The Phytophthora sojae Avirulence Locus Avr3c Encodes a Multi-Copy RXLR Effector with Sequence Polymorphisms among Pathogen Strains

Root and stem rot disease of soybean is caused by the oomycete Phytophthora sojae. The avirulence (Avr) genes of P. sojae control race-cultivar compatibility. In this study, we identify the P. sojae Avr3c gene and show that it encodes a predicted RXLR effector protein of 220 amino acids. Sequence and transcriptional data were compared for predicted RXLR effectors occurring in the vicinity of Avr4/6, as genetic linkage of Avr3c and Avr4/6 was previously suggested. Mapping of DNA markers in a F2 population was performed to determine whether selected RXLR effector genes co-segregate with the Avr3c phenotype. The results pointed to one RXLR candidate gene as likely to encode Avr3c. This was verified by testing selected genes by a co-bombardment assay on soybean plants with Rps3c, thus demonstrating functionality and confirming the identity of Avr3c. The Avr3c gene together with eight other predicted genes are part of a repetitive segment of 33.7 kb. Three near-identical copies of this segment occur in a tandem array. In P. sojae strain P6497, two identical copies of Avr3c occur within the repeated segments whereas the third copy of this RXLR effector has diverged in sequence. The Avr3c gene is expressed during the early stages of infection in all P. sojae strains examined. Virulent alleles of Avr3c that differ in amino acid sequence were identified in other strains of P. sojae. Gain of virulence was acquired through mutation and subsequent sequence exchanges between the two copies of Avr3c. The results illustrate the importance of segmental duplications and RXLR effector evolution in the control of race-cultivar compatibility in the P. sojae and soybean interaction.


Introduction
In the interaction between plant pathogens and their hosts, discrete genes control race-cultivar compatibility and disease outcome. Disease resistance resulting from the interplay of pathogen avirulence (Avr) genes and host resistance (R) genes was first described in the gene-for-gene hypothesis [1]. This type of plant disease resistance is now called effector triggered immunity (ETI) [2]. Typically, plant R-genes are part of the innate immune system and encode surveillance proteins that detect pathogen-specific molecules or alterations and thereby activate ETI [3]. Likewise, pathogen effector proteins can be broadly described as secreted proteins that suppress plant defenses and promote disease [4,5].
The identification of the first Avr genes from oomycetes, together with whole genome sequencing projects, revealed a special class of secreted effector proteins that are delivered into host cells [6,7,8,9,10,11]. These proteins contain a second targeting motif, downstream from the signal peptide, with the consensus sequence RXLR (Arg-X-Leu-Arg). The RXLR and associated dEER (Asp-Glu-Glu-Arg; with the leading Asp being variable) motifs somehow traffic the pathogen effector protein across the host plasma membrane [12,13]. The RXLR effectors constitute large super-families of rapidly evolving proteins in all oomycete genomes sequenced to date [14,15]. Predicted RXLR effector genes in P. sojae have been named Avirulence homologues, or Avh genes.
The central importance of RXLR effectors in determining the outcome of oomycete-plant interactions is becoming increasingly evident. Oomycete Avr genes identified and shown to encode RXLR effectors include: Avr1b-1, Avr1a and Avr3a from P. sojae [6,16], Avr3a, Avr4, and Avr-blb1 from Phytophthora infestans [9,17,18]; and ATR1 and ATR13 from Hyaloperonospora arabidopsis [7,8]. These findings demonstrate that the RXLR effector family is at the forefront in evolution and adaptation of oomycete plant pathogens towards their hosts. The genes encoding RXLR effectors are subject to unstable selective pressures that shape this rapidly evolving and highly diverse gene family [14,15,16]. Once an RXLR effector comes under R-gene mediated surveillance and causes ETI, it becomes an Avr gene. Evasion of ETI may be accomplished by a variety of mechanisms including amino acid changes, protein truncations, gene deletions, or by transcriptional silencing of the Avr gene. Thus, the massive redundancy and the high level of intra-and inter-specific variation of RXLR genes provide a diverse reservoir of effector capability to meet the ever changing selective pressures imposed by plant immune systems.
Soybean R-genes that control ETI to P. sojae are known as Rps (Resistance to P. sojae) genes [19]. The aim of the present study was to identify the Avr3c gene from P. sojae and to determine how this varies among strains with differential virulence on Rps3c. Candidate genes for Avr3c were selected based upon genetic mapping data, genome sequence assemblies, and the predicted RXLR secretome of P. sojae. Functional characterization of the Avr3c gene shows this to correspond to an RXLR effector that varies in amino acid sequence among P. sojae strains.

Results
Gene candidates for Avr3c are chosen from predicted RXLR effectors near Avr4/6 The completion of the P. sojae P6497 genome sequence [10], together with the mapping and recent identification of Avr4/6 [20,21] provided an opportunity to identify Avr3c. Previous genetic mapping work placed Avr3c and Avr4/6 on a single linkage group, separated by a distance 16 cM [22]. The overall average genetic to physical distance in P. sojae is approximately 35 kb/cM, but the Avr4/6 region has a high recombination frequency, estimated at 3 kb/cM [20,23]. Thus, a 16 cM distance may represent a physical range of 48 to 560 kb, depending upon whether this is calculated using the recombination frequency determined for Avr4/6 or the genome average. By examining the space around Avr4/6, we determined there are at least four additional RXLR effector genes predicted to occur within a 551 kb interval encompassing the gene, as shown in Figure 1A. Thus, Avh26, Avh27, Avh28, and Avh432 are predicted RXLR effector genes that lie in the vicinity of Avr4/6 and represent good candidates for Avr3c.
Sequence and transcript analysis of each of the candidate genes was performed on P. sojae strains P7064 and P7074, to detect any polymorphisms compared to the sequences from P. sojae P6497. These P. sojae strains differ in virulence characteristics on soybean plants carrying Rps3c, P. sojae P6497 is avirulent while P7064 and P7074 are virulent. The results are summarized in Table 1. This analysis showed that the Avh28 and Avh432 sequences are identical in virulent and avirulent strains of P. sojae, and that no transcriptional polymorphisms could be detected. The Avh26 sequence in P7064 and P7074 differs by a single nucleotide from that in P6497, causing an amino acid change in the predicted protein, but no transcript could be detected for Avh26 in any of the P. sojae strains. In contrast, comparison of the Avh27 sequence revealed many nucleotide and amino acid differences between virulent and avirulent P. sojae strains. Furthermore, Avh27 was determined to be present in multiple copies in each of the P. sojae strains. A pair of closely related sequences, named Avh27a and Avh27b, was assembled from trace files from whole genome sequence data ( Figure 1B). Each of the genes, Avh27a and Avh27b, differ in nucleotide and predicted amino acid sequence in P. sojae strain P7064 compared to P6497. The copy number of each gene was estimated by depth-sampling of trace files, because previous work has shown that this method presents a reliable way to estimate the number of copies of particular RXLR genes in P. sojae P6497. Sampling of trace files suggests that P. sojae P6497 contains two identical copies of Avh27a and a single copy of Avh27b ( Figure 1C). Analysis of genomic DNA of P. sojae P6497 using Avh27a-specific primers by real-time PCR resulted in a determination of 2.0560.404 copies per haploid genome, a value in agreement with the estimate from depth sampling.

Avh27a corresponds to Avr3c
To determine whether Avh27a or Avh27b cosegregate with Avr3c, we developed DNA markers for each of the genes and tested a collection of F 2 progeny that was additionally scored for virulence on Rps3c, as shown in Figure 1D. This analysis demonstrates that Avh27a precisely cosegregates with Avr3c, whereas recombination between Avh27b and Avr3c is evident. With all of the results pointing to Avh27a as the best candidate for Avr3c, we preformed DNA transformation by co-bombardment of soybean leaves, to test whether Avh27a interacts with Rps3c. Plasmid constructs of Avh27a and Avh27b, with and without native signal peptide, were bombarded into soybean leaves along with a reporter gene to measure cell viability, as shown in Figure 2. Results indicate that expression of Avh27a triggers cell death specifically on plants carrying Rps3c, regardless of the presence of the signal peptide. Transformation with Avh27b did not result in any differences in reporter gene expression in comparison to controls, indicating that this gene does not trigger cell death on Rps3c plants. Thus, Avh27a from P. sojae P6497 was renamed Avr3c P6497 .
The Avr3c transcript is expressed early during infection The expression of Avr3c was compared in a collection of P. sojae strains to determine whether any difference in transcript levels could be associated with virulence phenotypes. The Avr3a gene was also measured for comparison, because transcription of this gene is known to vary among P. sojae strains. Results shown in Figure 3A demonstrate that Avr3c is expressed in all P. sojae strains tested, in contrast to Avr3a. For P. sojae strains with detectable Avr3a transcripts, the expression of this gene differed from Avr3c. The expression of Avr3c peaked at 24 h after infection and declined rapidly, being scarcely detectable at the 48 h time point, whereas comparable levels of Avr3a transcript were present at 24 h and 48 h. Thus, expression of Avr3a and Avr3c differs markedly. Furthermore, the uniformity of expression of Avr3c among the different P. sojae strains indicates that virulence is not associated with loss of the Avr3c transcript, for any of the strains that were tested.
The expression of Avr3c in P. sojae P6497 was also compared to predicted RXLR genes represented on a commercially available microarray. Of the 15,820 P. sojae target sets on the array, we determined that 107 of these exactly match to one or more of the 385 predicted RXLR effectors from P. sojae [14]. Hybridizations were performed using mRNA isolated from germinating zoospores, infected plant tissues (compatible interaction), and in vitro grown mycelia. By applying a cut-off filter, we found that only 58 of the 107 array targets displayed hybridization intensities that were above background levels in one or more of the seven treatments. This set of 58 expressed RXLR effectors included Avr3c, Avr1b-1, and Avr4/6. The Avr3a gene was not among the 107 targets on the array. Expression patterns of the 58 RXLR effectors were arranged by cluster analysis, as shown in Figure 3B. Two broad groups that displayed contrasting expression patterns emerged from this analysis. The first group of 45 RXLR effectors showed highest relative expression during infection, while the second group contained 13 RXLR effectors that were preferentially expressed in germinating zoospores. Varying patterns of expression within each of the groups was also noted. For example, considering the first group of effectors with highest expression during infection, expression of many of these abruptly declined after the 24 h time point while others continued to be expressed at high levels at 48 h. The expression of Avr3c, Avr1b-1, and Avr4/6 reached their highest levels at 24 h after inoculation then declined at 48 h.

Avr3c is embedded in a tandem array
To determine the organization of the genetic space around Avr3c we examined the genome assembly of P. sojae P6497, and compared this to results from DNA blot and real-time PCR analyses, and depth-sampling of trace files. Results indicate that Avr3c and Avh27b are embedded within a 33.7 kb tandem repeat, as shown in Figure 4A. A list of predicted genes occurring in the replicated segment is provided in the Table S1. Each 33.7 kb repeat contains nine predicted genes, including one RXLR effector gene corresponding to Avh27b (one copy) or Avr3c (two copies). Other P. sojae strains examined appear to have a similar structural arrangement. However, comparison of conserved syntenic regions in three additional oomycete species indicates the Avr3c interval has diverged substantially in P. sojae ( Figure S1). The arrangement of many of the flanking genes remained conserved in P. ramorum, P. infestans, and Hyaloperonspora arabidopsis, but these regions do not encode an Avr3c ortholog nor any predicted RXLR effectors whatsoever.

Gain of virulence is caused by amino acid changes in the effector domain of Avr3c
Since there were no differences in expression of Avr3c among P. sojae strains, we sequenced this gene in each strain to determine any allelic differences, and tested whether virulent alleles evaded recognition by Rps3c. An alternate allele present in P7064, now called Avr3c P7064 , displayed 11 differences in its deduced amino acid sequence in comparison to the Avr3c P6497 protein, as shown trace files from whole genome shotgun sequencing of P. sojae P6497. The number of sequence matches with an expect (E) value,10 240 is indicated on the left axis. The predicted number of matches for a segment with one, two, or three copies per haploid genome is shown to the right of the graph, with the bars indicating the 95% confidence interval for each class. Results from a single copy reference gene, CL164, are shown for comparison. (D) Co-segregation of Avh27a but not Avh27b with Avr3c virulence phenotype in F 2 progeny. A cross of P. sojae P64976P7064 generated the F 2 progeny. The cultures were scored as virulent (V) or avirulent (A) on Rps3c soybean plants. Cleaved amplified polymorphic (CAP) markers, codominant and specific for each parental allele of Avh27a and Avh27b, were scored using genomic DNA from the parents and F 2 progeny. Shown is a photograph of an ethidium bromide-stained agarose gel of the CAP marker results. doi:10.1371/journal.pone.0005556.g001 in Figure 4B. Two other Avr3c alleles were discovered that showed DNA sequence differences compared to Avr3c P6497 . These two alleles, represented by Avr3c ACR8 and Avr3c ACR9 , differ in DNA sequence from each other but the predicted protein products encoded by their ORF are identical ( Figure 4B). The Avr3c ACR8 and Avr3c ACR9 protein products are indistinguishable from each other and nearly identical to Avr3c P7064 , differing by only four amino acids. Each of the three distinct Avr3c proteins were 220 amino acids in length and contained two predicted W-motifs within the effector domain. The P. sojae strain-specific amino acid differences in Avr3c were localized exclusively to the effector domain. In four strains of P. sojae, we could isolate two different copies of the Avr3c gene ( Figure 4C). Since copy number analysis indicated that Avr3c P6497 is present in two identical copies in P. sojae P6497, we conclude that the two copies of Avr3c have diverged slightly in sequence in the P. sojae strains ACR8, ACR9, ACR17, and ACR24. It was important to test each of the Avr3c alleles for interaction with Rps3c by the co-bombardment assay, because all P. sojae strains tested expressed this gene despite their differences in virulence on Rps3c-containing plants. These results demonstrate that Avr3c alleles from virulent strains of P. sojae do not cause cell death on Rps3c plants, in contrast to the Avr3c P6497 allele present in P6497 and other avirulent strains ( Figure 2). Thus, we conclude that Avr3c P6497 is recognized by Rps3c but Avr3c P7064 and Avr3c ACR8 are not.

Discussion
The identification of oomycete Avr genes is now happening at a rapid pace, facilitated by genome sequence data and the discovery of the RXLR host-targeting motif. This is occurring after more than 30 years of intensive research to find Avr factors in P. sojae and P. infestans. It is remarkable that at least nine different Avr genes corresponding to RXLR effectors have been described in three species of oomycete plant pathogens in the last five years [6,7,8,9,16,17,18,21]. Additional oomycete Avr genes that do not encode RXLR effectors have also been proposed, such as Avr3b-Avr10-Avr11 in P. infestans [24,25] and Avr1b-2 in P. sojae [6]. Likewise, many new Avr genes from fungal plant pathogens are now being identified or proposed [4].
Our finding that the Avr3c gene of P. sojae lies in close proximity to the Avr4/6 gene substantiates past studies describing genetic linkage of these avirulence determinants [22]. The clustering of Avr  genes in P. sojae and P. infestans was discovered by studying F 2 populations and hybrids [26,27,28,29]. With the completion of whole genome sequences and associated physical maps, identification of one Avr gene may facilitate the identification of additional Avr genes that are genetically linked, as we have demonstrated in this study.
Results from the co-bombardment analysis show that Avr3c is able to trigger cell death (measured by reduced GUS expression) specifically in plants carrying Rps3c. This occurred regardless of the presence of the signal peptide, as has been noted for other Avr genes [13,16]. Previous studies have demonstrated that the RXLR motif is necessary and sufficient for delivery of proteins into plant cells [12,13], and it is now widely accepted that the RXLR motif functions as a host targeting signal. Thus, expression of full-length Avr3c in plant cells likely results in the protein being exported via the secretory pathway and re-imported via the RXLR mediated pathway. Overall, the results suggest that Avr3c is recognized within the plant cell.
As more Avr genes are identified in P. sojae, it is possible to compare their expression patterns to one another and to other RXLR effectors in the genome. Our results in this area show that RXLR effectors have varying patterns of expression, but most are expressed at their highest levels during early infection and in germinating zoospores. These results are not surprising because there is evidence that RXLR effectors assist the colonization of plant tissues by suppression of host defense responses [21,30]. This will be especially crucial during the early stages of colonization, while P. sojae is growing biotrophically on the host. Previous work has shown that by 48 h after inoculation, P. sojae transits to a necrotrophic growth mode that is accompanied by expression of hydrolytic enzymes and toxins [31,32,33,34]. Despite the overall similarity of RXLR effector expression, we noted differences and found groups that shared distinct expression patterns. For example, the expression of Avr3c and Avr1b-1 abruptly decline at 48 h after infection, while Avr3a and many other effectors continue to be expressed well into this necrotropic phase. These expression differences may be related to varying functional roles that the effectors engage in with the host, and the need for the pathogen to coordinate effector gene expression with that of particular host targets.
The arrangement of the Avr3c locus provides yet another example of a multi-copy P. sojae Avr gene occurring in a tandem array of duplicated DNA segments. Recent studies have shown that P. sojae Avr1a and Avr3a are multi-copy genes that display copy number variation among different P. sojae strains [16]. Although we noted sequence polymorphisms within the Avr3c and Avh27b genes among different P. sojae strains, there was no evidence for copy number variation of the repetitive unit among strains, as described for Avr1a and Avr3a. The P. sojae strains that are virulent on Rps3c plants display amino acid polymorphisms in the effector domain of Avr3c. Evasion of immunity by amino acid substitutions within the effector domain has also been described for P. sojae Avr1b-1 [6], P. infestans Avr3a [9], and H. arabidopsis ATR13 [7] and ATR1 [8]; whereas premature stop codons cause truncated proteins and gain of virulence at the P. infestans Avr4 locus [17]. Transcriptional variation represents another mechanism for gain of virulence, as described for P. sojae Avr1b-1, Avr1a, and Avr3a [6,16].
The arrangement of Avr genes in clusters offers opportunities for variation generated by sequence exchanges via unequal crossing over, homologous recombination, or gene conversion. Our results suggest that such processes occurred at the P. sojae Avr3c locus. Gain of virulence on Rps3c plants requires mutations to each of the two identical copies of Avr3c P6497 that occur in avirulent strains of P. sojae. The P. sojae strains P7064 and ACR6 are virulent on Rps3c and carry two identical copies of the Avr3c P7064 allele. The mutations that define the Avr3c P7064 sequence could not have simultaneously arisen in the two Avr3c gene copies. Rather, these changes must have occurred in one copy and then spread to the second copy by sequence exchanges. It is also evident that four P. sojae strains, ACR8, ACR9, ACR17 and ACR24, have further evolved at this locus, by accumulating additional mutations in the Avr3c ACR9 and Avr3c ACR8 alleles. In these four P. sojae strains, the Avr3c ACR9 and Avr3c ACR8 alleles are clearly derived from Avr3c P7064 but no sequence homogenization has occurred between the two copies of Avr3c.
The finding that Avr3c is a multi-copy RXLR effector that acquired gain of virulence mutations that spread through sequence exchanges provides a new example of the plasticity of the RXLR effector family. Similar mechanisms for generating diversity at plant resistance (R) gene loci have long been known [35,36]. Thus, plant immune systems and pathogen effector systems mirror each other in an additional way, by relying on gene clusters and associated sequence exchange mechanisms to provide rapid and novel changes to genes that control immunity and virulence.

Phytophthora sojae isolates, plant materials and disease assays
Working stocks of P. sojae was routinely grown on 9cm 26% (v/ v) V8 agar plates at 25uC for 5 to7 days in the dark [37]. The source of P. sojae isolates used in this study is described in Table S2. For plant infection assays P. sojae cultures were grown on V8 media containing 0.9% agar. Axenic cultures for nucleic acid isolation were prepared by transferring 5 mm mycelial disks cut from the growing edge of each colony to vegetable juice (V8) agar plates layered with a disc of cellophane (BioRad). After growth, cellophane sheets overlaid with fully grown mycelial colonies were peeled off the media and flash frozen in liquid nitrogen for extraction.
Soybean (Glycine max) cultivar Williams (rps3c) and the Williams isoline L92-7857 (Rps3c), were from the collection at Agriculture and Agri-Food Canada (Harrow, Ontario) and used to evaluate the virulence of P. sojae cultures. Etiolated soybean seedlings were grown in vermiculite soaked in 3 mg/L fertilizer (15-30-15) at 25uC day (16 h) and 16uC night (8 h) temperatures for 7 days prior to harvest for disease assays. A mycelial plug (5 mm diameter), cut from the growing edge of 5 to 7 day old V8 grown P. sojae cultures, was transferred to each of 15 to 20 hypocotyls per cultivar, mycelial-side down, 2 to 3 cm from the base of the cotyledon. For light-grown soybeans, six soybean seeds were sown in 10 cm pots (a minimum of three pots per isolate) containing soilless mix (Pro-Mix 'BX', Premier Horticulture Ltd, Rivière-du-Loup, Canada) soaked with 3 mg/L fertilizer (20-20-20). Plants were grown in a controlled growth chamber with a 16 h photoperiod, 25uC day and 16uC night temperatures. Plants were grown for one week for use in virulence assays, or for two weeks for use in biolistics. P. sojae cultures were grown on 0.9% (v/v) V8 agar plates 5 to 7 days prior to light-grown plant inoculations. Mycelia inoculums were prepared by passing the actively growing edge of a culture through a 3 ml syringe attached to an 18-gauge needle. Soybean plants were inoculated in the mid-section of each hypocotyl by making a small incision for injection of the mycelial slurry. Inoculated plants were covered with plastic bags to maintain humidity for two days. Disease symptoms were allowed to develop for an additional four days before phenotypes were scored as resistant, susceptible or intermediate. A minimum of three independent replicates of the disease assay were performed for each P. sojae culture tested.

Microarray hybridization and analysis
P. sojae P6497 and the compatible soybean cultivar Harosoy, were used for all treatments. Methods for isolation of mRNA from germinating zoospores, infected plant tissues, and mycelia have been described [31,33]. Plant inoculations were performed by placing mycelia plugs on etiolated hypocotyls. A minimum of three biological replicates were performed for each of the seven treatments in the microarray experiment. Integrity, purity and concentration of RNA were verified by electrophoresis (Bioanalyzer, Agilent Technologies) before hybridization to high-density oligonucleotide arrays containing probe-sets for 15,820 predicted P. sojae genes (Affymetrix Soybean GeneChip). Data was normalized and analyzed using computer software (GeneSpring GX 7.3). Spot intensity from was interpreted through an RMA pre-processor. The following normalization steps were sequentially performed: (1) Data transformation: set measurements less than 0.01 to 0.01; (2) Per chip: normalize to a set of pre-determined genes, using all 15,820 P. sojae genes as control genes; (3) Per gene: normalize to median. To obtain a set of RXLR genes on the array, the nucleotide sequences of 385 predicted P. sojae Avh genes were downloaded from JGI P. sojae genome assembly1.1. To determine the Avh genes with corresponding probes on the microarray, computer software was used (www.affymetrix.com/analysis/netaffx/index.affx). Predicted Avh genes with sequences exactly matching to 11/11 probe-sets were considered as having a perfect probes on the microarray.

Plasmid construction and plant transient expression assays
Primers used in this study are provided in Table S3. The primer sets named Avh27gF/Avh27gR, Avh28gF/Avh28gR, and Avh26gF/Avh26gR were used to amplify the Avh27, Avh28 and Avh26, respectively. For each amplification product, PCR bands were excised and purified (QIAquick Gel Extraction Kit, Qiagen) prior to sequencing and sub-cloning. Each of the Avh27a and Avh27b constructs, with or without signal peptide, were amplified by using different combinations of primers. Products were digested with BglII and SphI and inserted into the PFF19 plant transient expression vector. Recombinant plasmids were purified for sequencing and for plant transient expression assays. Cobombardment assays were performed as previously described [32,38]. After bombardment, leaves were left at room temperature for 16 h then transferred to GUS staining solution and incubated in darkness at 37 C for 16 h [32,38]. Leaves were then washed in 70% (v:v) ethanol and photographed using a digital camera.

Genotyping of F 2 progeny
Fragments amplified by primer sets Avh27abF-SP/Avh27aR and A27bRT-F/A27bRT-R were digested by RsaI and XmnI, respectively. The assays provided co-dominant markers for scoring the genotype of P. sojae P6497 or P7064 alleles of Avh27a and Avh27b. The F 2 populations resulting from a cross of P. sojae P64976P7064 have been described [23,27]. A total of 40 F 2 individuals were selected, based upon their genotypes for Avh27a and Avh27b, and their virulence was assessed on Rps3c soybean plants.

Nucleic acid isolation and copy number determination
Methods for RNA isolation and reverse transcriptase PCR from P. sojae mycelium, zoospores, germinating zoospores, and P.sojaeinfected soybean tissues have been described [31,33]. Genomic DNA was isolated from P. sojae mycelial cultures using a modified CTAB (hexadecyl trimethyl ammonium bromide) method [39]. Gene copy number determinations were made by quantitative PCR, using an instrument that measures products in real-time (LightCycler, Roche, Laval, PQ, Canada, software version 3.5). A reference plasmid construct containing the P. sojae genes CL164 and Avr3c was used to develop a standard curve with data points ranging from 1 to 108 copies. The gene CL164 is present in single copy in the genome of P. sojae P6497. Primers specific for Avr3c and CL164 are available in Table S3. Amplification reactions (20 ml) were performed in duplicate with 106.6 or 1066 pg of input genomic DNA, 3.25 mM MgCl 2 , 0.5 mM of each primer and 2 ml of master mix (FastStart DNA Master SYBR Green I mix, Roche). The PCR parameters were as follows: an initial 10 min denaturation step at 95uC followed by 40 cycles of 15 s at 95uC, 10 s at 65uC and 12 s at 72uC. Specificity of the primers was verified by a melting curve analysis of the PCR products with a temperature gradient of 0.2uC/s from 68uC to 98uC and by conventional gel electrophoresis. Copy number of Avr3c was determined as a ratio of the estimated copies of Avr3c to that of the reference gene, CL164. The estimation of gene copy number by counting of trace file matches from whole genome shotgun sequences was performed as previously described [16].

Genome structure analysis
The 80 kb of genomic sequence encompassing Avh27a and Avh27b was downloaded from P. sojae genome assembly v3.0. Computer software (DNA Star, Lasergene) was used to search for imbedded repetitive segments. A replicate unit containing Avh27b was well-assembled, while another nearly identical unit containing Avh27a contained a gap in the assembly. After analysis and reassembly, it was determined that two copies of the replicate unit containing Avh27a were present in the genome of P. sojae P6497. The re-assembled Avh27 genomic region was compared to syntenic regions in P. ramorum (JGI P. ramorum v1.1 database), P. infestans (Broad Institute, Phytophthora Database) and Hyaloperonospora arabidopsis (VBI microbial database, H. arabidopsis assembly v3.0).

Data Deposition
The sequence data for the Avr3c alleles have been deposited to NCBI GenBank under the accession numbers: FJ705360, FJ705361, FJ705362 and FJ705363. Microarray expression data has been deposited to NCBI-GEO, series GSE15100. Figure S1 Conservation and interruption of synteny of the Avr3c region in P. sojae, P. ramorum, P. infestans, and Hyaloperonospora arabidopsis.