Soybean (Glycine max (L.) Merr. ) is the most important legume in the world. However, the rapid expansion of its cultivated areas has created new ecological niches for many pathogens. Among them, Phytophthora sojae (Kaufmann and Gerdemann) ranks as one of the most damaging soybean pests in the world. The most common method to control it is the introgression of resistance genes termed Rps (Resistance to P. sojae ) into elite cultivars. This imposes a high selection pressure on P. sojae leading to the development of new virulent pathotypes. Consequently, more durable sources of resistance are needed to manage P. sojae . A complementary approach resides in the exploitation of quantitative trait loci (QTL) associated with partial resistance (PR) which has been found to be more durable and effective against a broad spectrum of pathotypes (Karhoff et al., 2019). Several QTLs for PR of soybean against P. sojae have already been reported and are listed on SoyBase (Grant et al. , 2010). However, limited information is available about the precise nature and role of genes within those QTLs.

To exploit PR efficiently, an in‐depth characterization of the genetic regions involved is essential. Recent advances in high throughput genotyping by sequencing (GBS) techniques exploiting next generation sequencing (NGS) technologies provide abundant genome‐wide SNPs at low cost allowing precise mapping. These NGS advancements have also revolutionized transcriptome profiling (RNA‐seq), making possible the analysis of differentially expressed genes located within QTLs, an approach that has been proposed as a critical validation tool for candidate genes. Moreover, RNA‐seq, as a genotyping tool, represents an alternative reduced‐representation approach focusing on protein‐coding regions (Scheben et al. , 2017).

One of the main challenges in the study of PR against P. sojae has been the lack of reliable methods to precisely characterize the phenotypes. Indeed, the most common assays, the layer test and/or the tray test, provide mainly qualitative estimates of PR, which can be biased in the presence of Rps genes (Karhoff et al. , 2019). To overcome these constraints, a hydroponic assay, developed by Lebreton et al. (2018), reproduces the key steps of the soybean–P. sojae interaction and allows the simultaneous inoculation of isolates covering all pathotypes thus eliminating the possible effect of Rps genes.

Sources of horizontal resistance against P. sojae are limited, and the problem is accentuated in Canada, where early maturity soybean varieties are required because of the short growing season. For this reason, the early maturity line PI 449459 reported to exhibit a high level of PR represented a rare opportunity. By using optimized GBS and the new phenotyping approach, the present study aimed to identify QTLs conferring PR using a recombinant inbred line (RIL) population derived from early maturity parents differing for PR. In order to define with greater resolution the putative genes involved in PR within QTLs, an RNA‐seq approach (BioProject ID: PRJNA574764) coupled with bioinformatic prediction tools were exploited to detect variation in gene expression and/or sequences to identify the most relevant candidate genes linked to PR.

Evaluation of the F_5:6 RILs using the hydroponic assay coupled with a mixed inoculum carrying pathotypes to all common Rps genes provided a wide spectrum of responses going from plant death to severe to low root rot and to near absence of symptoms (Figure 1a). This response could be quantified using a single variable, the corrected dry weight (CDW; Stewart and Robertson, 2012), amenable to QTL analysis. When CDW (dry weight of inoculated vs control) of the RILs was measured at 21 dpi, a wide variation of phenotypes was obtained (Figure 1b).

A GBS approach on the F5 progeny of the cross between PI 449459 and Misty yielded a total of 1078 non‐redundant SNPs. The set of SNPs was subsequently used to construct a linkage map covering 2300 cM (over 93 % of the reference map) across 24 linkage groups representing 20 chromosomes with a marker density of one marker every 2.1 cM. Inclusive composite interval mapping using QTL IciMapping V 4.1 identified two QTLs inherited from PI 449459 and associated with PR to P. sojae in our RIL population. They were located on chromosomes 13 and 19 (designated QTL‐13 and QTL‐19) explaining 17.6% and 13.1% of the phenotypic variance, respectively (Figure 1c). The confidence intervals (defined using a one‐unit decrease of the peak LOD score) were mapped between 96.5 and 100.5 cM and were flanked by markers Chr13:28842184 and Chr13:30776191 (the nomenclature of markers is chromosome: physical position (bp)) for QTL‐13, and were mapped between 126.5 and 127 cM and were flanked by markers Chr19:50040258 and Chr19:50556102 for QTL‐19. The QTL‐13 and QTL‐19 intervals contained a total of 204 and 66 candidate genes, respectively. In order to reduce the number of candidate genes, the expression of these genes in response to infection and the predicted functional impact of nucleotide variants located within their coding regions were both investigated.

An RNA‐seq strategy to compare the different expression patterns of genes underlying the QTLs was performed by sequencing the transcriptomes of the resistant and susceptible parents, under both infected and control conditions. We focused on genes showing a > 5‐fold change in expression specific to P. sojae infection between the resistant and susceptible parents, and this analysis yielded four and two differentially expressed genes (DEGs) after infection for QTL‐13 and QTL‐19, respectively (Figure 1c).

In parallel, the sequence data from the RNA‐seq libraries were used to identify mutations inducing a modification in peptide sequence resulting in altered protein function. The effect of the SNPs was determined with the Variant Effect Predictor bioinformatic tool (Ensembl.org). One hundred and fifteen mis‐sense mutations, including two that induced a gain/loss of a stop codon and five inducing a frameshift, were found in 52 genes (Figure 1c). The analysis was further refined by examining SNPs predicted to modify the folding of proteins (score > 0.9) using PolyPhen‐2 (Kono et al. , 2018). This analysis uncovered 11 SNPs located in three and five candidate genes for QTL‐13 and QTL‐19, respectively.

Interestingly, NBS‐LRR Glyma.13G190400 and AP2/ERF‐type transcription factor Glyma.19G262700 were the only ones identified by both expression and nucleotide variation analyses making them extremely promising candidate genes for further functional characterization studies of PR to P. sojae . The RNA‐seq results for these genes were confirmed by qRT‐PCR with parental lines and RILs presenting susceptible and resistant phenotypes. AP2/ERF‐type transcription factors have been reported as playing a critical role in tolerance to biotic stress in several economically important crops such as rice, wheat, barley and soybean, leading Jisha et al. (2015) and Phukan et al. (2017) to recommend their use in breeding programmes. In our study, the AP2/ERF‐type transcription factor Glyma.19G262700 was up‐regulated 9‐fold in the resistant genotype during the early stage of the infection process (4 dpi) which is consistent with a role in resistance as reported in the studies mentioned above. Also, of interest, Glyma.13G190400 was expressed 7‐fold more in the susceptible genotype, which may favour a positive outcome for the pathogen (McHale et al. , 2006). Consistent with our results, Rasoolizadeh et al. (2018) observed a higher expression of NBS‐LRRs in a compatible interaction soybean–P. sojae .

In conclusion, this study proposes an integrated approach, exploiting a new phenotyping procedure, RNA‐seq analyses and SNP variants of predicted functional impact, to discriminate and prioritize high‐value candidate genes modulating complex quantitative traits such as those defining PR during plant–pathogen interactions.

大豆（大豆（L.）大豆- 。）是世界上最重要的豆科植物。但是，其耕地面积的迅速扩大为许多病原体创造了新的生态位。其中，大豆疫霉菌（Kaufmann和Gerdemann）被列为世界上最具破坏力的大豆害虫之一。控制它的最常见方法是将称为Rps（对大豆疫霉菌的抗性）的抗性基因渗入优良品种中。这对大豆假单胞菌施加了很高的选择压力，导致了新的致病性致病型的发展。因此，需要更持久的抗药性来控制大豆疫霉菌。一种补充方法在于与部分抗性（PR）相关的定量性状基因座（QTL）的开发，已发现它对多种病原体具有更强的持久性和有效性（Karhoff et al。，2019）。大豆对大豆疫霉的PR的几个QTL已有报道，并在SoyBase上列出（Grant等，2010）。但是，有关这些QTL中基因的确切性质和作用的信息有限。

为了有效利用PR，对涉及的遗传区域进行深入表征是必不可少的。利用下一代测序（NGS）技术的测序（GBS）技术进行高通量基因分型的最新进展以低成本提供了丰富的全基因组SNP，可进行精确定位。这些NGS的进步也彻底改变了转录组分析（RNA-seq），使得分析位于QTL内的差异表达基因成为可能，该方法已被提议作为候选基因的关键验证工具。此外，RNA-seq作为一种基因分型工具，代表了另一种以蛋白质编码区域为重点的简化表示方法（Scheben et al。，2017）。

对大豆疫霉菌进行PR研究的主要挑战之一是缺乏可靠的方法来精确表征表型。实际上，最常用的测定法，该层测试和/或托盘测试，提供PR，它可以在存在被偏置的主要定性估计RPS基因（Karhoff等人。，2019）。为了克服这些限制，Lebreton等人开发了一种水培测定法。（2018），再现了大豆与大豆疫霉菌相互作用的关键步骤，并允许同时接种涵盖所有病原体的分离株，从而消除了Rps基因的可能影响。

抗大豆疫霉菌水平抗性的来源数量有限，加拿大的问题更加突出，加拿大由于生长季节短而需要提早成熟的大豆品种。出于这个原因，据报道表现出高水平PR的早熟品系PI 449459代表了难得的机遇。通过使用优化的GBS和新的表型方法，本研究旨在使用衍生自不同PR的早熟父母的重组自交系（RIL）群体鉴定赋予PR的QTL。为了以更高的分辨率定义QTL中与PR相关的推定基因，利用RNA-seq方法（BioProject ID：PRJNA574764）结合生物信息学预测工具来检测基因表达和/或序列的变异，从而确定最相关的候选基因与PR相关的基因。

使用水培法对F _5：6 RIL进行评估，并结合对所有常见Rps基因携带病原体的混合接种物，提供了从植物死亡到严重腐烂到低根腐烂以及几乎没有症状的广泛响应（图1a） 。可以使用适合QTL分析的单个变量（校正后的干重（CDW； Stewart和Robertson，2012年））来量化此响应。当在21 dpi下测量RIL的CDW（接种干重与对照干重）时，获得了广泛的表型变异（图1b）。

针对PI 449459和Misty之间杂交的F5后代的GBS方法，共产生了1078个非冗余SNP。随后将这组SNP用于构建代表24个连锁组的2300 cM（占参考图谱的93％）的连锁图谱，该连锁组代表20条染色体，标记密度为每2.1 cM一个标记。使用QTL IciMapping V 4.1的包容性复合区间映射确定了两个从PI 449459继承并与PR到大豆疫霉菌相关的QTL 。在我们的RIL人口中。它们位于13号和19号染色体（分别称为QTL-13和QTL-19）上，分别解释了表型变异的17.6％和13.1％（图1c）。置信区间（使用降低的LOD峰值降低一个单位来定义）位于96.5和100.5 cM之间，并位于标记Chr13：28842184和Chr13：30776191的两侧（标记的命名法是染色体：物理位置（bp）） QTL-13的位置，映射在126.5和127 cM之间，并位于QTL-19的标记Chr19：50040258和Chr19：50556102的两侧。QTL-13和QTL-19间隔分别包含总共204和66个候选基因。为了减少候选基因的数量，

通过在感染和对照条件下对抗性和易感亲本的转录组进行测序，进行了一种RNA-seq策略，以比较QTL潜在基因的不同表达模式。我们集中研究了抗性和易感亲代之间大豆疫霉感染特异性表达变化> 5倍的基因，该分析在感染QTL-13和QTL-19后产生了四个和两个差异表达基因（DEG），分别（图1c）。

同时，来自RNA-seq库的序列数据用于鉴定引起肽序列修饰的突变，从而导致蛋白质功能改变。SNP的作用是用Variant Effect Predictor生物信息学工具（Ensembl.org）确定的。在52个基因中发现了115个错误义突变，包括两个诱导终止密码子增减的错义突变和五个诱导移码的错义突变（图1c）。通过使用PolyPhen-2（Kono et al。，2018）检查预测可改变蛋白质折叠（得分> 0.9）的SNP进一步完善分析。该分析发现了分别位于QTL-13和QTL-19的三个和五个候选基因中的11个SNP。

有趣的是，NBS‐LRR Glyma.13G190400和AP2 / ERF型转录因子Glyma.19G262700是通过表达和核苷酸变异分析鉴定的唯一基因，这使其成为大豆PR大豆进一步功能表征研究的极有希望的候选基因。这些基因的RNA-seq结果通过qRT-PCR与亲本系和RIL表现出易感和抗性表型的方法得到证实。据报道，AP2 / ERF型转录因子在水稻，小麦，大麦和大豆等几种经济上重要的农作物对生物胁迫的耐受性中起着至关重要的作用，这导致了吉沙等人的发现。（2015年）和Phukan等人（2015年）。（2017年）建议将其用于育种程序。在我们的研究中，在感染过程的早期阶段（4 dpi），AP2 / ERF型转录因子Glyma.19G262700的抗性基因型上调了9倍，这与抗药性中的作用一致。上面提到的研究。同样，有趣的是，Glyma.13G190400在易感基因型中表达高7倍，这可能有利于病原体的阳性结果（McHale等，2006）。与我们的结果一致，Rasoolizadeh等。（2018）观察到在兼容的大豆大豆互作中，NBS-LRRs的表达更高。

总之，本研究提出了一种综合方法，利用新的表型分析方法，RNA-seq分析和预测功能影响的SNP变体，来区分和优先考虑可调节复杂数量性状（如在植物病原体中定义PR的那些）的高价值候选基因。互动。