Introduction

Spring black stem and leaf spot (SBS) disease is a globally distributed disease of Medicago truncatula and Medicago sativa (alfalfa) [1]. Notably, SBS disease is one of the most severe foliar disease of alfalfa in Australia, Iran, Europe, and Canada [1,2,3]. The causal agent of SBS disease is Ascochyta medicaginicola, previously known as Phoma medicaginis. With the expansive genomic resources available for M. truncatula, this interaction presents an opportunity to study the host response to necrotrophic fungal pathogens of legumes [4]. The symptoms of SBS disease include necrotic lesions and chlorosis of the foliar tissue as well as the stems, which results in defoliation of the lower canopy. In alfalfa, yield losses are especially pronounced in the first or second harvest after a wet spring.

Complete resistance to SBS disease has not been observed. For resistant genotypes of M. sativa and M. truncatula, spore germination, penetration, and pycnidia development are delayed [2, 5]. Diseased plant material has higher amounts of the phytoestrogen coumestrol, which can impact livestock fertility [6]. South Australian Research and Development Institute (SARDI) maintains a large diverse collection of M. truncatula. Eighty-six of the SARDI M. truncatula accessions were screened for SBS disease response, and most were found to be susceptible; however, genotype-specific resistance was seen in 16 accessions, including SA27063, also known by the Medicago HapMap identifier HM078. On a 1 to 5 scale increasing in disease severity, HM078 has a mean disease rating of 1.64 against A. medicaginicola isolate OMT5, whereas the susceptible accession A17 (HM101) has a mean disease rating of 4.15 [2].

SBS-resistant accession SA27063 (HM078) and SBS-susceptible accessions A17 (HM101) and SA3054 were used as parents to generate two populations for quantitative trait locus (QTL) mapping that discovered rnpm1 (HM101 & HM078) and rnpm2 (SA3054 & HM078), two recessively inherited QTL which account for approximately 30% of the phenotypic variance for resistance to SBS disease of M. truncatula [7]. In addition, SA27063 (HM078) and SA3054 were also used as resistant and susceptible genotypes, respectively, in a microarray study of the host transcriptome at 12 h post inoculation (hpi) with A. medicaginicola [8]. In that study, Kamphuis et al. [8] found upregulation of the phenylpropanoid and octadecanoid pathways associated with defense responses. Another transcriptome study of SBS disease of alfalfa showed that several pathogenesis-related (PR) proteins were significantly upregulated upon infection with A. medicaginicola [9].

Plant defense responses to necrotrophic pathogens are complex and often differ from the host responses to biotrophic pathogens. There are two general arms of the plant immune system. First, an initial detection of pathogen associated molecular patterns (PAMPs) by transmembrane proteins called pattern recognition receptors (PRRs). PRR proteins are described as receptor-like kinases (RLKs) or receptor-like proteins (RLPs) that bind to PAMP ligands, and promote PAMP-triggered immunity (PTI). PTI includes callose deposition, lignification, an oxidative burst by reactive oxygen species (ROS), the production of PR proteins, the synthesis of antimicrobial compounds like phytoalexins, and production of plant hormones [10, 11]. Virulent pathogens possess effectors that dampen PTI. Plant disease resistance genes, also known as nucleotide-binding site and leucine-rich repeat (NLR) genes, function in sensor-helper pairs to detect effectors and initiate programmed cell death (PCD) [12]. The second arm of the plant immune system is the detection of these effectors by intracellular NLR proteins [13]. In the gene-for-gene model, NLR-mediated recognition of effector proteins results in effector-triggered immunity (ETI) and PCD. Specific NLR proteins have been shown to confer susceptibility to toxins of necrotrophic pathogens in the inverse gene-for-gene model also known as effector-triggered susceptibility (ETS) [14, 15]. Conversely, NLR proteins have also been found to confer resistance against necrotrophic fungi, such as the Dothideomycete pathogen Leptosphaeria maculans [16]. Resistance to necrotrophic pathogens has been associated with phytohormones such as salicylic acid (SA), jasmonic acid (JA), abscisic acid (ABA), and ethylene (ET), which regulate stress responses through signaling pathways [17]. For instance, the accumulation of JA in Arabidopsis thaliana has been associated with resistance to necrotrophic fungus Sclerotinia sclerotiorum [18]. Overall, plant immune responses need to be investigated in regard to specific pathosystems.

Comparative transcriptome analysis has been shown to be an effective method for identifying differentially expressed genes (DEGs) in response to plant-pathogen interactions. In this study, our objective was to identify candidate genes for SBS disease resistance for future validation in functional studies. We examined the host transcriptome of a resistant (HM078) and susceptible (A17) M. truncatula genotype at 24, 48, and 72 hpi with A. medicaginicola. We identified DEGs in the resistant and susceptible genotype compared to mock-treated samples at each time point and evaluated functionally enriched pathways. However, the number of DEGs was much lower in the resistant genotype. To identify candidate genes for disease resistance we examined the expression of SA and JA pathway genes, genes in QTL regions for disease resistance, RLKs, NLRs, and genes in functionally enriched pathways. We identified specific candidate genes based on five criteria; (1) among the top ten upregulated or downregulated genes in the resistant genotype, (2) upregulated DEGs over multiple time points in the resistant genotype, (3) DEGs in the susceptible genotype with higher constitutive expression in the resistant, (4) shared DEGs between resistant and susceptible with variable expression levels, or (5) genes in QTL regions rnpm1 and rnpm2 with contrasting expression profiles. We identified candidate genes for SBS disease resistance based on our comparative transcriptome analysis, functional annotations, and support from the literature. Overall, this study sheds light on the plant immune response to A. medicaginicola using contemporary genomic resources, and provides a number of strong candidate genes for SBS disease resistance to be validated in future studies.

Methods

Plant growth conditions

Germplasm of M. truncatula accessions A17 (HM101) and SA27063 (HM078) were obtained from the Medicago HapMap collection. Seed was scarified with 2 mL of concentrated sulfuric acid for 7 min, followed by washes with sterile de-ionized (DI) water. Seedlings were grown in autoclaved potting soil (Sun Gro Professional Growing Mix, Sun Gro Horticulture, Agawam, MA, USA) in a growth chamber at 22–24 °C with 16 h of light per day.

Inoculation procedure

Fungal cultures were maintained on potato dextrose agar (PDA) and exposed to ambient daylight on the benchtop throughout growth. Inoculum of A. medicaginicola was prepared from 4-week-old cultures by flooding plates with 5 mL of sterile DI water with 50 ppm Tween®20 surfactant (Sigma-Aldrich, St. Louis, MO) and dislodging conidia into suspension. Conidial suspensions were strained using a Falcon™ Cell Strainer with a 40 μm pore (Thermo Fisher Scientific, Waltham, MA, USA) to remove hyphal fragments. Conidial suspensions were quantified using a hemocytometer under 400x magnification and adjusted to 5 × 105 conidia/mL. The oldest trifoliate leaf originating from the node of the first secondary branch was marked with a white string tied to the petiole to be designated for inoculation. Approximately 1 mL of inoculum was atomized with a spray bottle at a distance of 15 cm away from the target leaf. Inoculated plants were placed in a humidity chamber at 100% relative humidity in the dark for 72 h following inoculation.

Microscopic evaluation of SBS disease at selected time points

Spore germination and fungal growth on the leaf surface was observed for each genotype and time point. Cross sections of infected leaves were made to evaluate hyphal penetration. A sliding microtome was used to take 10 μm cross sections to visualize fungal penetration. The infected leaf material was immersed in GFP Polyclonal Antibody, Alexa Fluor® 488 (496/518 nm) (Thermo Fisher Scientific, Waltham, MA, USA) in a phosphate buffered saline solution as previously described [19]. Alexa Fluor® 488 selectively binds to N-acetylglucosamine, the monomer component of chitin, allowing for the fluorescence of hyphae under GFP (482/524 nm) wavelengths.

RNA extraction and sequencing

At each time point (24, 48, and 72 hpi), three inoculated leaves and three mock-inoculated leaves were sampled from biological replicates of each genotype. A total of 36 inoculated leaves were harvested from SBS-resistant M. truncatula HM078 (n = 18) and SBS-susceptible A17 (n = 18) from 36 individual plants. Samples were not pooled between biological replicates. Leaves were harvested in low-light conditions. Tissue was stored at -80 °C until RNA extraction. Collected tissue was subjected to RNA extraction using the Qiagen RNeasy Mini Kit for Plants (Qiagen Inc., Valencia, CA, USA). For one sample an entire trifoliate leaf was homogenized in liquid nitrogen by mortar and pestle. Then, 75 mg of frozen tissue was sub-sampled, and added to 1 mL Buffer RLT (Qiagen). The rest of the protocol was followed according to the manufacturer’s specifications. Illumina RNA sequencing was conducted at the University of Minnesota-Twin Cities Genomics Center. TruSeq unique dual-indexed (UDI) stranded mRNA libraries were prepared, combined in a single pool, and sequenced on a single lane of NovaSeq S4 2 × 150-bp flow cell. Short-read RNA sequence data was uploaded to the Minnesota Supercomputing Institute for analysis.

RNA sequence read alignment and quantification

RNA sequence reads derived from mock and inoculated plant tissue samples were processed in a series of steps detailed in the associated code file. First, Cutadadpt v1.18 [20] was used to trim Illumina sequencing adapters, retaining RNA sequence reads with Phred-scaled quality scores above 30, and reads longer than 50 bp. FastQC reports of RNA sequence data statistics were summarized with MulitQC v1.14 [21]. The Mt5.0 reference genome of M. truncatula accession A17 was accessed from NCBI under RefSeq identifier GCF_003473485.1. The GFF file was converted to GTF format using the Cufflinks v2.2.1 function ‘gffread’ [22]. Next, STAR v2.5.3 [23] was used to perform spliced transcript alignments to the Mt5.0 genome with the parameter ‘sjdbOverhang 149’ and the parameter ‘—sjdbGTFfile’ set to the Mt5.0 GTF file. STAR v2.5.3 [23] was run in ‘twopassMode Basic’ and default parameters. The alignment data was quality checked using MulitQC v1.14 [21]. Samtools v1.9 [24] was used to merge and filter RNA sequence alignment files to include paired RNA sequence reads with unique alignments. For A17, 94.1% of reads aligned with a mean of 56.4 million reads per sample. For HM078, 91.3% of reads aligned with a mean of 55.9 million reads per sample (Additional file 2: Table S1). HTSeq-count v0.11.0 [25] was used to quantify the number of reads that mapped to each exon. Finally, a feature count matrix was generated showing the number of RNA sequence reads for each tissue sample that mapped uniquely to exons of each gene.

Differential expression and pathway analysis

A count-based differential expression analysis was performed as previously described [26]. The EdgeR v3.36.0 pipeline [27] was used to conduct the differential expression analysis in R v4.1.2 [28]. First, a DGEList object was created from the feature count matrix and a model matrix object was created to store the experimental design variables for each sample. Next, the EdgeR v3.36.0 function ‘filterByExpr’ was used to determine which genes had adequate counts for statistical analysis across all samples. The function ‘calcNormFactors’ was used to determine the scaling factors to normalize counts based on library sizes, and the function ‘estimateDisp’ was used to evaluate common and tagwise dispersions. Then, the ‘glmQLFit’ function was used to fit the negative binomial generalized linear model (GLM) for each gene. Quasi-likelihood F-tests were performed for specified group comparisons using the ‘glmQLFTest’ function to calculate statistical differences in expression. For evaluating DEGs an absolute log2FC > 1 and an adjusted p-value (FDR) < 0.05 were used as cutoffs. A lower than standard log2FC cutoff was chosen due to relatively low numbers of DEGs in the resistant genotype. Statistical comparisons were made for each accession within each time point between mock and inoculated samples. DEGs were evaluated for enriched pathways using g:Profiler [29, 30]. Representative GO (Gene Ontology) terms were evaluated by Biological Processes (BP), Molecular Function (MF), Cellular Component (CC), and Kyoto Encyclopedia of Genes and Genomes (KEGG). Significantly enriched terms were retained if the adjusted p-value was below 0.05, or -log10(adjusted p-value) > 1.3.

qPCR validation of RNA-seq gene expression data

To validate the RNA-seq results we performed quantitative RT-PCR (qPCR) on a set of eight genes. RNA was extracted using the Qiagen RNeasy Mini Kit for Plants (Qiagen Inc., Valencia, CA, USA), and synthesis of cDNA was performed using SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA, USA). qPCR was performed using PerfeCta SYBR Green FastMix (Quanta BioSciences, Beverly Hills, California, USA) following the manufacturer’s recommendations. Amplification was performed for Nepenthesin (MtrunA17_Chr1g0187841), MtKCS12 (MtrunA17_Chr2g0327721), MtPP2C (MtrunA17_Chr3g0105371), MtEDS1L-like (MtrunA17_Chr3g0118251), MtCYP93C19 (MtrunA17_Chr4g0046331), MtIFR (MtrunA17_Chr5g0404481), MtLAC7 (MtrunA17_Chr7g0240991), and MtP21-like (MtrunA17_Chr8g0386341). Primer pairs are detailed in Additional file 2: Table S2. qPCR was performed in triplicate for one biological replicate from each genotype at each time point, and mean Ct values for each sample were used for calculations. Relative quantification compared to the reference gene MtACTIN11 (MtrunA17_Chr7g0223901) was calculated using 2^(-ΔΔCT) and (2^(-Δδlog2CPM)) for qPCR and RNA-seq data, respectively. Pearson’s correlation between fold change expression values was performed using R v4.1.2 in R studio v1.4.1717 [28].

Results

Microscopic evaluation reveals invasive hyphae penetrating leaf tissue samples

To establish optimal harvest time points to capture the host-pathogen interaction, infected leaf cross sections of resistant (HM078) and susceptible (A17) M. truncatula genotypes were examined at 24, 48, and 72 hpi (Fig. 1). Invasive hyphae penetrating epidermal cells in cross sections of leaf samples were observed at each time point, except for the resistant genotype at 24 hpi. Notably, fewer invasive hyphae were observed on the resistant genotype in all samples along with reduced fungal growth on the leaf surface (Additional file 1: Figure S1). Overall, we concluded that the selected harvest time points were sufficient to capture the host response.

Fig. 1
figure 1

Cross sections of M. truncatula leaves infected with A. medicaginicola. Images were taken under GFP fluorescence (left) and RGB (right) for susceptible genotype A17 at (A) 24 hpi, (B) 48 hpi, and (C) 72 hpi, as well as the resistant genotype HM078 at (D) 24 hpi, (E) 48 hpi, and (F) 72 hpi. Red arrows indicate invasive hyphae penetrating leaf epidermal cells. Scale bars for (A-F) are 75, 50, 150, 50, 150, and 150 micrometers, respectively

Full size image

Transcriptome analysis showed the largest differences in expression occurred at 72 hpi

Mock and inoculated M. truncatula leaf tissue of an SBS-resistant and SBS-susceptible genotype was collected at 24, 48, and 72 hpi for RNA sequencing. A total of 5.7 billion paired-end reads, with a mean library depth of 61.9 million reads per sample (Q > 30), were generated. After mapping reads to the Mt5.0 reference genome, 25,084 genes had sufficient expression for statistical analysis across all time points. Across all samples, log2 counts per million (CPM) values ranged from − 4.87 to 15.54, with a mean of 1.95 log2CPM. Principal component analysis (PCA) of all samples revealed that genotype was the primary separating factor, followed by hpi (Additional file 1: Figure S2A). For both genotypes, PCA showed significant separation between mock and inoculated at 72 hpi. For the resistant genotype at 24 hpi, mock and inoculated samples overlap and show little variation, which may indicate transcription shifts due to pathogen inoculation were delayed (Additional file 1: Figure S2B). For the resistant genotype at 48 hpi, there was variability in mock and inoculated samples. For the susceptible genotype at 48 hpi, a single mock-inoculated sample appears to be an outlier, as it overlaps with the inoculated replicates (Additional file 1: Figure S2C).

Differential expression analysis shows less response in the resistant genotype

Differential gene expression from mock and inoculated, resistant and susceptible leaf tissue from three time points was analyzed (Fig. 2). Relevant statistics for the differential expression results at each time point are summarized for both host genotypes (Additional file 2: Table S3). Gene names and functional annotations are included when available. The resistant genotype HM078 had a total of 192 DEGs, including up and downregulation, which increased over time with 15, 27, and 150 DEGs at 24, 48, and 72 hpi, respectively (Additional file 1: Figure S3) (Additional file 2: Table S4). The susceptible genotype A17 had a total of 2,908 DEGs, including up and downregulation, with 393, 17, and 2,498 DEGs at 24, 48, and 72 hpi, respectively. The number of DEGs detected fits with observations made in the PCA. For instance, the high variability between biological replicates at 48 hpi likely contributed to low numbers of DEGs detected at this time point.

Fig. 2
figure 2

Number of DEGs for resistant and susceptible M. truncatula in response to A. medicaginicola. Venn diagrams of (A) Upregulated DEGs of resistant genotype HM078, (B) Upregulated DEGs of susceptible genotype A17, (C) Downregulated DEGs of resistant genotype HM078, and (D) Downregulated DEGs of susceptible genotype A17

Full size image

Unique DEGs in the resistant genotype highlight potential genetic factors involved in disease resistance

The majority of the top ten most upregulated DEGs in the resistant genotype in response to pathogen infection were not differential expressed in the susceptible genotype. These included MtPBP1, MtPrx28, a MATH domain-containing protein, and a MtRPP13-like coiled-coil plant disease resistance protein (Table 1). Both calcium-binding protein MtPBP1 and peroxidase MtPrx28 are involved in ROS signaling and defense responses [31,32,33,34]. MATH domain-containing proteins have been shown to regulate NLR turnover in A. thaliana [35, 36]. The top ten most upregulated DEGs in the resistant genotype occurred at 72 hpi, while the top ten downregulated DEGs occurred at all time points. Notably, MtPBP1 had a 600-fold increase in expression between mock and inoculated samples at 72 hpi. Overall, the identified genes are potentially strong candidates for SBS disease resistance.

Table 1 The ten most upregulated and downregulated DEGs in resistant genotype HM078
Full size table

Upregulated DEGs in the resistant genotype across multiple time points were identified, and the majority were not differentially expressed in the susceptible genotype (Table 2). These included the JA biosynthesis gene linoleate 9 S-lipoxygenase, MtLOX1-5, were upregulated across all three time points [37, 38]. A member of the 3-ketoacyl-CoA synthase family, MtKCS12, as well as a receptor-like kinase of the RLK-Pelle-DLSV family, were upregulated at 48 and 72 hpi. Genes in the 3-ketoacyl-CoA synthase family perform biosynthesis of very long chain fatty acids (VLCFA) [39, 40]. Transmembrane RLKs can act to recognize apoplastic pathogen effectors [41, 42]. No DEGs were shown to be downregulated in HM078 across multiple time points, although several were downregulated at 48 hpi and later upregulated at 72 hpi (Table 2). Overall, DEGs unique to the resistant genotype in response to pathogen infection highlight potential candidates for disease resistance.

Table 2 DEGs in the resistant genotype across multiple time points
Full size table

Six of the top ten most downregulated DEGs in the resistant genotype were not differential expressed in the susceptible genotype during pathogen infection. These included genes like MtCTSH, MtPPR, and MtRPM1-like (Table 1). MtCTSH, a pro-cathepsin H, is a lysosomal cysteine protease, and while little is known about its role in plant disease, cathepsin B mediates PCD [43,44,45]. MtPPR is a pentatricopeptide repeat-containing protein, which are known to mediate post-transcriptional regulation [46]. In A. thaliana, the cleavage of a PPR protein has been associated with susceptibility to fungal disease, and these proteins share characteristics with plant disease resistance genes [47, 48]. MtRPM1-like, a coiled-coil plant disease resistance protein, was also downregulated in the resistant genotype. Interestingly, RPM1 is required for resistance to Pseudomonas syringae in Arabidopsis thaliana and Glycine max [49, 50]. Overall, DEGs that are uniquely downregulated in the resistant genotype highlight potential genetic factors involved in the host response.

Unique DEGs in the susceptible genotype provide insight into the compatible host response

When evaluating candidate genes for disease resistance from transcriptome data, it is beneficial to compare expression levels between contrasting host genotypes. Furthermore, analyzing DEGs in the susceptible genotype can provide insight into the compatible plant immune response to a necrotrophic fungus. In the susceptible genotype, the top ten most upregulated DEGs included MtMYB and MtCHS-1 A (Table 3). The MYB transcription factor family is involved in regulating a variety of stress responses, while chalcone synthases (CHS) are a vital component of flavonoid biosynthesis. The most downregulated DEG in the susceptible genotype was a major facilitator superfamily (MFS) transporter, which transport a variety of substrates across membranes (Table 3). Only three DEGs were upregulated across all time points in the susceptible genotype; cytochrome P450 monooxygenase MtCYP76X2, chalcone synthase MtCHS-1 A, and alcohol dehydrogenase MtADH6. Cytochrome P450s conduct NADPH or O2 dependent hydroxylation, while alcohol dehydrogenase oxidizes ethanol. An additional 67 DEGs were upregulated across multiple time points (Additional file 2: Table S5). Overall, the majority of DEGs in the susceptible genotype were not shared by the resistant genotype, and reveal a drastically different host response.

Table 3 The ten most upregulated and downregulated DEGs in susceptible genotype A17
Full size table

Shared DEGs by both host genotypes include a variety of transcription factors

A total of 65 genes were differentially expressed in both the resistant and susceptible genotypes (Additional file 2: Table S6). The top upregulated DEG in both genotypes was RNA-binding protein ARP1. A variety of transcription factor gene families were among the shared DEGs, which included C2H2, AS2-LOB, Calmodulin-binding, Homeobox-WOX, WD40, WRKY, C2C2-Dof, and AP2-EREBP. Ten TFs were differentially expressed in both genotypes; seven were upregulated and two were downregulated. A WD40-type strictosidine synthase-like 10 transcription factor, MtSTR10-like, was downregulated in the susceptible genotype while upregulated in the resistant genotype. Strictosidine synthases have been implicated in terpenoid biosynthesis of phytoalexins [51, 52]. The resistant genotype showed higher upregulation for six of the seven upregulated transcription factors. For example, the C2H2 transcription factor MtZAT11 had a 36-fold increase in the resistant genotype, but only a 6-fold increase in the susceptible genotype at 72 hpi. C2H2 transcription factors have been shown to be involved in regulation of hormone pathways in response to biotic stress [53, 54]. Finally, for the two downregulated TFs, the resistant genotype showed less downregulation.

Functional enrichment analysis provides insight into contrasting host responses

Functional enrichment analysis of DEGs revealed that the resistant and susceptible genotypes activate distinct pathways. We examined upregulated and downregulated DEGs across all time points for each genotype, and reported significantly enriched terms (Fig. 3). In the resistant genotype, the most significant cellular components included ‘cell wall’, ‘external encapsulating structure’, and ‘extracellular region’ (Fig. 3A). The most significant biological processes were ‘cell wall organization’ and ‘external encapsulating structure or organization’. The most significant molecular function was ‘calcium ion binding’. Overall, due to fewer DEGs, the functional enrichment analysis in the resistant genotype was limited. For instance, KEGG enrichment analysis resulted in one significantly term, ‘Plant-pathogen interaction’.

Fig. 3
figure 3

Functional enrichment analysis of resistant and susceptible M. truncatula in response to A. medicaginicola. Significantly enriched GO terms were analyzed for (A) DEGs in the resistant genotype HM078, and (B) DEGs in the susceptible genotype A17. Upregulated and downregulated DEGs across all time points were included for each genotype. GO (Gene Ontology) terms were grouped by Biological Processes (BP), Molecular Function (MF), Cellular Component (CC), or Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways

Full size image

For the susceptible genotype, the most significant cellular components were ‘cell periphery’ and ’plasma membrane’ (Fig. 3B). The most significant biological processes were ‘secondary metabolite biosynthetic process’ and ‘flavonoid biosynthetic process’. The most significant molecular function was ‘oxidoreductase activity’. KEGG enrichment analysis showed engagement in ‘Biosynthesis of secondary metabolites’, ‘Metabolic pathways’, and ‘Flavonoid biosynthesis’.

Regulation of hormone pathways in response to A. Medicaginicola

The SA and JA pathways are crucial for plant immune responses. A previous study has shown that M. truncatula activates these pathways in response to A. medicaginicola [8]. In the compatible interaction with the susceptible genotype, there is a greater induction in the SA pathways, whereas the resistant genotype shows a rapid induction of the JA pathway [8]. Expression profiles for genes contributing to SA and JA biosynthesis and signaling were visualized in a heatmap (Additional file 1: Figure S4). Across all three time points, the resistant genotype upregulated the JA biosynthesis gene MtLOX1-5 (Table 4). MtLOX1-5 was highlighted earlier for being one of a few DEGs upregulated in the resistant genotype over multiple time points, although the susceptible genotype had higher constitutive expression. At 72 hpi, the susceptible genotype upregulated SA biosynthesis and signaling genes, such as numerous PR proteins, as well as genes in the JA pathway. Interestingly, isoflavone reductase (IFR) and phenylalanine ammonia lyase (PAL) were upregulated in the susceptible genotype, while the resistant genotype had higher constitutive expression in mock-inoculated samples (Additional file 1: Figure S5). These genes are known to participate in isoflavonoid biosynthesis of anti-fungal phytoalexins [55, 56].

Table 4 Differentially expressed genes in SA and JA pathways for the resistant and susceptible genotype
Full size table

Contrasting expression profiles in QTL provide candidate genes for disease resistance

QTL regions rnpm1 and rnpm2 described by Kamphuis et al. [7] for SBS disease resistance were examined for differentially expressed genes. We examined the expression of 130 genes within a 1 Mbp region across rnpm1, and 69 genes across approximately 440 kbp in the fine-mapped rnpm2 region [57]. Overall, differential expression of genes in these regions was only identified in the susceptible genotype (Additional file 2: Table S7). While no differential expression was observed across the QTL in the resistant genotype, there were genes with contrasting expression profiles between the resistant and susceptible genotypes (Fig. 4). In rnpm1, these include a Toll/Interleukin1 receptor-nucleotide binding site-leucine-rich repeat (TIR-NBS-LRR) disease resistance protein (MtrunA17_Chr4g0008981), which is constitutively expressed in HM078 at much higher levels than observed in A17. A Blast2GO annotation shows this gene has high similarity (78.79%) to the disease resistance protein RPS6 isoform X1. Conversely, TIR-NBS-LRRs (MtrunA17_Chr4g0009001, MtrunA17_Chr4g0009011) were expressed in A17 but had little to no expression in HM078. Finally, in rnpm2, a PAM16-like (MtrunA17_Chr4g0064871) gene was identified as having a contrasting expression profile between the resistant and susceptible genotype. Overall, genes with contrasting expression in the QTL regions for disease resistance may point to structural variation or transcriptional repression, and warrant further investigation.

Fig. 4
figure 4

Gene expression profiles for QTL regions. Heatmaps are displayed in log2CPM for QTL (A) rnpm1 and (B) rnpm2. Genes with contrasting expression profiles between resistant and susceptible genotypes are outlined with a box. Differentially expressed genes in specific tissues are indicated with asterisks. Sample ID abbreviations are SM: susceptible mock-inoculated, SI: susceptible inoculated, RM: resistant mock-inoculated, RI: resistant inoculated, followed by hours post inoculation (24, 48, or 72 hpi)

Full size image

Plant immune system receptors feature notable candidate genes for disease resistance

We examined RLKs among the DEGs of both genotypes (Additional file 2: Table S8) (Additional file 1: Figure S6). For the resistant genotype, these included nine RLKs from five classes (Table 5). Of these, an RLK-Pelle-LRR was the most upregulated, and an RLK-Pelle-DLSV was upregulated over time. Neither were differentially expressed in the susceptible genotype. In the susceptible genotype, there were 166 differentially expressed RLK genes, and only three shared between both genotypes (Additional file 2: Table S8).

Table 5 Differentially expressed RLKs in the resistant genotype
Full size table

Among the DEGs, we evaluated plant disease resistance genes (Additional file 1: Figure S7). In the resistant genotype, these included a TIR-NBS-LRR and three coiled-coil NLRs (Table 6). Notably, a CC-NBS plant disease resistance gene had the highest upregulation, with approximately a 75-fold increase in expression. This coiled-coil NLR gene has homology to RPP13-like protein 1 in Pisum sativum (amino acid identity = 73.35%), which lacks an LRR domain. A total of 64 plant disease resistance proteins were differentially expressed in the susceptible genotype, and none were shared between both genotypes (Additional file 2: Table S9).

Table 6 Differentially expressed NLR plant disease resistance genes in the resistant genotype
Full size table

Selection of candidate genes for SBS disease resistance

After evaluating differential expression in a resistant and susceptible genotype of M. truncatula in response to A. medicaginicola, candidate genes for SBS disease resistance were selected (Table 7). Genes were chosen based on the following criteria: among the top ten upregulated or downregulated genes in the resistant genotype, upregulation across multiple time points in the resistant genotype, differentially expressed in the susceptible genotype with higher constitutive expression in the resistant genotype, shared DEGs between both genotypes with different expression levels, genes that had contrasting expression profiles in QTL regions, uniquely upregulated RLKs or NLRs in the resistant genotype, and upregulated genes in functionally enriched pathways. Finally, candidate genes that have not been linked to plant disease resistance in the literature were excluded. The expression profiles of candidate genes was visualized in a heatmap (Additional file 1: Figure S8).

Table 7 Candidate genes for SBS disease resistance in M. truncatula HM078
Full size table

Validation of transcriptome sequencing data with qPCR

To validate our transcriptome sequencing results we performed qPCR on cDNA synthesized from the RNA samples collected throughout our experiment. We selected eight genes for validation across all time points for both genotypes. Relative quantification of target genes was performed compared to MtACTIN11, and RNA-seq fold change had a significant positive correlation (R = 0.8–0.98) with qPCR fold change for each gene tested (Additional file 1: Figure S9).

Discussion

In this study, we analyzed the transcriptomes of both resistant and susceptible M. truncatula accessions in response to inoculation with the necrotrophic pathogen A. medicaginicola at three time points. We observed an approximate 24-hour delay in fungal penetration of the resistant genotype, possibly due to physical barriers or antimicrobial compounds [58, 59]. Invasive hyphae were more difficult to identify at earlier time points on both genotypes, likely resulting in less dramatic transcriptional shifts in whole-leaf samples and overlap between mock and inoculated samples observed on PCA plots at 24 and 48 hpi. Variability between individual plants and uneven inoculation application may have contributed to dispersal between biological replicates, particularly at 48 hpi, reducing statistical power and limiting our ability to detect differential expression. PCA analysis showed the largest degree of separation between mock and inoculated samples at 72 hpi, which corresponded to the highest numbers of DEGs for both genotypes. Similar studies have also noted large differences in DEGs between host genotypes in response to fungal pathogens, likely due to the number of invaded cells being sampled, highlighting a limitation of RNA-seq experiments [60].

Transcriptomics of ascochyta blights reveal similar findings for several legume species

Ascochyta blights of multiple legume species have been previously investigated using similar RNA-seq studies [8, 61,62,63]. In M. truncatula, numerous PR proteins, as well as SA and JA hormone pathway genes were found to be upregulated at 12 hpi [8]. We identified upregulation of PR proteins, but primarily in the susceptible genotype. RNA-seq studies of ascochyta blight of chickpea, Cicer arietinum, have been conducted, and found that RLKs, NBS-NLRs, transcription factors, as well as SA, JA, ET, and ABA hormone pathway genes likely contribute to disease resistance [62, 63]. In our study, we found upregulation of genes in these families, including numerous transcription factor families, as well as unique NBS-NLRs and RLKs in the resistant genotype. Finally, Ascochyta blight of grass pea, Lathyrus sativus, was investigated using transcriptomics, which described candidate genes for disease resistance in the SA, ET, and ABA hormone pathways, as well as cell wall remodeling genes, peroxidase, PR proteins, and detoxifying genes [61]. Our findings align with previous studies in relation to the functional description of many candidate genes for disease resistance. Both cell wall remodeling and ROS response genes were found to be important factors for resistance to A. medicaginicola. However, we also identified candidate genes that were not highlighted in previous studies, such as calcium-binding protein MtPBP1 and VLCFA synthesis gene MtKCS12.

Functional enrichment in the resistant genotype suggest antifungal mechanisms

Functional enrichment analysis of DEGs revealed differences in the host response of resistant and susceptible genotypes to A. medicaginicola. In the resistant genotype, significantly enriched terms related to the extracellular region, cell wall, and calcium ion binding, which were all unique to the resistant genotype. Upregulated DEGs in extracellular region included peroxidase 28 (Prx28) and ribonuclease T2 (RNASET2). Prx28 regulates redox signaling pathways for defense responses including cell wall thickening and PCD [33, 34]. RNASET2 is thought to inhibit pathogen colonization at infection sites [64, 65]. Upregulated DEGs in the cell wall included xyloglucan/xyloglucosyl transferase (XET), β-galactosidase (BGAL), and ascorbate oxidase (AAO). XET cross-links xyloglucans to strengthen the cell wall [66,67,68]. BGAL hydrolyze β-galactosides to modify the cell wall [69, 70]. AAO produces ROS that mediate defense signaling [71, 72]. In the resistant genotype, the most upregulated DEG overall was PINOID-BINDING PROTEIN 1 (PBP1), which similar to calmodulin (CML), contains domains that bind calcium ions [73, 74]. Calcium ion elevations activate MAPK signaling cascades, the oxidative burst, and the hypersensitive response [75,76,77]. Resistant M. truncatula genotype HM078 has been observed to have a hypersensitive-like response after inoculation with A. medicaginicola that could be attributed to an oxidative burst [7]. Overall, the function of genes in these pathways shed light on potential antifungal mechanisms in the resistant genotype.

Plant hormone pathways engaged during the host response to A. medicaginicola

Plant hormones, such as SA and JA, are crucial signaling molecules to regulate defense responses [78]. JA is synthesized from fatty acids in the octadecanoid pathway by enzymes including OPDA reductase, lipoxygenase, allene oxide synthase, and lipase [79]. JA signaling mediates defense responses against necrotrophic fungi, resulting in lignin formation, synthesis of PR proteins, flavonoids, terpenoids, and phytoalexins [80]. In the resistant genotype, MtLOX1-5, was upregulated from 24 to 72 hpi, while the susceptible genotype, upregulated JA biosynthesis genes at 72 hpi. On the other hand, SA is synthesized from phenylalanine in the phenylpropanoid pathway by a series of enzymes resulting in isoflavonoid phytoalexins, lignin, benzoic acid, phenylpropenes, and coumarins [56]. Chalcone synthase (CHS), isoflavone reductase (IFR), 4-coumarate-CoA ligase (4CL), and phenylalanine ammonia lyase (PAL) mediate flavonoid biosynthesis. In the susceptible genotype, phenylpropanoid pathway genes were enriched, including MtCHS-1 A, MtCHS-1B, MtIFR, MtPAL, and Mt4CL-2. Kamphuis et al. [8] found an induction of SA in resistant and susceptible genotypes, but found the resistant genotype HM078 contained constitutively higher levels of isoflavonoids. This was supported by our finding that MtPAL and MtIFR were upregulated at 72 hpi in the susceptible genotype, however, the resistant genotype had higher constitutive expression. Overall, both SA and JA likely play crucial roles as signaling molecules and the host response to A. medicaginicola.

Candidate genes in QTL identified based on contrasting expression profiles

QTL rnpm1 and rnpm2 were examined for their role in SBS disease resistance, focusing on gene expression patterns. DEGs were only detected at 72 hpi in the susceptible genotype. These QTL are known to be inherited recessively, which may support the inverse gene-for-gene model [7]. This paradigm is illustrated by the LOV1 gene in oat that confers sensitivity to the fungal toxin victorin, while also providing resistance to the crown rust fungus [15]. Genes expressed in the susceptible genotype and not expressed in the resistant genotype are of particular interest. In rnpm1, TIR-NBS-NLR genes MtrunA17_Chr4g0009001 and MtrunA17_Chr4g0009011 were found to be expressed only in the susceptible genotype, supporting this concept. In rnpm2, which does not contain NLRs, a gene orthologous to AtPAM16 showed no expression in the resistant genotype and high expression in the susceptible genotype. PAM16 is known to play a role in plant immunity, as shown in A. thaliana mutants lacking this gene, which exhibit enhanced disease resistance [81, 82]. Backcrossing studies with an AtPam16 knockout mutant (muse5-1) indicated that the recessive inheritance of the resistant phenotype aligns with the recessive inheritance pattern observed for rnpm2. Overall, genes with differing expression patterns between the resistant and susceptible genotypes in these QTL regions are candidate genes for further study.

RPP13-like plant disease protein likely involved in incompatible host response.

A promising plant disease resistance gene, potentially involved in the incompatible host response was identified. This coiled-coil class NLR gene was uniquely upregulated in the resistant genotype. Notably, this disease resistant protein is similar to RPP13-like protein 1, which confers broad spectrum resistance to biotrophic pathogens Melampsora lini (flax rust), as well as Hyaloperonospora arabidopsidis and Peronospora parasitica (downy mildew) in A. thaliana [83, 84]. Coiled-coil class NLRs have also been shown to play important roles in plant immunity to necrotrophic pathogens, for example, overexpression of GbCNL130 confers resistance to Verticillium dahliae in cotton [85]. Pathogen ligand binding to these proteins results in downstream defense responses. For instance, the activation of HOPZ-ACTIVATED RESISTANCE 1 (ZAR1) causes a calcium ion influx, ROS production, and cell death conferring resistance to Pseudomonas syringae in A. thaliana [86, 87]. Future directions will include investigating the specific role of this coiled-coil class NLR.

Conclusion

Examining host-pathogen interactions between M. truncatula and the necrotrophic fungal pathogen A. medicaginicola has the potential to illuminate molecular factors that could be used to enhance disease resistance to Ascochyta blights in legumes. We performed a transcriptome analysis for a resistant (HM078) and susceptible (A17) M. truncatula genotype infected with A. medicaginicola to evaluate the host response and identify candidate genes for disease resistance. We examined DEGs, functionally enriched pathways, hormone pathways, RLKs, NLRs, and QTL regions for SBS disease resistance. We identified a number of candidate genes for disease resistance with support from the literature. After functional validation of candidate genes, future studies will explore engineering SBS disease resistance in the economically important forage crop alfalfa.