The Parastagonospora nodorum necrotrophic effector SnTox5 targets the wheat gene Snn5 and facilitates entry into the leaf mesophyll.

Parastagonospora nodorum is an economically important necrotrophic fungal pathogen of wheat. Parastagonospora nodorum secretes necrotrophic effectors that target wheat susceptibility genes to induce programmed cell death (PCD). In this study, we cloned and functionally validated SnTox5 and characterized its role in pathogenesis. We used whole genome sequencing, genome-wide association study (GWAS) mapping, CRISPR-Cas9-based gene disruption, gain-of-function transformation, quantitative trait locus (QTL) analysis, haplotype and isoform analysis, protein modeling, quantitative PCR, and laser confocal microscopy to validate SnTox5 and functionally characterize SnTox5. SnTox5 is a mature 16.26 kDa protein with high structural similarity to SnTox3. Wild-type and mutant P. nodorum strains and wheat genotypes of SnTox5 and Snn5, respectively, were used to show that SnTox5 not only targets Snn5 to induce PCD but also facilitates the colonization of the mesophyll layer even in the absence of Snn5. Here we show that SnTox5 facilitates the efficient colonization of the mesophyll tissue and elicits PCD specific to host lines carrying Snn5. The homology to SnTox3 and the ability of SnTox5 to facilitate the colonizing of the mesophyll also suggest a role in the suppression of host defense before PCD induction. © No claim to US Government works New Phytologist

Introduction

Plant pathogenic fungi secrete effectors that contribute to virulence during host infection. These effectors are secreted into the apoplast or internalized into the cytoplasm where they manipulate the host cell’s biological processes to promote colonization (Lo Presti et al., 2015; Franceschetti et al., 2017). However, plants have evolved plant innate immunity including resistance (R) receptors that recognize effectors, resulting in effector‐triggered immunity (ETI) and pattern recognition receptors (PRRs) that recognize pathogen‐associated molecular patterns (PAMPs), resulting in PAMP‐triggered immunity (PTI) (Jones & Dangl, 2006). Typically, the resistance response results in programmed cell death (PCD) surrounding the infection site via a localized hypersensitive response (HR) that occurs along with, or as a result of, physiological processes, including accumulation of reactive oxygen species (ROS), lipid peroxidation, ion fluxes and deposition of callose on infection sites (Dodds & Rathjen, 2010; Balint‐Kurti, 2019).

Localized PCD is highly effective against biotrophic fungal pathogens because biotrophs typically require a living cell to extract nutrients. The PCD response is less effective against necrotrophic pathogens because they thrive and acquire nutrients made available by cell death. Necrotrophic fungal pathogens release necrotrophic effectors, a group of effectors that target host genes to trigger necrosis, providing nutrients to the pathogen (Friesen & Faris, 2010; Faris & Friesen, 2020). Unlike classical gene‐for‐gene interactions described for biotrophic pathosystems (Flor, 1971), necrotrophic interactions result in necrotrophic effector‐triggered susceptibility (Liu et al., 2009). In the past two decades, several necrotrophic effector–host susceptibility gene interactions have been described, including those found in the Parastagonospora nodorum–wheat pathosystems (Faris & Friesen, 2020).

Parastagonospora nodorum is a necrotrophic fungal pathogen of wheat that causes septoria nodorum blotch (SNB). Extensive research over the last two decades has shown that P. nodorum deploys several proteinaceous necrotrophic effectors during the infection process. To date, nine such interactions have been identified, including SnToxA–Tsn1 (Friesen et al., 2006; Faris et al., 2010), SnTox1‐Snn1 (Liu et al., 2004; Shi et al., 2016), SnTox2‐Snn2 (Friesen et al., 2007), SnTox3‐Snn3‐B1 (Friesen et al., 2008), SnTox3‐Snn3‐D1 (Zhang et al., 2011; Zhang et al., 2021), SnTox4‐Snn4 (Abeysekara et al., 2009), SnTox5‐Snn5 (Friesen et al., 2012), SnTox6‐Snn6 (Gao et al., 2015) and SnTox7‐Snn7 (Shi et al., 2015). Even though SnTox2, SnTox6 and SnTox7 were initially hypothesized to be three different effectors that targeted three separate sensitivity genes, Richards et al. (2022) recently showed that these three effectors were in fact the same protein, which was designated as SnTox267. Owing to the number of characterized interactions, the wheat–P. nodorum system is recognized as a model for the study of necrotrophic specialist pathogens (Oliver et al., 2012; Faris & Friesen, 2020).

Currently four P. nodorum necrotrophic effector genes have been cloned and functionally characterized, including SnToxA, SnTox1, SnTox267 and SnTox3 (Friesen & Faris 2021). SnToxA is nearly identical to the ToxA gene found in Pyrenophora tritici‐repentis and Bipolaris sorokiniana and it encodes a 13.2 kDa mature protein that targets Tsn1 indirectly to cause necrosis (Ciuffetti et al., 1997; Friesen et al., 2006; McDonald et al., 2018). SnTox3 was the second P. nodorum necrotrophic effector gene cloned (Liu et al., 2009), encoding for a mature 17.5 kDa protein that targeted Snn3‐B1 (Friesen et al., 2008) and Snn3‐D1 (Zhang et al., 2011; Zhang et al., 2021) to induce necrosis. SnTox3 also functioned to suppress the host defense through a direct interaction with TaPR1 proteins (Breen et al., 2016; Sung et al., 2021). SnTox1 encoded a 10.3 kDa protein that targeted Snn1 directly (Shi et al., 2016) to trigger an oxidative burst, upregulation of PR genes and DNA laddering (Liu et al., 2012). Liu et al. (2016) showed that in addition to targeting Snn1, SnTox1 had the ability to bind chitin and protect the fungal cell wall from wheat chitinases. The most recent necrotrophic effector gene that was cloned in this system was SnTox267, which encoded for a 27.4 kDa mature protein that targeted three sensitivity genes in two different pathways to induce necrosis (Richards et al., 2022).

SnTox5 is also a proteinaceous necrotrophic effector and targets the susceptibility gene Snn5 (Friesen et al., 2012). Snn5 was mapped to chromosome 4B in the Lebsock×PI94749 (LP749) population by using culture filtrates of the SnTox5‐producing P. nodorum isolate Sn2000 (Friesen et al., 2012). Here we used whole genome sequencing of 197 P. nodorum isolates in a genome‐wide association study (GWAS) to identify, clone and functionally validate the SnTox5 gene from P. nodorum isolate Sn2000. We also used laser confocal microscopy to characterize the role that SnTox5 plays in P. nodorum leaf colonization. This study provides further characterization of how P. nodorum is using its arsenal of necrotrophic effectors to target the host defense pathways to complete its pathogenic life cycle.

Materials and Methods

Genome‐wide association study (GWAS) analysis

Phenotypic and genotypic data for GWAS analysis were generated using 197 P. nodorum isolates as described by Richards et al. (2019). Details for phenotyping and genotyping are included in Supporting Information Methods S1 and S2. The GWAS analysis was performed to identify significant marker–trait associations as discussed in Methods S3.

Identification of candidate genes

The candidate region for SnTox5 was screened for genes encoding small secreted proteins using signalp v.4.1 (Petersen et al., 2011) and effectorp v.1.0 (Sperschneider et al., 2016). A gene encoding a small secreted protein that contained the marker with the most significant marker–trait association from GWAS analysis was considered the top candidate for SnTox5.

Deletion of SnTox5 in the virulent isolate Sn2000

The disruption of SnTox5 was carried out using a CRISPR‐Cas9 ribonucleoprotein‐mediated technique as described in Foster et al. (2018). In brief, the FASTA sequence of SnTox5 in Sn2000 was input into E‐CRISP (http://www.e‐crisp.org/E‐CRISP/) to select the primer template for the small guide RNA (sgRNA). Oligonucleotides were purchased from Eurofins Genomics (Louiville, KY, USA). The sgRNA was synthesized using the sgRNA synthesis kit NEB#E3322 from New England Biolabs following the manufacturer’s protocol. The resulting sgRNA was purified before complexing with Cas9‐NLS via the RNA clean and concentrator‐25 kit from Zymo Research (Irvine, CA, USA) following the manufacturer’s instructions.

Primers, Tox5HygDonor F1 and Tox5HygDonor R1 (Table S1) were designed to amplify the complete hygromycin resistance gene as the donor DNA. Each primer consisted of a 40 bp sequence homologous to the flanking region adjacent to the protospacer adjacent motif (PAM) site and 3 bp upstream of the PAM site that is incorporated into the ends of the hygromycin resistance gene, cpc‐1:hyg, which was amplified from the pDAN vector (Liu et al., 2012) as the template.

Fungal protoplast generation and transformation were performed as described in Liu & Friesen (2012). Cas9‐NLS were complexed with sgRNAs and then mixed with the donor DNA that was transformed into protoplasts of Sn2000. Protoplasts were plated on regeneration medium agar supplemented with hygromycin B. Regenerated colonies were picked and screened for SnTox5 disruption using the primers SnTox5_pENTR_F1_bac and SnTox5_pENTR_R1 (Table S1) and for the presence of the hygromycin resistance gene. Two SnTox5‐disrupted strains and one strain with an ectopic insertion of the hygromycin resistance gene were used for downstream phenotypic and quantitative trait locus (QTL) analysis on the LP29 and LP749 populations, respectively, as described in Methods S4.

Expression of SnTox5 in the avirulent P. nodorum isolate Sn79‐1087

A construct consisting of SnTox5 and its promoter region (Fig. S1) was developed as described in Methods S5. The SnTox5 construct was linearized using PmeI and transformed into Sn79‐1087 as described in Liu & Friesen (2012). Transformants were screened for the presence of the gene through PCR amplification using primers SnTox5_PENTR_F and SnTox5_PENTR_R (Table S1). The gain‐of‐function transformant, Sn79+Tox5‐3, and wild‐type Sn79‐1087 were used in further experiments, as described in Methods S6.

Homology between SnTox3 and SnTox5

Protein blast was performed against the NCBI nonredundant protein sequence database using the amino acid sequence of SnTox5 as the query. The structural homology and pairwise alignment between SnTox3 and SnTox5 was performed using phyre2 (Kelley et al., 2015) and ucsf chimera v,1.15 (Pettersen et al., 2004). In addition, disulfide bridges in SnTox5 that formed between cysteine residues were predicted using the web‐based application dianna1.1 (http://clavius.bc.edu/˜clotelab/DiANNA/).

Population genetics and haplotype analysis of SnTox5

The genome sequences of SnTox5 for each isolate of the natural population were assembled as described in Methods S7. Isolates with coverage ≤ 50% were considered to lack the gene. Genomic sequences of SnTox5 for isolates that contained complete coverage were then imported into dnasp v.6 for population genetic analysis. The predicted SnTox5 amino acid sequences for each haplotype were aligned using the web‐based sequence aligner MultAline (Corpet, 1988) to identify amino acid sequence variation in SnTox5.

Statistical analysis of variation in disease caused by different isoforms of SnTox5

Isolates harboring 14 active isoforms were grouped based on the amino acid residues at the 155th and 156th positions of SnTox5, and ANOVA and least significant differences (LSDs) were calculated for each group based on the average disease reaction on LP29. To account for the unbalanced sample size, a type III ANOVA was calculated using the package car v.3.0‐10 (Fox & Weisberg, 2019) and the resulting sum of ssquare was used to calculate the LSDs at 0the .05 experimental significance level using the package agricolae v.1.33 (De Mendiburu, 2009) in the R programming environment in both analyses.

Temporal expression profile of SnTox5 during the infection process

Secondary leaves of 14‐d‐old ‘Lebsock’ were inoculated with Sn2000 and three samples of leaf tissue were collected at 4, 12, 24, 48, 72, 96 and 120 h post‐inoculation (hpi). Total RNA was extracted from leaves and purified using the RNeasy Plant Mini Kit (Qiagen) according to the manufacturer’s protocol. RNA was quantified using a Qubit and 300 ng of total RNA from each time point was used to develop cDNA using the GoScript™ Reverse Transcription System (Promega). With the use of gene‐specific primers SnTox5_qPCR_F and SnTox5_qPCR_R, quantitative PCR (qPCR) was performed for all time points with three biological and three technical replicates. The P. nodorum actin gene was amplified as the internal control using previously published primers (Liu et al., 2009) (Table S1).

Laser confocal microscopic analysis of the infection process involving the SnTox5‐Snn5 interaction

A construct with the mCherry gene coupled to the promoter of SnTox1 was cloned into the pFPL‐Cg vector containing a geneticin resistance cassette using the Gateway cloning system (Gong et al., 2015). The construct was linearized using PmeI and concentrated to 1 µg µl−1. Parastagonospora nodorum strains Sn2000, Sn2kΔTox5‐15, Sn79+Tox5‐3 and Sn79‐1087 were transformed with the linearized construct, as explained in Liu & Friesen (2012), and transformants were inoculated onto 2‐wk‐old plants of the SnTox5 differential wheat line LP29. In addition, Lebsock was inoculated with P. nodorum strains Sn2000 and Sn2kΔTox5‐15. Two leaf samples were collected at 4, 12, 24, 48, 72, 96 and 120 hpi and a 2.5‐cm‐long cutting from the middle portion of each secondary leaf was placed on a glass slide as described in Solanki et al. (2019). Ecomount mounting media (Biocare Medical, Pacheco, CA, USA) was applied to the sample and a coverslip was placed on the sample without introducing any air bubbles. The slides were dried overnight under room temperature before preservation at 2–8°C. Three such replicates were conducted for each isolate–wheat line combination.

All the prepared slides were observed under an LSM700 laser scanning confocal microscope (Zeiss Thronwood, Thronwood, NY, USA) using ×20 and ×40 objectives where images captured under ×20 were used for calculation of fungal volume, and images captured under ×40 were used to characterize the features of the infection process. Two different channels were used where the red channel (Ex555/Em 639 nm) was assigned to capture the fluorescence emitted by mCherry and the green channel (Ex488/Em555 nm) was used for autofluorescent detection of the leaf structure. To observe the infection process in different tissues within the leaf, Z‐stack images were taken at different depths of the leaf from the upper to the lower epidermis with the use of zen v.11 (Zeiss Thronwood). The two‐dimensional images were processed for microscopic characterization and three ‐dimensional images were reconstructed for the volume analysis using imaris v.9.6. Animated figures were created using the web‐based application biorender (BioRender.com). Average fungal volume was calculated after constructing the surface of the inoculated wheat leaf sample as at a minimum of two infection sites per sample at each time point in one experimental replicate. Three such replicates were conducted, meaning the average volume was calculated for six infection sites per time point under the ×20 objective lens.

Microscopic analysis of infection caused by Sn2000 and Sn2kΔTox5‐15 on Snn5 mutants of LP29

Mutants of Snn5 were developed by treating LP29 seed with 0.25% ethyl methanesulfonate (EMS) (Methods S8). Red fluorescent protein‐tagged Sn2000 and Sn2kΔTox5‐15 were inoculated onto LP29 and LP29Δsnn5 as described earlier, and samples were collected at 120 hpi. The samples were used in microscopic analysis and the volume analysis was done as described in the previous section.

Results

GWAS identifies an SnTox5 candidate gene

To identify candidate genes for SnTox5, an association mapping approach was used to detect significant marker–trait associations in the P. nodorum genome using a natural population of 197 P. nodorum isolates. The genotypic data for the GWAS analysis was generated by aligning the whole genome shotgun sequence reads of 197 P. nodorum isolates to the SnTox5‐producing Sn2000 reference genome (Richards et al., 2018). A total of 402 601 high confidence markers were used in GWAS analysis. The 197 P. nodorum isolates were phenotyped on LP29, where the average disease reaction ranged from 0 to 4.33 (Table S2). The GWAS analysis was performed using both Gapit and Tassel v.5 and the most significant marker trait association was identified for a single nucleotide polymorphism (P = 6.71E–11) on chromosome 8 at the 53 300 bp position that resided in the gene Sn2000_06735 and therefore this gene became our top SnTox5 candidate (Fig. 1).

Fig. 1

Genome‐wide association mapping (GWAS) analysis of SnTox5. (a) Manhattan plot of the GWAS performed using the phenotypic data from each isolates of the United States Parastagonospora nodorum population on differential line LP29. The x‐axis labels denote chromosome numbers of the P. nodorum genome and the y‐axis represents the –log10 transformation of the P‐value for significance of the marker–trait association. The horizontal dotted line represents the Bonferroni significance threshold at the 0.05 level of probability. (b) The genomic location of SnTox5 using the Sn2000 genome sequence as a reference. (c). Amino acid sequence of SnTox5 from Sn2000. Purple and red double‐headed arrows represent the predicted signal peptide and the pro‐sequence, respectively. The green line represents the putative Kex2 protease site and the asterisk (*) represents the stop codon.

The Sn2000_06735 gene spanned from 53 219 to 53 872 bp on chromosome 8 of the Sn2000 genome (Fig. 1) and was a 654 bp intron‐free gene with a putative TATA box 171 bps upstream of the start site (Fig. S1). The gene encoded a predicted small secreted protein consisting of 217 amino acids, with the first 22 amino acids predicted to be a signal peptide. A putative Kex2 protease site was identified from the 68th to the 71st amino acid, marking a putative 49‐amino‐acid pro‐sequence following the signal peptide. Sn2000_06735 also contained six cysteine residues predicted to form three disulfide bridges (Fig. 1). blastp analysis against the NCBI database showed that Sn2000_06735 had 45.13% homology to SnTox3. Pairwise alignment between the two mature protein sequences showed that the positions of the six cysteine residues were conserved, and phyre2 used the crystal structure of SnTox3 (Outram et al., 2020) as the best template to model 98% of the mature protein structure of Sn2000_06735 with 100% confidence, showing that Sn2000_06735 had high structural homology to SnTox3 (Fig. 2). The secondary structure predicted by phyre2 showed the mature Sn2000_06735 protein consisted of an α‐helix and 11 β‐strands, of which eight were predicted to form a β‐barrel structure like SnTox3 (Figs 2, S2). The α‐helix was predicted at a relatively low confidence level and was not predicted by Chimera (Fig. S2).

Fig. 2

Structural homology between SnTox5 and SnTox3. (a) The three‐dimensional (3D) structure of mature SnTox5 predicted by Phyer2 using the crystal structure of SnTox3 as a template. Arrows represent the β‐sheets. (b) The predicted 3D structure of mature SnTox5 (Pink) was superimposed on the crystal structure of the mature SnTox3 (Blue). The majority of the two proteins showed structural homology. (c) Pairwise alignment of two mature proteins.

Deletion of Sn2000_06735 converts virulence to avirulence in the presence of Snn5

To validate Sn2000_06735 as SnTox5, Sn2000_06735 was disrupted in the P. nodorum isolate Sn2000 by inserting the hygromycin resistant cassette (hyg) into Sn2000_06735 using a CRISPR‐Cas9‐mediated gene disruption (Fig. S3). The two mutants designated Sn2kΔTox5‐10 and Sn2kΔTox5‐15 as well as an isolate with an ectopic insertion of hyg, designated Sn2k‐ect7, and the wild‐type isolate Sn2000 were inoculated onto LP29 (Snn5). Both Sn2000 and Sn2k‐ect7 were able to induce typical necrotic lesions (Fig. 3a). However, the two Sn2000_06735‐disrupted strains failed to cause necrosis on LP29 (Fig. 3a). This suggested that Sn2000_06735 was targeting Snn5 to cause disease, and therefore Sn2000_06735 will hereafter be referred to as SnTox5.

Fig. 3

Phenotypic and quantitative trait locus (QTL) analysis validation of SnTox5. (a) Phenotype of LP29 (Snn5) inoculated with Sn2000, Sn2k‐ect7 and Sn2000 SnTox5 gene‐disruption mutants Sn2kΔTox5‐10 and Sn2kΔTox5‐15. (b) Phenotype of LP29(Snn5) inoculated with the avirulent isolate Sn79‐1087 (bottom) and the gain‐of‐function transformant Sn79+Tox5‐3 (top). (c). Infiltration of culture filtrate of Sn79+Tox5‐3 on the parental lines of the LP749 population, including Lebsock (top) and PI94749 (bottom). (d) QTL analysis on the LP749 population using strains Sn2000, Sn2k‐ect7, Sn2kΔTox5‐10, Sn2kΔTox5‐15 and Sn79+Tox5‐3, illustrating the significance of Tsn1 and Snn5 in the presence and absence of the SnTox5‐Snn5 interaction.

Snn5, the susceptibility target for SnTox5, was originally mapped using the LP749 double haploid population via infiltration of culture filtrates containing SnTox5 (Friesen et al., 2012). The LP749 population was therefore inoculated with Sn2kΔTox5‐10, Sn2kΔTox5‐15, Sn2k‐ect7 and Sn2000, and a significant QTL, previously described by Friesen et al. (2012), was identified on chromosomes 4B at the Snn5 locus for both the wild‐type and ectopic strains (Fig. 3d). The Snn5 locus explained 33% and 32% of the variation in disease with logarithm of odds (LOD) values of 10.31 and 9.65 for the disease caused by Sn2000 and Sn2k‐ect7, respectively (Fig. 3d; Table 1). The significance of the Snn5 locus was eliminated for the two SnTox5 gene disruption mutants, further validating Sn2000_06735 as SnTox5.

Table 1

Composite interval mapping (CIM) analysis of quantitative trait loci (QTLs) associated with Sn2000 (wild‐type), Sn2k‐ect7 (isolate with an ectopic insertion of the hygromycin resistance gene, cpc‐1:hyg), Sn2kΔTox5‐10, Sn2kΔTox5‐15 ( SnTox5 disruption mutants of Sn2000) and Sn79+Tox5‐3 (Sn79‐1087 transformed with SnTox5) for the inoculation of the LP749 double haploid wheat population derived from the Lebsock (Snn5) × PI94749 (snn5) cross.

Isolate	LODa		R 2
	Snn5	Tsn1	Snn5	Tsn1
Sn2000	10.31*	2.37	0.33	0.10
Sn2k‐ect7	9.65*	3.03*	0.32	0.11
Sn2kΔTox5‐10	0.21	7.82*	0.01	0.26
Sn2kΔTox5‐15	0.14	12.30*	0.01	0.36
Sn79+Tox5‐3	27.00*	0.15	0.66	0.01

aA permutation test with 1000 iterations resulted in a logarithm of the odds (LOD) threshold of 3.00 at P = 0.05.

*, significant QTL at the 0.05 level of probability.

Insertion of SnTox5 converts avirulence to virulence in the presence of Snn5

Parastagonospora nodorum isolate Sn79‐1087 is avirulent on LP29, and the SnTox5 gene is completely absent (Fig. 3b). Transformation of Sn79‐1087 with SnTox5, along with its native promoter, converted the transformed strains Sn79+Tox5‐3 and Sn79+Tox5‐4 into virulent strains on LP29 (Fig. 3b). Sn79+Tox5‐3 caused an average disease reaction of 3.0 on Lebsock and an average disease reaction of 0 on PI94749, whereas Sn79‐1087 (wild‐type) showed no disease on either line. The LP749 population segregated for disease caused by Sn79+Tox5‐3 and subsequent analysis revealed a QTL at the Snn5 locus with a LOD value of 27.00, explaining 66% of the disease variation (Fig. 3d; Table 1). No other QTLs were identified for this transformant, showing that SnTox5 was sufficient to cause disease in the presence of Snn5. The QTL analysis data from the LP749 population using the SnTox5 disruption mutants and the SnTox5 gain‐of‐function transformants further validated that Sn2000_06735 was SnTox5.

Nucleotide diversity of SnTox5 varies across geographical regions

The 197 P. nodorum isolates of the natural population were screened for presence/absence of SnTox5 using whole genome resequencing and 149 (75.6%) contained the SnTox5 gene. Of the 149 isolates, 128 had a complete SnTox5 sequence and were therefore used for further haplotype analysis. The 128 isolates consisted of 22 nucleotide haplotypes. Haplotypes 17,18, 19, 20, 21 and 22, which were found in 11 of the isolates, contained a premature stop codon (Table 2). Of the remaining 117 isolates with functional SnTox5 haplotypes, 47 were found in the Upper Midwest population (44.76% of the Upper Midwest), 47 were found in the South/East population (70.15% of the South/East isolates), seven were found in the Oregon population (87.5% of Oregon isolates), and 17 were found in the Oklahoma population (100% of Oklahoma isolates). The Upper Midwest population, which included the reference isolate Sn2000, consisted of five SnTox5 haplotypes defined by six nonsynonymous and two synonymous polymorphisms (Table 3) and a nucleotide diversity (Pi) of 0.001 09. The South/East population consisted of nine SnTox5 haplotypes defined by seven nonsynonymous and one synonymous polymorphism and a nucleotide diversity of 0.003 27, higher than that of the Upper Midwest population. The calculated SnTox5 pN/pS ratio for the Upper Midwest population and the South/East population showed that SnTox5 was undergoing purifying selection in the Upper Midwest but strong diversifying selection in the South/East, probably adapting to the locally planted cultivars (Table 3).

Table 2

Average disease reaction type of isolates producing the different isoforms of SnTox5 on LP29 with in the United States population of Parastagonospora nodorum.

Isoform	Haplotype	No. of isolates with the haplotype	Average disease reaction typeb	Range of average disease reaction type
Isoform 1	Haplotype 1,10	36	2.27 a	0.50–4.33
Isoform 2	Haplotype 2,12	23	2.96 b	2.17–3.67
Isoform 3	Haplotype 3	16	3.36 c	2.67–4.33
Isoform 4	Haplotype 4	12	2.23 a	1.17–3.25
Isoform 5	Haplotype 5	7	3.07	1.50–4.33
Isoform 6	Haplotype 6	6	2.63	1.67–3.83
Isoform 7	Haplotype 7	4	2.71	2.33–3.00
Isoform 8	Haplotype 8	3	3.50	3.33–3.67
Isoform 9	Haplotype 9	3	2.72	1.83–3.17
Isoform 10	Haplotype 11	1	2.50	–
Isoform 11	Haplotype 13	1	2.50	–
Isoform 12	Haplotype 14	1	3.17	–
Isoform 13	Haplotype 15	1	2.17	–
Isoform 14	Haplotype 16	1	2.17	–
Isoform 15	Haplotype 17a	1	0.17	–
Isoform 16	Haplotype 18a	1	1.17	–
Isoform 17	Haplotype 19a	1	0.17	–
Isoform 18	Haplotype 20a	1	0.50	–
Isoform 19	Haplotype 21a	4	0.36	0.00–0.50
Isoform 20	Haplotype 22a, 28ac	4	0.72	0.50–1.00
Isoform 21	Haplotype 23c	1	2.50	–
Isoform 22	Haplotype 24c	1	2.67	–
Isoform 23	Haplotype 25c	1	3.17	–
Isoform 24	Haplotype 26c	1	1.00	–
Isoform 25	Haplotype 27c	1	3.67	–
Isoform 26	Haplotype 29c	1	2.67	–
Isoform 27	Haplotype 30c	1	1.50	–
Isoform 28	Haplotype 31c	1	2.17	–
Isoform 29	Haplotype 32c	1	3.33	–
Isoform 30	Haplotype 33c	1	2.67	–
Isoform 31	Haplotype 34c	1	3.50	–
Isoform 32	Haplotype 35c	1	3.17	–
Isoform 33	Haplotype 36c	1	3.33	–
Isoform 34	Haplotype 37c	1	3.50	–
Isoform 35	Haplotype 38c	1	3.17	–
Isoform 36	Haplotype 39c	1	3.83	–

aHaplotypes that contain a premature stop codon for SnTox5.

bANOVA of average disease reaction was performed only for the isoforms represented by more than 10 isolates. Least significant difference was calculated at the 0.05 level of probability. Average disease reactions followed by the same letter are not significantly different at the 0.05 level probability.

cHaplotypes with a sequence coverage between 50% and 100% for SnTox5.

Table 3

Calculation of pN/pS ratios for the entire SnTox5 gene and the region of the gene that encodes for the mature protein in the Upper Midwest and South/East populations of Parastagonospora nodorum.

	Upper Midwest (n = 47)		South/East (n = 47)
	Entire gene	Mature protein encoding regiona	Entire gene	Mature protein encoding regiona
Synonymous SNPs	2	2	1	1
Synonymous sites (average)	154.55	108.22	156.97	107.63
Nonsynoymous SNPs	6	3	7	6
Nonsynonymous sites (average)	487.45	341.78	494.03	342.37
pN/pS (sites/average sites)b	0.95	0.47	2.22	1.89

aBase pairs 213–654 of the SnTox5 encode for the mature protein. The sequence resulted from the cleavage of signal peptide and pro‐domain was considered for the calculation of pN : pS ratio.

bpN/pS ratios were calculated using the equation, pN/pS = (nonsynonymous SNPs/nonsynonymous sites (average))/(synonymous SNP/synonymous sites (average)). pN/pS < 1 indicates that the SnTox5 is undergoing purifying selection and pN/pS > 1 indicates that the gene is undergoing diversifying selection in the population.

SnTox5 isoform variation contributes to quantitative degrees of virulence

The 22 haplotypes encoded 20 different isoforms of SnTox5, with six having a premature stop codon (Fig. S4). Three active and five inactive isoforms of SnTox5 were identified in the Upper Midwest population (n = 56), where isoform 1 was the most prevalent at 60.71% (Fig. 4). The Southern/Eastern population (n = 48) harbored nine isoforms (eight active and one inactive) with isoforms 2 and 3 representing 35.42% and 33.33%, respectively (Fig. 4). The P. nodorum population from Oklahoma (n = 17) harbored eight isoforms (Fig. 4). Isoform 5 was the most prevalent in the Oklahoma population, consisting of 35.29%. The Oregon population consisted of only seven isolates, all of which harbored isoform 2 (Fig. 4).

Fig. 4

Prevalence of isoforms of SnTox5 in the Upper Midwest, South/East, Oregon and Oklahoma populations of Parastagonospora nodorum. Distribution of isoforms of SnTox5 in different P. nodorum populations from four regions of the United States showed that each population consisted of multiple isoforms of SnTox5, except for the Oregon population, with a varying degree of prevalence. SnTox5 isoform1 was the most prevalent isoform of the Upper Midwest population whereas isoform 5 was the most prevalent in the Oklahoma population. Isoform 2 was the most prevalent in both the Oregon and South/East populations. Isoforms marked with a ‘α’ represent an inactive form of SnTox5 with a premature stop codon.

Alignment of 14 active SnTox5 isoforms showed that amino acid substitutions were frequent at the 155th and 156th positions (Fig. 5). Three amino acids, including threonine (T), arginine (R) and lysine (K), were observed at the 155th position and two amino acids, including asparagine (N) and serine (S), were observed at the 156th position (Table 4). Isolates of the 14 isoforms were assembled into groups based on the amino acids at the 155th and 156th positions, and amino acid combinations of T‐N, R‐S, K‐S and K‐N showed average disease reactions of 2.31, 2.95, 2.95 and 3.20, respectively (Table 4). Isolates with R‐S, K‐S and K‐N substitutions caused significantly higher average disease reaction on LP29 compared with the T‐N substitution, suggesting that variation in amino acid residues at these two positions may have contributed to the higher virulence of the isolates producing them.

Fig. 5

A portion of the amino acid sequence of the active isoforms of SnTox5 representing the critical substitutions that contribute to the variation in disease using isoform 1 as a reference. Purple arrows indicate the physical position of the substitutions and red arrows indicate the physical position of the cysteine residues. The 155th position consisted of either threonine (T), arginine (R) or lysine (K), whereas the 156th position consisted of either asparagine (N) or serine (S) and was flanked by two cysteines at the 153rd and 161st positions which were predicted to form a disulfide bond. Substitutions T155R and T155K contributed to an increase in averaged disease reaction type, whereas N156S contributed to the variation in average disease reaction type on LP29.

Table 4

Amino acid substitutions at the 155th and 156th positions contribute to the variation in average disease reaction on LP29 caused by the Parastagonospora nodorum isolates harboring active isoforms of SnTox5.

Amino acid at the position		Isoforms representeda	No. of isolates with the substitution	Average disease reaction typeb
155th	156th
T	N	1, 4, 6, 11, 15	56	2.31 a
R	S	2, 10, 12	26	2.95 b
K	S	5, 13	8	2.95 b
K	N	3, 7, 8, 9	26	3.20 b

aIsolates from these isoforms that represent identical substitutions were pooled together for the mean comparison of average disease reaction.

bLeast significant difference (LSD) was calculated at P < 0.05 probability. Numbers followed by the same letter were not significantly different at the 0.05 level of probability.

SnTox5 expression peaks after penetration but before visible lesions

To examine the expression profile of SnTox5 throughout the infection cycle, reverse transcription quantitative polymerase chain reaction was performed using in‐planta samples of Lebsock inoculated with Sn2000 collected at 4, 12, 24, 48, 72, 96 and 120 hpi. SnTox5 expression peaked at 24 hpi, before the onset of lesions where SnTox5 was expressed approximately six times that of the actin gene. The expression of SnTox5 gradually decreased with the progression of the disease before it returned to levels like that of actin at 120 hpi where the pathogen had already colonized the mesophyll tissue (Fig. 6; Table S3).

Fig. 6

Temporal expression pattern of SnTox5 in planta on Lebsock inoculated with Sn2000. The x‐axis shows the sample collection time points in h post‐inoculation (hpi) for quantitative PCR and the y‐axis represents the expression of SnTox5 relative to the expression of the actin gene. Error bars represent the SEM from three replications for each time point.

Laser confocal microscopy shows that SnTox5 facilitates complete colonization of LP29

To visualize the effect of the SnTox5‐Snn5 interaction on penetration and colonization of the leaf cell layers, fluorescently labeled P. nodorum strains Sn2000, Sn2kΔTox5‐15, Sn79+Tox5‐3 and Sn79‐1087 were inoculated onto LP29. The infection process of each strain was observed using laser scanning confocal microscopy at seven different time points post‐inoculation. Germination of spores of all four strains was visible within 4 hpi (Figs 7, 8). At 12 hpi, penetration of the leaf surface was also visible for all four strains; however, it was clear that strains that contained SnTox5 had increased penetration compared with strains that lacked SnTox5 (Figs 7, 8).

Fig. 7

Laser confocal microscopy of the infection process of red fluorescent protein (RFP)‐tagged Parastagonospora nodorum strains Sn2000 and Sn2kΔTox5‐15 on wheat differential line LP29 (Snn5). (a) Micrographs of wheat leaves obtained through confocal imaging at 4, 12, 24, 48, 72, 96 and 120 h post‐inoculation (hpi) of Sn2000 and Sn2kΔTox5‐15. Fungal spores and hyphae are displayed in red. Wheat cells are displayed in various colors depending on the autofluorescence emitted by the degrading chloroplast, where the green color indicates healthy cells and the yellow to red color indicates cells that are undergoing programmed cell death. Separate z‐stack micrographs taken at the epidermis and the mesophyll tissue at 48–120 hpi showed that Sn2kΔTox5‐15 failed to advance into the mesophyll tissue. (b) Schematic drawings of transverse sections of each micrograph of (a). The purple line separates the epidermis from the mesophyll tissue. Bars, 60 µm.

Fig. 8

Laser confocal microscopy of the infection process of red fluorescent protein (RFP)‐labeled Parastagonospora nodorum strains Sn79‐1087 (avirulent) and Sn79+Tox5‐3 on wheat differential line LP29 (Snn5). (a) Micrographs of wheat leaves obtained through confocal imaging at 4, 12, 24, 48, 72, 96 and 120 h post‐inoculation (hpi) of Sn79‐1087 and Sn79+Tox5‐3. Fungal spores and hyphae are displayed in red. Wheat cells are displayed in various colors depending on the autoflorescence emitted by the degrading chloroplast, where the green color indicates healthy cells and the yellow to red color indicates cells undergoing program cell death. Separate z‐stack micrographs taken at the epidermis and the mesophyll tissue at 72–120 hpi showed that Sn79‐1087 failed to reach the mesophyll tissue as SnTox5 mutant of Sn2000. Transfer of SnTox5 to Sn79‐1087 enables the fungus to reach the mesophyll tissue. (b) Schematic drawings of transverse sections of each micrograph of (a). The purple line separates the epidermis from the mesophyll tissue. Bars, 60 µm.

At 48 hpi, all strains were able to colonize the epidermal tissue. Sn2000 and Sn79+Tox5‐3 had initiated colonization of the vascular and mesophyll tissue at 48 hpi, where fungal hyphae grew around the mesophyll tissue before the induction of PCD. Progressive colonization of the mesophyll was observed from 48 to 96 hpi, with the first evidence of PCD being observed at 72 hpi. At the 72 hpi time point, mesophyll cells were surrounded by mycelium, and chloroplasts had changed from green to yellow, indicating cellular disruption (Figs 7, 8). Deterioration of the chloroplasts was followed by the shrinking and total collapse of the surrounded mesophyll cells at the 96 and 120 hpi (Figs 7, 8). By 120 hpi, much of the mesophyll tissue was colonized by both Sn2000 and Sn79+Tox5‐3. It was evident that Sn2000 colonized more tissue than that of Sn79+Tox5‐3 at 120 hpi, indicating that additional effectors may be present in Sn2000 that are effective in LP29 but that are not present in the avirulent Sn79‐1087 (Figs 7, 8). Even though Sn2kΔTox5‐15 and Sn79‐1087 lacked SnTox5, both penetrated the epidermis by 24 hpi. Colonization of the epidermis was observed by 96 hpi, similar to Sn2000 and Sn79+Tox5‐3. However, unlike Sn2000 and Sn79+Tox5‐3, Sn2kΔTox5‐15 and Sn79‐1087 were never able to colonize the mesophyll or vascular tissue of the leaf (Figs 7, 8).

Typically, macroscopic necrotic lesions were first visible on the leaves at 48–72 hpi, correlating with the colonization of the mesophyll tissue by Sn2000 and Sn79+Tox5‐3. Until P. nodorum started to colonize the mesophyll tissue, the leaf remained green. Both Sn2kΔTox5‐15 and Sn79‐1087 failed to reach mesophyll tissue and therefore failed to produce lesions on inoculated leaves as expected.

As a proxy for fungal fitness, the fungal volumes of Sn2000 and Sn2kΔTox5‐15 were measured at 12, 24, 48, 72, 96 and 120 hpi (Fig. 9a). The volume of Sn2000 gradually increased on LP29 over time as expected; however, the volume of Sn2kΔTox5‐15 on LP29 remained constant from 12 to 120 hpi. The volume of Sn2000 started to increase significantly at 24 hpi, coinciding with the upregulation of SnTox5 and fungal colonization of the mesophyll layer (Figs 7, 9a).

Fig. 9

Volume of the fungus calculated through laser confocal microscopy on LP29 (Snn5) (a) and Lebsock (Snn5 and Tsn1) (b) for the strains Sn2000 (+SnTox5, +SnToxA) and Sn2kΔTox5(–SnTox5, +SnToxA) at various time points post‐inoculation. The x‐axis represents h post‐inoculation (hpi) and the y‐axis represents the volume of the fungus in µm3. Error bars represent the SEM of three replications. The volume of Sn2000 increased linearly after 24 hpi on both wheat lines. The increase in volume of Sn2kΔTox5‐15 on LP29 was negligible during the experiment. However, linear increase in volume of Sn2kΔTox5‐15 was observed after 72 hpi on Lebsock as a result of the establishment of the SnToxA‐Tsn1 interaction.

SnTox5 facilitates colonization of the mesophyll layer of LP29 even in the absence of Snn5

To further analyze the function of the SnTox5, we inoculated Sn2000 and the Sn2kΔTox5‐15 mutant on LP29 and its Snn5 disruption mutant LP29Δsnn5. At 120 hpi, Sn2000 was able to colonize both the epidermis and the mesophyll tissue as described above (Fig. 10). In the Sn2000 inoculation of LP29Δsnn5, Sn2000 was able to progress into the mesophyll tissue; however, unlike the Sn2000‐LP29 inoculation, the fungus was not able to induce PCD and was only observed to advance into the first two cell layers of the mesophyll tissue. As mentioned earlier, SnTox5 mutants of Sn2000 were only able to penetrate and colonize the epidermal tissue of LP29. At 120 hpi, spores of Sn2kΔTox5‐15 were able to form germ tubes on LP29Δsnn5. However, Sn2kΔTox5‐15 was not even able to penetrate the epidermal tissue of LP29Δsnn5, even up to 120 hpi (Fig. 10). Similar results were also observed for the inoculation of Sn79‐1087 and Sn79‐1087+Tox5 on LP29 and LP29Δsnn5 (Fig. S5). More work will need to be done on this interaction to understand the role of Snn5 in penetration in the absence of SnTox5.

Fig. 10

Laser confocal microscopy of red fluorescent protein (RFP)‐tagged Sn2000 and Sn2kΔTox5‐15 inoculated on LP29 and LP29Δsnn5 at 120 h post‐inoculation (hpi) (a). Micrographs of red fluorescent protein (RFP)‐tagged Sn2000 (+SnTox5) and Sn2kΔTox5‐15 (–SnTox5) inoculated on LP29 (Snn5) and LP29Δsnn5 (snn5) at 120 hpi. (b) Schematic drawings of transverse sections of each micrograph of (a). (a(i), b(i)) Sn2000 induced programmed cell death (PCD) and colonized the mesophyll tissue of LP29. (a(ii), b(ii)) Sn2000 was able to colonize the epidermis and hyphae were advanced into the mesophyll tissue but failed to induce PCD on LP29Δsnn5 as a result of the lack of functional Snn5. (a(iii), b(iii)) Sn2kΔTox5 colonized the epidermal tissue but failed to progress to the mesophyll tissue as it lacked SnTox5. (a(iv), b(iv)) Sn2kΔTox5 formed a penetration structure on LP29Δsnn5. However, Sn2kΔTox5 was not able to penetrate the epidermis as Sn2000ΔTox5 and LP29Δsnn5 lacked SnTox5 and Snn5. Therefore, these results showed that establishment of SnTox5‐Snn5 is essential for the P. nodorum to colonize the mesophyll of LP29 and lack of either partner of the interaction has a deleterious effect of fungal growth. Bars, 100 µm.

As a proxy for fungal fitness, the fungal volume was calculated for each of these combinations. Owing to the SnTox5‐Snn5 interaction, the fungal volume of Sn2000 on LP29 was significantly higher than that of all the other combinations at 120 hpi (Fig. S6). Although less than the Sn2000‐LP29 combination, the volume of Sn2000 in LP29Δsnn5 was significantly higher than the volume of Sn2kΔTox5‐15 in LP29 or LP29Δsnn5, indicating a role for SnTox5 in pathogenic fitness. No significant difference was observed between the other two combinations. Therefore, our results show not only that SnTox5 targets Snn5 to induce PCD to gain nutrients, but also that SnTox5 facilitates colonization of the mesophyll even in the absence of Snn5.

Laser confocal analysis of Sn2000 and Sn2kΔTox5‐15 on Lebsock

To further evaluate the additive nature of the SnTox5‐Snn5 and SnToxA‐Tsn1 interactions, we analyzed the infection process of Sn2000 and Sn2kΔTox5‐15 on Lebsock, which carries both Tsn1 and Snn5. Both Sn2000 and Sn2kΔTox5‐15 were able to penetrate and colonize the epidermis of Lebsock within 24 hpi (Fig. 11). Sn2000 was able to penetrate the mesophyll layer by 48 hpi and Sn2kΔTox5‐15 reached the mesophyll tissue at 96  hpi (Fig. 11), showing that the SnTox5‐Snn5 interaction was facilitating a more rapid colonization of the mesophyll tissue than was the SnToxA‐Tsn1 interaction alone.

Fig. 11

Laser confocal microscopy of the infection process of red fluorescent protein (RFP)‐labeled Parastagonospora nodorum strains Sn2000 and Sn2kΔTox5‐15 on wheat cv Lebsock (Snn5 and Tsn1). (a) Micrographs of wheat leaves obtained through laser confocal imaging at 4, 12, 24, 48, 72, 96 and 120 h post‐inoculation (hpi) of Sn2000 and Sn2kΔTox5. Fungal spores and hyphae are displayed in red. Wheat cells are displayed in various colors depending on the autofluorescence emitted by the degrading chloroplast, where the green color indicates healthy cells and the yellow to red color indicates cells that are undergoing programmed cell death. (b) Schematic drawings of transverse sections of each micrograph of (a) clearly illustrate that Sn2000 was able to advance into the mesophyll tissue and colonize the mesophyll tissue rapidly compared with that of Sn2kΔTox5‐15. Bars, 60 µm.

Calculation of the fungal volume of Sn2000 and Sn2kΔTox5‐15 in Lebsock clearly showed that even though both strains were able to colonize the mesophyll tissue, Sn2000 had significantly higher fungal volume compared with that of the Sn2kΔTox5‐15, starting at 48 hpi and continuing through the 120 hpi time point (Fig. 9b). In addition, an increase in fungal volume was observed at 24 hpi for Sn2000, whereas the fungal volume did not increase until 72 hpi for Sn2kΔTox5‐15 (Fig. 9b). These results show that although the SnToxA interaction is ultimately sufficient to facilitate the colonization of the mesophyll, SnTox5 is much more efficient in this regard. Together, the two interactions act synergistically, showing that these effectors have a unique role as well as the overlapping one of targeting their corresponding susceptibility targets to induce PCD, making the pathogen more efficient at completing the pathogenic life cycle.

Discussion

Parastagonospora nodorum is a necrotrophic fungal pathogen that deploys a plethora of necrotrophic effectors (NEs) to induce PCD on susceptible wheat lines. In this study, we performed GWAS on a P. nodorum natural population of 197 isolates, to identify Sn2000_06735 as a candidate for SnTox5. We subsequently used CRISPR/Cas9‐mediated gene editing to disrupt the gene and gain‐of‐function transformation to complement the gene showing that Sn2000_06735 was both sufficient and necessary to cause disease on LP29, the Snn5 differential line. In addition, a QTL analysis showed the presence and absence of the QTL on Snn5 for the inoculation of gain‐of‐function and disruption mutants, respectively, on the LP749 population.

SnTox5 encodes a small secreted protein that consists of a secretion signal and a putative pro‐sequence that is cleaved at a predicted Kex2 protease cleavage site, which showed homology to SnTox3 (Liu et al., 2009; Outram et al., 2020). The Kex2 protease is unique to fungi and Outram et al. (2020) showed that Kex2‐processed pro‐domain (K2PP) effectors were common in pathogenic fungi, including P. nodorum effectors SnTox3 and SnToxA. Outram et al. (2020) demonstrated experimentally that the Kex2‐processed pro‐domain was critical for SnTox3 folding and activity, and the validation of SnTox5 provides another K2PP effector interaction for further study of this class of effectors.

Outram et al. (2020) presented the crystal structure of SnTox3, which was used by the protein structuring server phyre2 to model the 3D structure of both SnTox5 and SnTox267 (Richards et al., 2022). phyre2 was able to model 98% of the structure of the mature SnTox5 with 100% confidence, whereas it was able to model only 35% of the mature SnTox267 with 92.6% confidence, which suggests structural homology between SnTox3 and SnTox5 and, to a lesser extent, SnTox3 and SnTox267. In addition, cysteines and their positions in SnTox3 and SnTox5 were similar and localized to one end of the protein. Like the predicted structure of SnTox5 (Sn2000_06735), the crystal structure of the mature SnTox3 consisted of 10 β‐strands, with eight of these forming a β‐barrel (Outram et al., 2020), a structure that is probably conserved in SnTox5. This β‐barrel structure was previously identified in a family of bacterial pore‐forming toxins and SnTox3 was the first fungal effector to be identified with a β‐barrel (Outram et al., 2020). Guo et al. (2018) showed that Magnaporthe oryzae avirulence and ToxB‐like (MAX) effectors AVR1‐CO39 and AVR‐PikD were recognized upon direct interaction with a heavy metal‐associated (HMA) integrated domain (ID) in the two rice nucleotide binding leucine‐rich repeat (NLR) receptors, RGA5 and Pikp‐1. Elucidation of the crystal structures showed that the RGA5HMA/AVR1‐CO39 and the Pikp‐1 HMA/AVR‐PikD interactions both involved the alignment of antiparallel β‐strands between the MAX effectors and the corresponding HMA domains, indicating a conserved mechanism in rice to recognize effectors, leading to resistance. Here we show two P. nodorum effector proteins whose genes are probably paralogs. Each protein harbors eight antiparallel β‐strands that form a unique β‐barrel structure. These proteins are each ‘recognized’ by different host genes/proteins resulting in susceptibility as this interaction involves a necrotrophic pathogen. This indicates the potential of a common host vulnerability that is being targeted by P. nodorum using this protein structure.

The prevalence of the SnTox5 gene (75.6%) in this natural population was slightly higher than that of SnToxA (63.4%) and SnTox3 (58.9%) (Richards et al., 2019). By contrast, the presence of SnTox5 was less compared with the prevalence of SnTox1 and SnTox267 in the same population, which was 95.4% for each (Richards et al., 2019, 2022). The ability to target multiple host susceptibility genes such as that for SnTox267 or SnTox3 (Friesen et al., 2008; Zhang et al., 2011; Richards et al., 2022), the existence of a secondary function such as the ability of SnTox1 to bind chitin (Liu et al., 2016) or SnTox3 to target PR1 proteins to suppress defense (Sung et al., 2021), or the prevalence of host susceptibility genes in the planted wheat of a given region (Richards et al., 2019) are all likely to govern the frequency of an effector gene in a fungal population. Therefore, prevalence of SnTox5 in the majority of P. nodorum isolates collected from the Upper Midwest, Oklahoma, Oregon and the South/East regions of the US suggests the prevalence of Snn5 in wheat planted in these regions or a secondary function that drives the maintenance of this gene.

Richards et al. (2019) showed that SnTox3 was under purifying selection in the South/East population and diversifying selection in the Upper Midwest population. Here we show that SnTox5 was under purifying selection in the Upper Midwest but diversifying selection in the South/East population, the opposite of SnTox3. Because the Upper Midwest wheat region is predominately spring wheat and the South/East wheat region is predominately winter wheat, the locally planted cultivars are vastly different. These differences include the complement of effector targets present in the local cultivars (Crook et al., 2012). Additionally, it is likely that population‐specific alleles of Snn3 and Snn5 exist that may be driving the diversification of both SnTox3 and SnTox5 in their respective populations.

Parastagonospora nodorum isolates harboring a diversity of active isoforms of the SnTox5 protein were evaluated for disease reaction on LP29. Isolates harboring SnTox5 isoforms predominately from the South/East winter wheat‐growing regions were significantly more virulent on LP29 than isolates harboring isoforms that were prevalent in the Upper Midwest spring wheat and durum wheat regions. No isolates harboring the most virulent two isoforms were identified in the Upper Midwest population, indicating that the genetic background of the winter wheat of the South/East, including but not necessarily limited to Snn5, was probably the selection pressure driving this diversity and the increased virulence.

The importance of critical amino acid residues in P. nodorum–wheat effector–target gene interactions have been reported for SnToxA (Meinhardt et al., 2002; Lu et al., 2014) and SnTox3 (Sung et al., 2021). A total of 14 active isoforms of SnTox5 were identified in our natural population. Isolates carrying isoforms with T155K and T155R amino acid substitutions caused significantly higher disease. Threonine is neutral in its hydrophobicity, whereas both lysine and arginine are highly hydrophilic residues. Therefore, an increase in the hydrophilicity at the 155th position appears to result in an increase in virulence of the isolates producing these isoforms. Amino acid residues at the 156th position were also variable and consisted of either serine or asparagine. The highest average disease reaction was observed when positions 155 and 156 were occupied by K and N compared with K and S, respectively; however, these differences were not significant. Based on our results, we hypothesize that the amino acid residues at position 155 and possibly 156 are under selection and are critical to the effectiveness of the protein in targeting Snn5 or, equally plausible, a secondary target involving defense suppression as the pathogen initiates colonization of the mesophyll.

SnTox5 expression peaked early, with its highest expression at 24 hpi, which is like that of SnTox267 (Richards et al., 2022), but earlier than that of SnToxA, SnTox1 (Liu et al., 2012) and SnTox3 (Liu et al., 2009). This expression pattern indicated that SnTox5 is probably involved in the early colonization of the leaf, including the initial colonization of the mesophyll, which is initiated at 24 hpi and continues through 120 hpi.

Laser scanning confocal microscopy was used to evaluate the importance of SnTox5 in the various stages of infection. Strains with or without SnTox5 were able to colonize the plant epidermis, indicating that SnTox5 was not necessary to colonize the epidermal layer of LP29. The visible differences in colonization began at 48 hpi where the strains producing SnTox5 were able to begin colonizing the mesophyll layer but those strains not producing SnTox5 did not colonize the mesophyll, even by 120 hpi. Similarly, SnTox5 gain‐of‐function transformations in the avirulent Sn79‐1087 background allowed this strain to colonize the mesophyll as well as induce PCD. Both comparisons showed that SnTox5 was facilitating the colonization of the mesophyll layer of the leaf.

To further investigate the function of SnTox5 in the presence and absence of Snn5, Sn2000 (SnTox5) and Sn2kΔTox5‐15 (no SnTox5) were inoculated on LP29 (Snn5) and LP29Δsnn5 (no Snn5). When comparing Sn2000 on LP29Δsnn5 with Sn2kΔTox5 on LP29, the expectation was that these two inoculations would have similar outcomes as the SnTox5‐Snn5 interaction was incomplete in each. However, Sn2000 was clearly able to advance into the mesophyll of LP29Δsnn5, whereas Sn2kΔTox5 was only able to colonize the epidermis of LP29, showing again that SnTox5 was facilitating the colonization of the mesophyll in the absence of Snn5. These results were further supported by the fungal volume for each combination where the highest fungal colonization was observed when both SnTox5 and Snn5 were present and the second highest fungal colonization was achieved when SnTox5 was present, even in the absence of Snn5. Together these combinations clearly showed that SnTox5 was triggering PCD via the targeting of Snn5 but was probably first facilitating the early colonization of the mesophyll.

In the combination of Sn2kΔTox5 on LP29Δsnn5, as a result of the lack of SnTox5, our expectation was that the Sn2kΔTox5 strain would just colonize the epidermal layer like Sn2kΔTox5‐15 on LP29. However, Sn2kΔTox5‐15 failed to even penetrate the epidermal layer in any of the leaves examined. Our only explanation for this phenomenon was that Snn5 was somehow involved in communication with the pathogen. However, working out a model to explain this result will require further work, including the cloning of Snn5, which is under way.

These results indicate that, like SnTox1 and SnTox3, SnTox5 has a secondary function, that is, facilitating the colonization of the mesophyll, probably by delaying the defense response. It is interesting to note that colonization of the mesophyll is evident at 48 hpi; however, no PCD is observed until the 72 hpi time point. As SnTox3 also suppresses the host defense response through direct interaction with PR1 class wheat proteins (Breen et al., 2016; Sung et al., 2021), and as SnTox5 is highly similar in structure to SnTox3, we are hypothesizing that SnTox5 is also involved in suppression of early host defense, resulting in a less hostile colonization environment. This initial colonization of the mesophyll is followed by the onset of PCD as well as the progression of additional colonization and completion of its pathogenic life cycle.

The P. nodorum modulation of the defense response is somewhat analogous to the ‘latent necrotroph’ Zymoseptoria tritici (Sánchez‐Vallet et al., 2015) where, during its latent phase, the necrotrophic pathogen colonizes the leaf tissue while suppressing the early onset of the defense response before actively triggering defense and PCD during its necrotrophic phase (Rudd et al., 2015). Parastagonospora nodorum is a necrotrophic specialist that uses multiple necrotrophic effectors that target corresponding susceptibility genes that trigger the defense response, resulting in PCD to gain nutrients. However, recent evidence involving SnTox1, SnTox3 and SnTox5 shows that this pathogen has evolved multiple means of host manipulation by using effectors with modes of action beyond just triggering PCD.

To evaluate the role of SnTox5 and Snn5 in the presence of SnToxA and Tsn1, we used laser confocal microscopy to collect fungal volume data and to visualize the pathogen movement of P. nodorum on the durum wheat line Lebsock (Tsn1/Snn5) (Fig. 11). Sn2000 (SnToxA/SnTox5) was able to initiate colonization of the mesophyll tissue by 48 hpi, whereas Sn2kΔTox5‐15 (SnToxA only) failed to breach the mesophyll layer until 96 hpi. Therefore, even though SnToxA can target Tsn1 to induce PCD, SnTox5 initiates an earlier and more efficient colonization of the mesophyll, resulting in earlier PCD and therefore faster acquisition of cellular nutrients that are needed to complete the fungus’ life cycle. This combination reiterates that SnTox5 is facilitating early mesophyll colonization, a role that is not duplicated by SnToxA.

Together, our evidence shows that SnTox5 is facilitating efficient colonization of the mesophyll layer, possibly through the suppression of host defense, resulting in a more efficient build‐up of biomass. This build‐up of biomass in the mesophyll is followed by the triggering of host defense through the targeting of Snn5. The PCD triggered by the SnTox5‐Snn5 interaction results in organelle and cell membrane disruption and the release of cellular nutrients needed for the pathogen to reproduce.

Author contributions

TLF, GKK and JKR designed the research; GKK, and JKR performed the research; GKK, JKR and NAW performed the data analysis and interpretation; JDF, SSX and KLDR provided plant populations and marker data; PB provided expertise in microscopy; ZL provided oversight; TLF and GKK wrote and organized the manuscript. All authors edited and approved the manuscript.

Fig. S1 Nucleotide sequence of SnTox5 and its promoter region.

Fig. S2 Detailed secondary structure of the mature SnTox5.

Fig. S3 PCR confirmation of SnTox5 gene deletion in Sn2kΔTox5 strains.

Fig. S4 Pairwise alignment of 20 isoforms of SnTox5.

Fig. S5 Laser confocal microscopy of Sn79‐1087 and Sn79+Tox5 on LP29 and LP29Δsnn5.

Fig. S6 Volume analysis of Sn2000 and Sn2kΔTox5‐15 on LP29 and LP29Δsnn5.

Methods S1 Disease phenotyping.

Methods S2 Whole genome sequencing and variant identification.

Methods S3 Genome‐wide association mapping.

Methods S4 QTL analysis of the LP749 population.

Methods S5 Development of construct of SnTox5 using Gateway cloning system.

Methods S6 Inoculation and infiltration of gain‐of‐function transformants.

Methods S7 Extraction of SnTox5 sequence for population genetics and haplotype analysis.

Methods S8 Development of Snn5 mutants of LP29.

Table S1 Primers used in this study.

Table S2 Phenotypic data used in GWAS analysis.

Table S3 Temporal expression of SnTox5.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Click here for additional data file.