specifically that Avr1b and Avr1k are both capable of triggering Rps1k mediated immunity [6].However, the caveat is that these suggestions are based upon genetic isolines that have been developed for the various Rps genes.Soybean Rps genes remain poorly characterized at the molecular level but it is known that the Rps1 locus consists of a large cluster of nucleotide-binding, leucinerich repeat type of immune receptor genes [44,45].Thus, there are alternative possibilities to account for the apparent dual specificity of Rps1k and Rps1c, such as separate but closely linked R-genes that recognize each of the Avr factors individually.Indeed, the variety trial data presented in Table 2 suggests the existence of a soybean line that possess both Rps1k and Rps1c, an occurrence that would be unlikely if these two genes are truly allelic.
A model for the structure and expression of the Avr1a/1c locus in different strains of P. sojae is presented in Figure 10.We acknowledge that our results present uncertainties with regard to this model, but we feel that sufficient evidence supports our conclusion that Avr1c occurs in the Avr1a gene cluster and corresponds to the RXLR effector Avh275c.These findings have practical importance in soybean breeding and disease management, especially since Rps1c is among the most widely deployed sources of P. sojae resistance presently in use.

Plant materials, Phytophthora sojae strains and virulence scoring
Soybean (Glycine max) cultivar Williams (rps) and the corresponding isolines L75-3735 (Rps1c), L75-6141 (Rps1a) from the collection at Agriculture and Agri-Food Canada were used to evaluate the virulence of pathogen cultures and for transient expression by particle bombardment.The cultivar Harosoy (Rps7) and the corresponding isolines Haro14 (Rps1c) Harosoy 63 (Rps1a) were also used to replicate the experiments in a separate genetic background.For virulence assays, 10 soybean seeds were sown in 10 cm pots containing soil-less mix (Pro-Mix 'BX', Premier Horticulture Ltd, Riviere-du-Loup, Canada).A total of three pots per isolate were tested for each replicate, and a minimum of three independent biological replicates were performed for each P. sojae culture tested.Pots were watered with 3 mg/L 20-20-20 (N-P-K) fertilizer.Plants were grown in a controlled growth chamber with supplemental light (16 h photoperiod with 25uC day and 16uC night temperatures) for one week before inoculations.
Phytophthora sojae isolates were from the collection at Agriculture and Agri-Food Canada, London, ON.The origin of each strain has been described [3,7].Performing sexual crosses of P. sojae strains ACR106P7076 and the generation of F 1 and F 2 progeny has been described [5,33].All of the isolates were routinely maintained on 2.5% (v/v) vegetable (V8) juice medium at 25uC in the dark.The P. sojae cultures were transferred to 0.9% (v/v) V8 agar plates for 5 to 7 days prior to green plant inoculations.Methods of hypocotyl inoculation and virulence scoring have been described [3].

Nucleic acid isolation, RT-PCR and DNA blot hybridization
The purification of RNA from P. sojae mycelium and transcript analysis by reverse transcriptase PCR (RT-PCR) was carried out as previously described [46].Total RNA was treated with DNAse I (Invitrogen).The RT-PCR was conducted on 1 mg RNA using a reverse transcriptase (SuperScriptIII, Invitrogen) system according to the manufacturer's instructions.Genomic DNA from P.sojae mycelium was isolated by phenol chloroform extraction and ethanol precipitation.Genomic PCR amplifications were performed using 15 ng DNA, 0.5 mM primers, 0.25 mM dNTPs, Taq polymerase and supplied buffer.The following PCR program was used: 94uC for 2 min, 40 cycles of 94uC for 40 s, 58uC for 40 s, 72uC for 30 to 60 s (varied with the PCR product size) and a final extension of 72uC for 10 min.The P. sojae actin gene was used as a control.
Genomic DNA isolation from P. sojae mycelial cultures, digestion with the restriction enzyme PmlI, electrophoretic separation, blotting and hybridization was performed according to standard protocols [47].For hybridization, a DIG-labeled probe corresponding to the full length Avr1a open reading frame was prepared.This probe hybridizes to Avr1a, Avr1c, and Avh72.

Transient expression assays
Soybean plants were grown in the growth chambers as described above.Leaves from 14 d old plants were harvested for co-bombardment.Primers used for cloning Avr1c segments into the transient expression vector pFF19 are provided in Supplementary Table S2.The three different Avr1c open reading frames (Avh275c P6497 , Avh275c P7064 and Avh275c P7074 ) and the Avr1a open reading frame were amplified using specific primers and cloned into the 35S promoter-derived plant expression vector pFF19 using BamHI/PstI restriction sites.The signal peptides were omitted from all constructs.Soybean leaves were transformed by co-bombardment with plasmids encoding a glucuronidase (GUS) reporter gene and a test construct, using tungsten beads.Cobombardment assays were performed as previously described [48].For the control, the test gene corresponded to a frame-shift mutant of Avr1a.Leaves were washed in 70% ethanol and photographed using a digital camera.Sample images were obtained and processed, and the volume of GUS-positive spots was calculated as the intensity (measured in absorbance units) multiplied by the area (mm 2 ) [5].

Figure 1 .
Figure 1.Identification of SNPs linked to Avr1c by bulked segregant analysis and deep sequencing.A, The procedure for discovery of candidate SNPs linked to Avr1c is shown.Selected F 2 progeny from a cross of P. sojae strains ACR106P7076 were pooled according to their virulence phenotype.The avirulent (A) Pool1 and virulent (V) Pool2 composite DNA samples were deeply sequenced.Sequence reads were aligned against the reference genome, and SNPs were identified and filtered based upon quality scores.High quality SNPs were further filtered according to the predicted SNP frequencies for Pool1 and Pool2.After processing, only three SNPs passed all requirements.B, Genome location of three candidate SNPs.These three candidate SNPs occur in close physical proximity in the reference genome assembly.All three sites fall within a 92 kb segment on Scaffold_7 (V5.0).Reference (Ref) allele and alternate (Alt) alleles for three SNPs are shown.The SNP frequencies (Freq) in each pool are also shown.doi:10.1371/journal.pone.0089738.g001

Figure 2 .Figure 3 .
Figure 2. Genetic and physical mapping of Avr1c region.A, Physical map of Avr1c region.The position of the three identified SNP markers, Avr1a, and 25 predicted Avh genes are shown.B, Genetic analysis of the Avr1c region.The position of DNA markers and their recombination frequency (RF%) with Avr1c in a segregating F 2 population (n = 28) is indicated; Mbp, mega base pair.C, Predicted arrangement of the Avr1a locus in P. sojae reference strain P6497.doi:10.1371/journal.pone.0089738.g002

Figure 4 .Figure 5 .
Figure 4. Sequence read coverage of the Avr1a/Avr1c region from re-sequencing of P. sojae strains ACR10 and P7076.An assembly of the Avr1a region from reference strain P6497 was used to align sequence reads from parental strains ACR10 and P7076.The DNA segment corresponding to the Avr1c gene lacks sequence reads in parental strain P7076.The longest segment of identity between the Avr1a and Avr1c DNA sequences is 98 nucleotides, which is shorter than the read length of 100 nucleotides.doi:10.1371/journal.pone.0089738.g004

Figure 6 .
Figure 6.Analysis of alleles of Avr1c, and transcripts of Avr1c and Avr1a in P. sojae strains.Shown at the top of the figure are the Avr1c alleles present in each of the selected P. sojae strains.Virulence phenotype of the P. sojae strains on Rps1c or Rps1a plants is shown as virulent (V) or avirulent (A).Reverse transcriptase polymerase chain reaction (RT-PCR) analysis was performed using mRNA from mycelia cultures, and Avr1a and Avr1c specific primers, to test for transcripts of these two genes.Results from RT-PCR of the control gene Actin are also shown.doi:10.1371/journal.pone.0089738.g006

Figure 7 .
Figure 7. Transient expression of Avr1c and Avr1a triggers cell death in Rps1c soybean plants.Measurement of cell death in soybean leaves by co-bombardment and transient expression of a GUS reporter together with a test gene.Test genes that cause cell death reduce GUS expression and blue staining.A, Leaves of Williams (rps) and the isoline L75-3735 (Rps1c) tested with each of the three Avr1c alleles, and Avr1a.B, Leaves of Williams (rps) and the isoline L75-6141 (Rps1a) tested with each of the three Avr1c alleles.Control test gene in each experiment corresponds to a synthetic Avr1a sequence with a frame-shift mutation.Results show means and standard errors of three independent biological replicates, with a minimum of three leaves per treatment, per replicate.doi:10.1371/journal.pone.0089738.g007

Table 1 .
Whole genome re-sequencing of composite and strain specific P. sojae libraries.

Table 2 .
Rps1c is the most prevalent known source of P. sojae resistance in soybean lines entered in Ontario variety trials 1 .

Table S1 A
list of 24 candidate Avh genes tested for co-segregation with Avr1c in the P. sojae cross ACR106P7076.(PDF)