Genetics of Variable Disease Expression Conferred by Inverse Gene-For-Gene Interactions in the Wheat-Parastagonospora nodorum Pathosystem

Host gene-effector interactions in the wheat-P. nodorum pathosystem display additivity, epistasis, differential effector gene expression, and complex interplay of genetic regulation.

Abstract

The wheat-Parastagonospora nodorum pathosystem involves the recognition of pathogen-secreted necrotrophic effectors (NEs) by corresponding wheat NE sensitivity genes. This inverse gene-for-gene recognition leads to necrotrophic effector-triggered susceptibility and ultimately septoria nodorum blotch disease. Here, we used multiple pathogen isolates to individually evaluate the effects of the host gene-NE interactions Tan spot necrosis1-Stagonospora nodorum ToxinA (Tsn1-SnToxA), Stagonospora nodorum necrosis1-Stagonospora nodorum Toxin1 (Snn1-SnTox1), and Stagonospora nodorum necrosis3-B genome homeolog1-Stagonospora nodorum Toxin3 (Snn3-B1-SnTox3), alone and in various combinations, to determine the relative importance of these interactions in causing disease. Genetic analysis of a recombinant inbred wheat population inoculated separately with three P. nodorum isolates, all of which produce all three NEs, indicated that the Tsn1-SnToxA and Snn3-B1-SnTox3 interactions contributed to disease caused by all four isolates, but their effects varied and ranged from epistatic to additive. The Snn1-SnTox1 interaction was associated with increased disease for one isolate, but for other isolates, there was evidence that this interaction inhibited the expression of other host gene-NE interactions. RNA sequencing analysis in planta showed that SnTox1 was differentially expressed between these three isolates after infection. Further analysis of NE gene-knockout isolates showed that the effect of some interactions could be masked or inhibited by other compatible interactions, and the regulation of this occurs at the level of NE gene transcription. Collectively, these results show that the inverse gene-for-gene interactions leading to necrotrophic effector-triggered susceptibility in the wheat-P. nodorum pathosystem vary in their effects depending on the genetic backgrounds of the pathogen and host, and interplay among the interactions is complex and intricately regulated.

Wheat (Triticum aestivum) currently supplies 20% of the world’s calorie intake. Parastagonospora (syn. ana, Stagonospora; teleo, Phaeosphaeria) nodorum (Berk.) Quaedvleig, Verkley, and Crous is a necrotrophic fungal pathogen that causes the disease septoria nodorum blotch (SNB; formerly called Stagonospora nodorum blotch). SNB can lead to severe yield losses, reaching upwards of 50% (Eyal et al., 1987), and decrease grain quality.

Biotrophic pathogens require living tissue to proliferate and complete their life cycle. Plants have innate immune systems to combat biotrophic pathogens that involve the recognition of pathogen-produced molecules, such as effectors and pathogen-associated molecular patterns (PAMPs). Recognition of PAMPs, which are usually conserved molecules that serve essential functions, occurs by way of pattern recognition receptors (PRRs) in the host and leads to the activation of the PAMP-triggered immunity (PTI) pathway characterized by an oxidative burst, activation of defense response genes, and sometimes localized cell death (Jones and Dangl, 2006; Day et al., 2011; van Schie and Takken, 2014). Recognition of pathogen-produced effectors by plant resistance genes occurs in a gene-for-gene manner (Flor, 1956). This recognition leads to activation of the effector-triggered immunity (ETI) pathway and provides a second layer of defense against biotrophic pathogens. The ETI responses largely overlap with those of the PTI pathway resulting in the restriction of biotrophic pathogen growth (van Schie and Takken, 2014).

Necrotrophic pathogens like P. nodorum induce cell death by secreting necrotrophic effectors (NEs) that, when recognized by the host, elicit the activation of programmed cell death and other components of the host defense response (Friesen and Faris, 2010; Winterberg et al., 2014; Shi et al., 2016b). However, because the pathogen can survive the various components of this defense response and feed on the dying cells, the result is host susceptibility (pathogen virulence). Therefore, these interactions are referred to as “inverse gene-for-gene” because the recognition of NEs by dominant host sensitivity genes leads to necrotrophic effector triggered susceptibly (NETS) as opposed to ETI as is observed in host-biotroph gene-for-gene interactions (Friesen and Faris, 2010; Oliver et al., 2012; Shi et al., 2016b).

To date, a total of nine host sensitivity gene-NE interactions have been characterized in this pathosystem: Tan spot necrosis1 (Tsn1)-Stagonospora nodorum ToxinA (SnToxA; Friesen et al., 2006, 2009; Liu et al., 2006; Faris and Friesen, 2009; Zhang et al., 2009; Faris et al., 2010, 2011), Stagonospora nodorum necrosis1 (Snn1)-Stagonospora nodorum Toxin1 (SnTox1; Liu et al., 2004b, 2004a, 2012; Reddy et al., 2008; Shi et al., 2016b), Stagonospora nodorum necrosis2 (Snn2)-Stagonospora nodorum Toxin2 (SnTox2; Friesen et al., 2007, 2009; Zhang et al., 2009), Stagonospora nodorum necrosis3-B genome homeolog1 (Snn3-B1)-Stagonospora nodorum Toxin3 (SnTox3; Friesen et al., 2008; Liu et al., 2009; Shi et al., 2016a), Stagonospora nodorum necrosis3-d genome homeolog1 (Snn3-D1)-SnTox3 (Zhang et al., 2011), Stagonospora nodorum necrosis4 (Snn4)-Stagonospora nodorum Toxin4 (SnTox4; Abeysekara et al., 2009, 2012), Stagonospora nodorum necrosis5 (Snn5)-Stagonospora nodorum Toxin5 (SnTox5; Friesen et al., 2012), Stagonospora nodorum necrosis6 (Snn6)-Stagonospora nodorum Toxin6 (SnTox6; Gao et al., 2015), and Stagonospora nodorum necrosis7 (Snn7)-Stagonospora nodorum Toxin7 (SnTox7; Shi et al., 2015).

The Tsn1-SnToxA, Snn1-SnTox1, and Snn3-B1-SnTox3 interactions have been studied more intensively due to the cloning of host sensitivity and/or pathogen NE genes. The NE SnToxA interacts with the wheat sensitivity gene Tsn1, which encodes a protein consisting of an N-terminal Ser/Thr protein kinase domain and C-terminal nucleotide binding and Leu-rich repeat domains (Friesen et al., 2006; Faris et al., 2010). The ToxA gene is unique compared with the other NEs in the wheat-P. nodorum system in that it was likely horizontally transferred to Pyrenophora tritici-repentis (Friesen et al., 2006) and has recently been identified in Bipolaris sorokiniana (Friesen et al., 2018; McDonald et al., 2018) and Cochliobolus heterostrophus (Lu et al., 2015), making ToxA an important virulence factor for multiple host-pathogen interactions.

The Snn3-B1-SnTox3 interaction has been shown to contribute significantly to SNB in some host genetic backgrounds (Friesen et al., 2008). Snn3-B1 is located on the short arm of chromosome 5B, and fine mapping has been done to initiate the cloning process (Shi et al., 2016a). SnTox3 was cloned by Liu et al. (2009) and encodes for a secreted protein. After recognition, host defense genes are up-regulated leading to programmed cell death. However, these host genes differ from those associated with the Tsn1-SnToxA interaction (Winterberg et al., 2014).

Snn1 is located on the short arm of chromosome 1B (Liu et al., 2004b) and was cloned by Shi et al. (2016b). The Snn1 gene product is a wall-associated kinase protein that is similar in structure to PRR proteins typically involved in early recognition of pathogens and the up-regulation of the PTI pathway. Snn1 was the first susceptibility gene to be cloned with a PRR-like structure, suggesting that necrotrophic pathogens have evolved the ability to hijack both the PTI and the ETI pathways. SnTox1 was cloned by Liu et al. (2012) and encodes a protein that is Cys-rich and contains a functional chitin-binding motif. SnTox1 is unique from other known NEs in that it is a dual-function protein (Liu et al., 2016). SnTox1 binds chitin and protects the pathogen from wheat chitinases (Liu et al., 2016) in addition to interacting directly with the Snn1 protein (Shi et al., 2016b). The important role SnTox1 plays in penetration, colonization, and sporulation may be the reason that ∼95.4% of the U.S. P. nodorum natural population and 84% of P. nodorum isolates worldwide carry SnTox1 (T. L. Friesen and J. K. Richards, unpublished data), which is much higher than the frequency of Snn1 in wheat (McDonald et al., 2013).

It has been shown that when multiple interactions occur between wheat and P. nodorum, there is a higher level of disease severity with the interactions contributing additively to overall disease (as reviewed by Friesen and Faris, 2010). Variable expression of NE genes also contributes to disease severity and adds another layer of complexity to the wheat-P. nodorum pathosystem. Faris et al. (2011) found that increased SnToxA expression in the P. nodorum isolate BBC03Sn5 compared with LDN03Sn4 led to the Tsn1-SnToxA interaction contributing to a higher level of disease in BBC03Sn5 than LDN03Sn4. Gao et al. (2015) found that, despite the presence of the SnTox1 gene in the isolate Sn6, the Snn1-SnTox1 interaction was not a significant contributor to disease due to the lack of SnTox1 transcription. Recently, Phan et al. (2016) found that when SnTox1 is eliminated from the isolate SN15, SnTox3 expression is increased, implying that SnTox1 may suppress SnTox3. Although multiple studies have looked at SnTox1 expression in single isolates, no research has been published comparing SnTox1 expression across multiple P. nodorum isolates on a single host population.

The cloning of the three NE genes SnToxA, SnTox1, and SnTox3 makes it feasible to dissect and quantify the relative effects of the Tsn1-SnToxA, Snn1-SnTox1, and Snn3-B1-SnTox3 interactions, but their effects have not been evaluated in a single host genetic background. Here, we evaluated multiple P. nodorum isolates that produce all three NEs for their ability to cause SNB, and quantified the roles of each compatible interaction in causing disease. We show that the roles of these interactions in contributing to SNB vary depending on the isolate, and that relationships can range from additive to epistatic, with even some evidence of antagonistic effects of one interaction leading to inhibition of others. Our results suggest that a transcriptionally regulated balance of NE production is maintained in P. nodorum.

RESULTS

Marker Analysis and Linkage Map Construction

A population of 190 recombinant inbred lines (RILs) derived from a cross of the common hexaploid wheat line Sumai 3 and a Chinese Spring–Triticum turgidum ssp. dicoccoides chromosome 5B disomic chromosome substitution line (CS-DIC 5B), hereafter referred to as the CDS population, was developed and genotyped with simple sequence repeat (SSR) and single nucleotide polymorphism (SNP) markers to generate genetic linkage maps representing the whole genome. The maps consisted of 98 SSRs, 2,098 SNPs, and the three NE sensitivity gene loci (Tsn1, Snn1, and Snn3-B1; see below), for a total of 2,199 markers including 848 unique loci (Supplemental Table S1). The maps were used to quantify the effects of the NE sensitivity loci using quantitative trait loci (QTL) analysis methods.

Genetic Analysis and Mapping of the NE Sensitivity Genes Tsn1, Snn3-B1, and Snn1

CS-DIC 5B carries Snn1 but not Tsn1 or Snn3-B1; therefore, it was sensitive to SnTox1 but not SnToxA or SnTox3 (Fig. 1). Sumai 3 harbors Tsn1 and Snn3-B1 but lacks Snn1 and was therefore sensitive to SnToxA and SnTox3 but insensitive to SnTox1. The segregation ratios and χ² analysis of the NE reactions in the CDS population are presented in Supplemental Table S2. The NE sensitivity genes Tsn1, Snn3-B1, and Snn1 mapped to chromosome arms 5BL, 5BS, and 1BS, respectively (Supplemental Table S1), as expected.

Figure 1.

Leaves of Sumai 3 and CS-DIC 5B infiltrated with SnTox3, SnTox1, and SnToxA. Sumai 3 is sensitive to SnTox3 and SnToxA and insensitive to SnTox1, whereas CS-DIC 5B is sensitive to SnTox1 and insensitive to SnTox3 and SnToxA.

Analysis of Inverse Gene-For-Gene Interactions using P. nodorum Isolates that Produce All Three NEs

The P. nodorum isolates Sn4, Sn5, and SN15, all of which are known to harbor the SnToxA, SnTox1, and SnTox3 genes, were inoculated onto the CDS population to evaluate the relative effects of compatible Tsn1-SnToxA, Snn1-SnTox1, and Snn3-B1-SnTox3 interactions in causing SNB. At least three replications of inoculations for each isolate were conducted, and Bartlett’s χ² tests for homogeneity between replicates within each isolate indicated that the variance among the replicates was not significantly different (Supplemental Table S3). Therefore, mean reaction type scores for each isolate were used for further analysis. The reaction types were obtained using the 0–5 lesion type scale, where 0 = highly resistant and 5 = highly susceptible, as defined in Liu et al. (2004b).

P. nodorum Isolate Sn4

For Sn4, CS-DIC 5B had an average reaction type score of 2.10 (moderately resistant) and Sumai 3 had an average score of 3.80 (susceptible; Fig. 2; Tables 1 and 2; Supplemental Fig. S1). The average reaction type score for the CDS population was 2.96, and individuals in the population had average reaction types that ranged from 1.38 to 4.00 (Table 1; Supplemental Fig. S1).

Figure 2.

Leaves of CS-DIC 5B and Sumai 3 inoculated with different Parastagonospora nodorum isolates. CS-DIC 5B has the NE sensitivity gene Snn1, whereas Sumai 3 has Snn3-B1 and Tsn1. P. nodorum isolates Sn4, Sn5, and SN15 contain the NE genes SnTox1, SnTox3, and SnToxA.

Table 1. Average lesion-type reactions of the parental lines CS-DIC 5B and Sumai 3, along with the CDS population average and range to P. nodorum isolates Sn4, Sn5, SN15, Sn2000, and Sn2000KO6-1.

Isolate	CS-DIC 5B	Sumai3	Population Average	Population Range
Sn4 (SnTox1, SnTox3, SnToxA)	2.10	3.80	2.96	1.38-4.00
Sn5 (SnTox1, SnTox3, SnToxA)	1.25	3.25	2.80	1.13-4.00
SN15 (SnTox1, SnTox3, SnToxA)	2.00	2.67	2.32	0.50-4.50
Sn2000 (SnTox1, SnToxA)	2.00	2.56	2.24	0.06-4.07
Sn2000KO6 (SnTox1)	2.10	0.60	2.02	0.00-4.00

Table 2. The different genotypic classes in the CS-DIC 5B × Sumai 3 recombinant inbred population and their average reaction score to the P. nodorum isolates Sn4, Sn5, and SN15.

Genotypea	No. RIL	Sn4 Average Reaction Typeb	Sn5 Average Reaction Type	SN15 Average Reaction Type
CS-DIC 5B	—c	2.00 ± 0.41	1.25 ± 0.29	2.00 ± 0.25
Sumai 3	—	3.63 ± 0.38	3.25 ± 0.29	2.67 ± 0.29
Snn1/Snn3-B1/Tsn1	14	3.18a	3.04bcd	2.53ab
snn1/snn3-B1/tsn1	13	2.30b	2.09e	1.08d
Snn1/Snn3-B1/tsn1	18	3.07a	2.88cd	2.47bc
Snn1/snn3-B1/Tsn1	9	3.32a	3.08bc	2.46bc
Snn1/snn3-B1/tsn1	17	2.41b	1.98e	2.12c
snn1/Snn3-B1/tsn1	15	2.97b	2.76d	2.24bc
snn1/Snn3-B1/Tsn1	24	3.34a	3.39a	2.92a
snn1/snn3-B1/Tsn1	8	3.02a	3.20ab	2.20bc
lsd0.05	—	0.39	0.29	0.40

^aFor clarity, gene symbols in bold represent the NE sensitivity allele.

^bNumbers followed by the same letter in the same column are not significantly different at the 0.05 level of probability.

^c—, Nonapplicable

To analyze the effects of different combinations of the interactions together and individually, the 118 RILs were divided into eight genotypic classes based on their allelic compositions at the Tsn1, Snn1, and Snn3-B1 loci (Table 2). The reaction type mean for lines with Snn1 as the only NE sensitivity gene (Snn1/snn3-B1/tsn1 lines) was not significantly different from lines with no NE sensitivity genes (snn1/snn3-B1/tsn1), suggesting that the Snn1-SnTox1 interaction did not significantly contribute to disease caused by Sn4. However, lines with only Tsn1 (snn1/snn3-B1/Tsn1) and lines with only Snn3-B1 (snn1/Snn3-B1/tsn1) were significantly more susceptible than lines containing no NE sensitivity genes (snn1/snn3-B1/tsn1), but the reaction type means for SnTox3 and/or SnToxA sensitive lines (lines with Snn3-B1 and/or Tsn1) were not significantly different from each other (P < 0.05). These results suggest that the Tsn1-SnToxA and Snn3-B1-SnTox3 interactions played significant roles in SNB caused by Sn4, but their effects were not additive because the presence of both Snn3-B1 and Tsn1 did not make plants significantly more susceptible than plants with one of the two genes.

A second method of analyzing the effects of compatible interactions was conducted using the QTL identification approach known as composite interval mapping (CIM). Analysis of SNB reaction type means caused by Sn4 indicated that the Snn3-B1 and Tsn1 loci were both significantly associated with SNB susceptibility (Fig. 3; Table 3). The Snn3-B1 locus had a logarithm of the odds (LOD) of 4.17 and explained 17.3% of the disease variation, and Tsn1 had a LOD of 4.13 and explained 19.2% of the disease variation. The Snn1 locus was not significantly associated with reaction to Sn4, which agreed with the average reaction type analysis of the different genotypic combinations (Table 2).

Figure 3.

Composite interval regression maps of chromosomes 1B and 5B in the CS-DIC 5B × Sumai 3 recombinant inbred population containing QTL associated with Parastagonospora nodorum isolates Sn4, Sn5, and SN15. A cM scale is indicated on the left of the image. The critical LOD threshold is indicated by the dotted lines, and the LOD scale is on the top along the x axis.

Table 3. Composite multiple interval mapping analysis of susceptibility to SNB caused by P. nodorum isolates Sn4, Sn5, and SN15 in the CS-DIC 5B × Sumai 3 population.

Gene	Chromosome Arm	Genetic position (cM)	LODa			R2b			Source
			Sn4	Sn5	SN15	Sn4	Sn5	SN15
Snn1	1BS	1.4	0.31	0.54	1.72	0.002	0.039	0.020	CS-DIC 5B
Snn3-B1	5BS	1.2	4.17	5.27	5.58	0.173	0.204	0.207	Sumai 3
Tsn1	5BL	63.9	4.13	9.59	4.43	0.192	0.324	0.171	Sumai 3

^aLOD, determined by the execution of 1,000 permutations on marker and phenotypic datasets; cutoff value yielded was 3.25 for detection of significant QTL.

^bR² = coefficient of determination.

P. nodorum Isolate Sn5

CS-DIC 5B and Sumai 3 had average disease reaction scores of 1.25 (resistant) and 3.25 (susceptible) to Sn5, respectively (Fig. 2; Tables 1 and 2; Supplemental Fig. S1). The CDS population had an average reaction score of 2.80 with individuals in the population ranging from 1.13 to 4.00 (Table 1; Supplemental Fig. S1).

Analysis of the reaction type means of the eight genotypic classes revealed the Snn1/snn3-B1/tsn1 lines were not significantly different in their reaction to Sn5 compared with the lines with none of the NE sensitivity genes (snn1/snn3-B1/tsn1) indicating that, as with Sn4, the Snn1-SnTox1 interaction did not play a significant role in the development of SNB (Table 2). However, lines with only Tsn1 (snn1/snn3-B1/tsn1) or only Snn3-B1 (snn1/Snn3-B1/tsn1) were significantly more susceptible than the snn1/snn3-B1/tsn1 lines, and lines containing only Tsn1 were significantly more susceptible than lines with only Snn3-B1. However, in the presence of Snn1, lines with either Tsn1 or Snn3-B1 were not significantly different. In addition, lines with Snn3-B1 and Tsn1 but not Snn1 (snn1/Snn3-B1/Tsn1) were significantly more susceptible than lines with all three NE sensitivity genes. These results suggest that, like Sn4, the Snn3-B1-SnTox3 and Tsn1-SnToxA interactions play significant roles in SNB development, and the Snn1-SnTox1 interaction does not. However, the role of Snn1-SnTox1 is more complicated in Sn5 and may contribute to resistance, or suppression of susceptibility, in the presence of some other interactions.

CIM analysis indicated that both Tsn1 and Snn3-B1 were significantly associated with SNB susceptibility (Fig. 3; Table 3). Tsn1 had a LOD of 9.59 and explained 32.4% of the disease variation, whereas Snn3-B1 had a LOD of 5.27 and explained 20.4% of the disease variation.

P. nodorum Isolate SN15

CS-DIC 5B and Sumai 3 had average disease reaction scores of 2.00 (moderately resistant) and 2.67 (moderately susceptible) to SN15, respectively (Fig. 2; Tables 1 and 2; Supplemental Fig. S1). The CDS population had an average disease score of 2.32, and average disease scores ranged from 0.50 to 4.50 (Table 1; Supplemental Fig. S1).

RILs containing at least one NE gene were significantly more susceptible to SN15 than lines with none of the NE sensitivity genes (snn1/snn3-B1/tsn1), indicating that all three interactions played significant roles in the development of SNB (Table 2). However, lines with both Snn3-B1 and Tsn1, but not Snn1 (snn1/Snn3-B1/Tsn1), were as susceptible as lines with all three NE sensitivity genes, and lines with Snn1 in addition to one other NE sensitivity gene (Snn1/Snn3-B1/tsn1 and Snn1/snn3-B1/Tsn1) were not significantly different from lines with only a single NE sensitivity gene. This indicated that the Tsn1-SnToxA and Snn3-B1-SnTox3 interactions played significant roles and their effects were additive, but the Snn1-SnTox1 interaction was associated with SNB development only in the absence of the other two interactions.

CIM analysis showed that Tsn1 and Snn3-B1, but not Snn1, were significantly associated with SNB susceptibility (Fig. 3; Table 3). Snn3-B1 had a LOD of 5.58 and explained 20.7% of the disease variation, whereas Tsn1 had a LOD of 4.43 and explained 17.1% of the disease variation.

Analysis of Interactions using P. nodorum Isolate Sn2000 and an Sn2000 SnToxA-Knockout Mutant

The CDS population was evaluated for reaction to SNB caused by P. nodorum isolate Sn2000, which contains the NE genes SnTox1 and SnToxA (Liu et al., 2004b, 2012). CS-DIC 5B and Sumai 3 had average disease reaction types of 2.00 (moderately resistant) and 2.56 (moderately susceptible), respectively (Fig. 4; Tables 1, 4, and 5; Supplemental Fig. S2). The CDS population had an average reaction type of 2.24 and a population range of 0.06 to 4.07 (Table 1; Supplemental Fig. S2).

Figure 4.

Leaves of CS-DIC 5B and Sumai 3 inoculated with different Parastagonospora nodorum isolates. CS-DIC 5B has the NE sensitivity gene Snn1, whereas Sumai 3 has Snn3-B1 and Tsn1. P. nodorum isolate Sn2000 contains the NE genes SnTox1 and SnToxA, whereas Sn2000KO6 contains the NE gene SnTox1.

Table 4. The different genotypic classes in the CS-DIC 5B × Sumai 3 recombinant inbred population and their average reaction score to the P. nodorum isolate Sn2000.

Genotype	No. RIL	Sn2000 Average Reaction Typea
CS-DIC 5B	—b	2.00 ± 0.60
Sumai 3	—	2.56 ± 0.86
Snn1/Tsn1	23	2.66ab
snn1/tsn1	28	1.12c
Snn1/tsn1	35	2.27b
snn1/Tsn1	32	2.88a
lsd0.05	—	0.47

^aNumbers followed by the same letter in the same column are not significantly different at the 0.05 level of probability.

^b—, Nonapplicable

Table 5. Single-trait multiple interval mapping analysis of susceptibility to SNB caused by P. nodorum isolates Sn2000 and Sn2000KO6-1 in the CS-DIC 5B × Sumai 3 population.

Gene	Chromosome arm	Genetic position (cM)	LODa		R2b		Source
			Sn2000	Sn2000KO6	Sn2000	Sn2000KO6
Snn1	1BS	1.4	5.42	12.90	0.071	0.302	CS-DIC 5B
QSnb.fcu-4B	4BL	55.7	4.86	15.60	0.101	0.344	CS-DIC 5B
Tsn1	5BL	63.9	14.18	—c	0.327	—	Sumai 3

^aLOD, determined by the execution of 1,000 permutations on marker and phenotypic datasets; cutoff value yielded was 3.25 for detection of significant QTL.

^bR² = coefficient of determination.

^c—, Nonsignificant.

Analysis of the reaction type means of the four genotypic classes revealed that lines containing at least one NE sensitivity gene (Tsn1 and/or Snn1) were significantly more susceptible to Sn2000 than lines containing neither of the NE sensitivity genes (snn1/tsn1), indicating that both the Snn1-SnTox1 and Tsn1-SnToxA interactions play a significant role in the development of SNB (Table 4). Lines with only Tsn1 were significantly more susceptible than lines with only Snn1, indicating that the Tsn1-SnToxA interaction played a more significant role than the Snn1-SnTox1 interaction. Lines with either Tsn1 or Snn1 were not significantly different than lines with both genes (Snn1/Tsn1) indicating that the additive effects of these two interactions were not significant in SNB caused by Sn2000.

CIM analysis of the CDS population infected with isolate Sn2000 revealed three loci significantly associated with SNB susceptibility (Fig. 5; Table 5). The locus with the largest effect was Tsn1, which had a LOD of 14.18 and explained a total of 32.7% of the disease variation. The locus with the second largest effect was the Snn1 locus, which had a LOD of 5.42 and explained 7.1% of the disease variation. QSnb.fcu-4B was also associated with susceptibility to Sn2000 with a LOD of 4.89, and it explained 10.1% of the disease variation.

Figure 5.

Composite interval regression maps of chromosomes 1B and 5B in the CS-DIC 5B × Sumai 3 recombinant inbred population containing QTL associated with Parastagonospora nodorum isolates Sn2000 and Sn2000KO6-1. A cM scale is indicated on the left of the image. The critical LOD threshold is indicated by the dotted lines, and the LOD scale is on the top along the x axis.

Liu et al. (2012) generated an SnToxA-knockout strain of Sn2000, designated as Sn2000KO6-1, and it was used to evaluate the CDS population to determine the effects of the Snn1-SnTox1 interaction in the absence of a compatible Tsn1-SnToxA interaction. CS-DIC 5B and Sumai 3 had average reaction types of 2.10 (moderately resistant) and 0.60 (resistant), respectively (Fig. 4; Tables 1 and 6; Supplemental Fig. S2). The CDS population had an average disease score of 2.02 and a population range of 0.00 to 4.00 (Table 1; Supplemental Fig. S2).

Table 6. The different genotypic classes in the CS-DIC 5B × Sumai 3 recombinant inbred population and their average reaction score to the P. nodorum isolate Sn2000KO6-1.

Genotype	No. RIL	Sn2000KO6 Average Reaction Typea
CS-DIC 5B	—b	2.10 ± 0.42
Sumai 3	—	0.60 ± 1.08
Snn1	58	2.51a
snn1	60	1.57b
lsd0.05	—	0.36

^aNumbers followed by the same letter in the same column are not significantly different at the 0.05 level of probability.

^b—, Nonapplicable

Analysis of the reaction type means of the two genotypic classes revealed that lines containing Snn1 were significantly more susceptible in their reaction to Sn2000KO6-1 compared with lines without (Table 6), indicating that the Snn1-SnTox1 interaction played a significant role in the development of SNB. CIM analysis of the CDS population infected with isolate Sn2000KO6-1 revealed that the Snn1 locus had a LOD of 12.90 and explained 30.2% of the disease variation (Fig. 5; Table 5). QSnb.fcu-4B was also associated with Sn2000KO6-1 susceptibility with a LOD of 15.60 and explained 34.4% of the disease variation.

Expression Analysis of SnTox1 in Planta using RNA Sequencing

RNA sequencing analysis was conducted to determine whether SnTox1 expression differed between P. nodorum isolates Sn4, Sn5, and SN15. Previous research on SnTox1 expression has shown that expression is greatest around 48 h and then begins to decline (Liu et al., 2012); therefore, we inoculated the RIL CDS37, which contains all three NE sensitivity genes (Snn1/Snn3-B1/Tsn1), and began tissue collection at 48 h post inoculation (hpi). RNA sequencing of P. nodorum isolates Sn4, Sn5, and SN15 at in planta postinoculation time points of 48, 72, and 96 hpi yielded an average of 13,005,965 reads per library and ranged from 1,093,752 to 25,560,987 reads (Supplemental Table S4). The majority of reads corresponded to wheat mRNA and, therefore, did not map to the reference genome of isolate LDN03-Sn4. The average percent of reads mapping to the fungal genome ranged from 0.03% to 0.53% with an average of 0.11%. Differential expression analysis revealed that SnTox1 was significantly up-regulated in isolate SN15 compared with isolates Sn4 and Sn5 at 48 hpi (P = 0.0031 and P = 0.0033, respectively; Fig. 6). No expression was detected in isolates Sn4 and Sn5 at 48 hpi, whereas SN15 had a normalized read count of 14.06 (Fig. 6). No significant differences in SnTox1 expression were detected at 72 hpi or 96 hpi between the three isolates. At 72 hpi, SN15, Sn4, and Sn5 had normalized read counts of 3.13, 4.87, and 1.02, respectively. At 96 hpi, SN15, Sn4, and Sn5 had normalized read counts of 1.70, 2.18, and 1.03, respectively. These results indicate P. nodorum isolate SN15 expresses SnTox1 at comparatively higher levels during early in planta time points, but expression then declines over time to levels comparable with those observed in isolates Sn4 and Sn5.

Figure 6.

RNA sequencing–evaluated expression of SnTox1 in the susceptible line CDS37 (Snn1/Snn3-B1/Tsn1) inoculated with Parastagonospora nodorum isolates SN15, Sn4, and Sn5 at 48, 72, and 96 h after inoculation. The expression level is shown as the average normalized read count (n = 2) from RNA sequencing. se bars are shown. *Significant differential expression calculated from a Wald test implemented in DESeq2 (P < 0.01) compared with other isolates within the same time point.

Comparison with all genes encoding predicted secreted proteins, as well as predicted effectors, revealed that SnTox1 is relatively highly expressed in isolate SN15 at 48 h postinoculation. Out of the total 1239 genes encoding predicted secreted proteins, expression of 305 genes was detected with SnTox1 being the second highest expressed gene, behind CJJ16_03445, a glycosyl hydrolase. Among all 219 genes encoding predicted effectors, expression of a total of 38 genes was detected with SnTox1 being expressed the highest (Supplemental File S1). These results indicate P. nodorum isolate SN15 is expending considerable energy in the expression of SnTox1 at 48 hpi compared with the rest of the secretome and effectorome.

SnTox1 Expression was Increased in the Absence of SnToxA

To investigate whether SnTox1 expression increases in the SnToxA-disrupted strain Sn2000KO6-1 compared with the wild-type isolate Sn2000, or whether the change in QTL magnitude was due to the absence of SnToxA, we measured SnTox1 expression in both Sn2000-and Sn2000KO6-1–infected plants using reverse transcription-quantitative PCR (RT-qPCR). Leaf tissue samples were collected at 12, 24, 48, 72, and 96 hpi for the RIL CDS37, which contains all three genes (Snn1/Snn3-B1/Tsn1). SnToxA expression was also evaluated and was only expressed in Sn2000 and not Sn2000KO6-1, which lacks SnToxA (data not shown). SnTox1 expression was significantly increased in CDS37 inoculated with Sn2000KO6-1 compared with Sn2000-inoculated plants, except for at the 12 hpi time point (Fig. 7). The highest level of SnTox1 expression occurred at 72 hpi, where expression was increased 4-fold in plants inoculated with Sn2000KO6-1 compared with the wild-type Sn2000 (Fig. 7).

Figure 7.

RT-qPCR–evaluated transcriptional expression of SnTox1 in the susceptible line CDS37 (Snn1/Snn3-B1/Tsn1) inoculated with Parastagonospora nodorum isolates Sn2000 and Sn2000KO6-1 at 12, 24, 48, 72, and 96 h after inoculation. The expression of SnTox1 was normalized to the expression of Act1. Average gene expression was calculated from six biological samples in three technical replicates (n = 6). se bars are shown. *Significant differences (P < 0.05) between Sn2000 and Sn2000KO6-1 at that time point using a t test.

Promoter Region and Protein Sequence Comparison

The promoter region of SnTox1 was extracted from the SN15 and Sn4 reference genome sequences (Syme et al., 2016; Richards et al., 2018). Due to the lack of a contiguous genome sequence in Sn5, the SnTox1 region of that isolate was PCR amplified and sequenced using the Sanger method. The resulting sequence alignment included 915 bp upstream of the start codon and 138 bp of coding sequence. Sequence comparison revealed a 401-bp deletion in the putative promoter region in isolate SN15 located 268 bp upstream of the start codon (Supplemental Fig. S3). Protein alignments revealed that Sn4 and Sn5 harbor identical SnTox1 isoforms, whereas the isoform present in SN15 contained seven amino acid substitutions (Supplemental Fig. S4). These results indicate that extensive polymorphism within the putative regulatory region or between protein isoforms may account for the differences in expression observed for SnTox1 in isolate SN15.

Although no significant differences in the expression of SnToxA and SnTox3 were observed (Supplemental Fig. S5; Supplemental Fig. S6), their respective putative promoter regions and protein sequences were also compared among the three isolates. A 795-bp region upstream of the start codon of SnToxA was aligned between the three isolates, revealing a total of four SNPs and a single base pair deletion (Supplemental Fig. S7). Additionally, a 43-bp insertion was found in isolate SN15, 369 bp upstream of the start codon. The protein sequence of SnToxA was conserved between isolates Sn4 and Sn5; however, the SN15 isoform differed by two amino acids (Supplemental Fig. S8). A higher level of conservation between isolates was observed at the SnTox3 locus. In the 822 bp upstream of the start codon, isolates Sn4 and Sn5 harbored identical sequences (Supplemental Fig. S9). The same genomic region in isolate SN15 differed only by a single SNP. Additionally, all three isolates share identical protein isoforms (Supplemental Fig. S10).

DISCUSSION

Relative Effects of the Three Compatible Host-NE Interactions

The relationships between biotrophic pathogens and their hosts have been studied extensively with the occurrence of multiple gene-for-gene interactions normally leading to a similar level of resistance as a single interaction. Less research is available on how host susceptibility gene-NE interactions differ between isolates in their importance for disease contribution and how different NE genes interact and regulate one another. In this study, we evaluated the relative effects of the Snn1-SnTox1, Snn3-B1-SnTox3, and Tsn1-SnToxA interactions in a single biparental population using multiple fungal isolates/strains. The effects of the three host gene-NE interactions among the three isolates that produced all three NEs were similar in some ways, but there were obvious and interesting dissimilarities as well, which are summarized in Supplemental Table S5. The effects of the Snn3-B1-SnTox3 and Tsn1-SnToxA interactions were associated with SNB caused by all three isolates. However, the effects of the Snn1-SnTox1 were more subtle; therefore, it was investigated in more detail through SnTox1 expression analysis using RNA sequencing and RT-qPCR.

Snn3-B1-SnTox3 and Tsn1-SnToxA

The effects of the Snn3-B1-SnTox3 and Tsn1-SnToxA interactions were significantly associated with SNB caused by the three isolates that produced all three NEs. However, the effects of these two interactions differed among these isolates. The effects of the Snn3-B1-SnTox3 interaction on disease were fairly consistent for the three isolates explaining from 17.3% to 20.7% of the variation, whereas the effects of the Tsn1-SnToxA interaction varied much more ranging from 17.1% to 32.4%. Faris et al. (2011) evaluated the effects of the Tsn1-SnToxA interaction in SNB development caused by isolates Sn4 and Sn5 in the BR34 × Grandin population and showed that the interaction explained more of the variation in SNB caused by Sn5 than Sn4, which agrees with the results of the current study where Tsn1-SnToxA explained 17.9% and 32.4% of the variation for Sn4 and Sn5, respectively.

Other dissimilarities in the relative effects of the Snn3-B1-SnTox3 and Tsn1-SnToxA interactions among isolates were observed in the genotypic classification analyses. The two interactions contributed equally to disease caused by Sn4 and SN15, and Tsn1-SnToxA contributed more to disease caused by Sn5 than did Snn3-B1-SnTox3, but only in the absence of the Snn1-SnTox1 interaction.

Previous research involving these three isolates in the BR34 × Grandin population showed that the effects of the Snn2-SnTox2 and Tsn1-SnToxA interactions were additive, and that the Snn2-SnTox2 interaction was epistatic to Snn3-B1-SnTox3 (Friesen et al., 2007, 2008; Faris et al., 2011). Phan et al. (2016) recently showed in a study using isolate SN15 that Snn1-SnTox1 was epistatic to Snn3-B1-SnTox3. However, interactions between Snn3-B1-SnTox3 and Tsn1-SnToxA have not been evaluated. Here, our results showed that the effects of Snn3-B1-SnTox3 and Tsn1-SnToxA were not additive except for disease caused by SN15 in the absence of a compatible Snn1-SnTox1 interaction. Together, these results indicate that host-NE gene interactions are complex and range from additive to epistatic. From our sequencing data of SnTox3 within each isolate, there are no differences in the promotor and coding region sequences, which correspond to the observed contribution of the Snn3-B1-SnTox3 interactions being similar in each isolate. Our initial hypothesis for the difference in SnToxA expression and contribution to SNB between isolates Sn4, Sn5, and SN15, with Sn5 having significantly higher expression at 48 hpi and a higher QTL magnitude, was that Sn5 contains a different protein isoform or promotor sequence of SnToxA than the other two isolates. However, our sequencing analysis between the three isolates showed no differences in the promotor and amino acid sequences in SnToxA, with only very minor differences in the sequences for SN15. The difference in expression and disease contribution from the Tsn1-SnToxA interaction when plants are infected with Sn5 may be due to transposable elements and other genetic factors near SnToxA, which will be discussed below. However, no full genome sequence of Sn5 exists. Additionally, other host-pathogen interactions may be involved in regulating virulence dictated by host sensitivity gene-NE interactions.

Snn1-SnTox1

The Snn1-SnTox1 interaction did not contribute to SNB caused by Sn4 or Sn5, and in SN15, it contributed to disease only in the absence of compatible Snn3-B1-SnTox3 and Tsn1-SnToxA interactions indicating that the latter two are epistatic to Snn1-SnTox1. This is contrary to the findings of Phan et al. (2016) who, using the same isolate, reported that the Snn1-SnTox1 interaction was epistatic to Snn3-B1-SnTox3. A possible explanation for the differences between the two studies is likely the presence of different host alleles at the Snn1 and/or Snn3-B1 loci, with Sumai 3 containing an allele associated with a stronger reaction to a compatible Snn3-B1-SnTox3 interaction (Shi et al., 2016a) than the Snn3-B1 donor parent in Phan et al. (2016).

Not only did the Snn1-SnTox1 interaction contribute little to susceptibility, some evidence suggested that the involvement of the Snn3-B1-SnTox3 and Tsn1-SnToxA interactions in disease was reduced in the presence of Snn1-SnTox1. For example, the Tsn1-SnToxA interaction contributed more than Snn3-B1-SnTox3 to SNB caused by Sn5, but only in the absence of Snn1-SnTox1. Also observed in the Sn5 analysis, the presence of Snn3-B1-SnTox3 and Tsn1-SnToxA led to more disease than when all three interactions were present, indicating that a compatible Snn1-SnTox1 interaction actually reduced disease under that scenario. Also, the finding that Snn3-B1-SnTox3 and Tsn1-SnToxA were additive only in the absence of Snn1-SnTox1 in SNB caused by SN15 would suggest further that a compatible Snn1-SnTox1 may inhibit disease when in the presence of some other interactions.

Another reason for the discrepancy between the Phan et al. (2016) study and ours may pertain to the nontypical phenotype of the compatible Snn1-SnTox1 interaction. The Snn1-SnTox1 interaction is unique in that it not only produces lesions, but also necrotic flecks that sometimes, but not always, lead to larger lesions (Liu et al., 2004b, 2012, 2016; Shi et al., 2016b). The flecks do not fit the descriptions of lesions in the 0–5 lesion type rating scale used in this study and are therefore largely unaccounted for even though they undoubtedly lead to a decrease in the photosynthetic material on the leaf. Perhaps digital imaging of infected leaves or similar methods would lead to more accurate scores and hence the association of the Snn1-SnTox1 interaction with the development of SNB. It is interesting to note that Xu et al. (2004) found a significant correlation between SNB resistance and the flecking caused by the Snn1-SnTox1 interaction in synthetic hexaploid wheat lines. This would lend support to our findings noted above that Snn1-SnTox1 may sometimes contribute to resistance. However, it has been clearly demonstrated that SnTox1 is an NE that, when recognized by the Snn1 gene product, leads to host responses that result in NETS in an inverse gene-for-gene fashion (Liu et al., 2004a, 2012, 2016; Chu et al., 2010; Shi et al., 2016b). It may be that the putative PTI pathway activated by a compatible Snn1-SnTox1 interaction may antagonize or inhibit the response of the ETI pathways activated by other interactions such as Snn3-B1-SnTox3 and Tsn1-SnToxA. More work is needed to decipher the molecular interplay of these interactions.

SnTox1 Is Differentially Expressed between Isolates as Shown using RNA Sequencing

To examine whether the Snn1-SnTox1 interaction was being masked by the other interactions or whether SnTox1 was expressed at lower levels in Sn4 and Sn5 compared with SN15, we performed an RNA sequencing experiment. The data from RNA sequencing indicate that when comparing the three isolates, SnTox1 expression was highest in SN15 at 48 hpi but was not significantly different between the three isolates at 72 and 96 hpi. This corroborates our genetic analyses where the Snn1/snn3-B1/tsn1 lines had significantly more disease than snn1/snn3-B1/tsn1 lines. These findings indicate that the level of disease contributed by a compatible Snn1-SnTox1 interaction varies among these three isolates and is likely due to differences in SnTox1 expression in the pathogen and not host differences. Previously, it was shown that SnTox1 acts as a dual-function protein and protects the fungus from wheat chitinases during the infection process (Liu et al., 2016). Isolates that contain SnTox1 may express this NE at low levels for protection; however, the pathogen may down-regulate SnTox1 to prevent it from playing an important role in eliciting a host response. This may be occurring in P. nodorum isolates Sn4 and Sn5. However, we observed that SnTox1 expression was higher in SN15 along with the Snn1-SnTox1 interaction contributing significantly to disease. Our analysis of the SN15 secretome and effectorome showed that SnTox1 is highly expressed among all secreted proteins and is the highest expressed effector at 48 hpi. Some isolates, such as SN15, may express SnTox1 at high levels so the SnTox1 protein can provide both protection along with eliciting disease, whereas other isolates may use other NE for eliciting disease. Potentially, pathogens may have the ability to ‘determine’ which host sensitivity genes are present through feedback mechanisms and increase the expression of the corresponding NE genes to maximize disease while balancing the cost of expression. One explanation for the differences in SnTox1 expression between isolates may be epigenetic regulation. Recently, Haueisen et al. (2019) examined the expression profiles after infection of three isolates of Zymoseptoria tritici, a hemibiotrophic wheat pathogen. Many of the differentially expressed genes between the isolates were effector candidate genes and were located within 2 kb of transposable elements. Located next to SnTox1 is a short truncated molly-type retrotransposable element (Liu et al., 2012). Whether epigenetic transcriptional regulation of the transposable element close to SnTox1 plays a role in influencing SnTox1 expression would need to be further studied to validate or disprove if this type of regulation network is also occurring in P. nodorum. Another potential mechanism governing the differences in SnTox1 expression may be transcription factors (TFs), which are further discussed below.

Differences in NE Gene Promoter and Protein Sequences May Influence the Importance of Different Sensitive Gene-NE Interactions in Causing Disease and Variation between Isolates

Amino acid and promoter sequence alignments for SnTox1 revealed multiple polymorphisms in SN15 compared with Sn4 and Sn5, with the latter two being nearly identical. The 401-bp deletion in the promotor region of the SN15 SnTox1 gene might be important for transcription factor binding, with the deleted sequence promoting higher SnTox1 expression. Additionally, the SnTox1 protein isoform differed between SN15 compared with Sn4 and Sn5, with seven amino acid changes between them. These changes could influence SnTox1 protein binding with the Snn1 protein, with the SN15 isoform having a higher binding affinity. Further studies are needed to test these hypotheses, along with identifying transcription factors and other regulators of SnTox1.

Effects of the Inverse Gene-For-Gene Interactions using the NE Gene-Disrupted Mutant Isolates

Infection of the CDS population with isolate Sn2000, which does not have SnTox3, indicated that both the Snn1-SnTox1 and Tsn1-SnToxA interactions were associated with SNB susceptibility. This agreed with the findings of Chu et al. (2010) who used the same isolate on a different wheat population. However, Chu et al. (2010) showed that the effects of the two interactions were largely additive, whereas they were not additive in the CDS population. Chu et al. (2010) also inoculated their population with the SnToxA-knockout isolate Sn2000KO6-1 and found that the effects of the Snn1-SnTox1 interaction increased from explaining 22% of the variation in the wild-type isolate to 50% in the SnToxA-knockout isolate. Their findings agree with the results of this research where the Snn1-SnTox1 interaction went from explaining 7.1% of the variation in Sn2000 to 30.2% in the SnToxA-knockout isolate. The increase in necrotic flecking and disease in Snn1 plants infected with Sn2000KO6-1 compared with the wild type was a result of increased levels of SnTox1 transcription, which agrees with previous findings that NE gene transcriptional expression levels dictate the levels of NETS that occur in wheat-P. nodorum inverse gene-for-gene interactions (Faris et al., 2011; Phan et al., 2016; Virdi et al., 2016).

We demonstrated that when SnToxA is eliminated from Sn2000, the expression of SnTox1 is significantly increased. Manning and Ciuffetti (2015) reported that ToxA is epistatic to other effector genes in P. tritici-repentis. Further research is needed to determine whether SnToxA is truly epistatic to SnTox1 or if other mechanisms in the pathogen are leading to changes in SnTox1 expression in the absence of SnToxA.

Previous research has suggested that SnTox1 is epistatic to SnTox3 (Phan et al., 2016). Further studies are needed to determine whether this was an isolate-specific case or if this is true for all P. nodorum isolates that contain SnTox1 and SnTox3, because the opposite was observed in this study. Recently published literature has reported that multiple transcription factors are responsible for NE gene regulation. The C2H2 zinc finger transcription factors P. nodorumCon7 (PnCon7), a zinc finger transcription factor, and Stagonospora nodorum StuA, an ASM-1, Phd1, StuA, EFG1, and Sok2 domain transcription factor, have been reported to be regulators of SnTox3 (IpCho et al., 2010; Lin et al., 2018). Lin et al. (2018) found through silencing of PnCon7, there was a corresponding decrease in SnTox3 expression. Although PnCon7 was found to not be directly correlated with SnTox1 expression, silencing of PnCon7 led to decreased SnTox1 and SnToxA expression. Additionally, the zinc finger transcription factor PnPf2 is a conserved signaling component that regulates both SnToxA and SnTox3 expression (Rybak et al., 2017).

It might be possible that some TFs regulate multiple NE genes, or that there are global regulators of NE genes. When one NE gene is eliminated from an isolate, there would be up-regulation of the other NE genes due to fewer promoter sites competing for TF binding. To test this hypothesis, expression levels for multiple NE genes in multiple isolates would need to be determined and compared with knockout strains of different combinations of NE genes. Additionally, no common TF that binds to a conserved promoter region in SnTox1, SnTox3, and SnToxA has yet been identified. As stated, whether NE genes are truly regulating one another (i.e. are epistatic to one another) or if they are controlled by global regulators has yet to be determined. Findings from this study along with previously published research suggest that regulation within P. nodorum for NE genes is complex and that we are just beginning to unravel this conundrum.

CONCLUSION

Overall, the results of this study show that all three host sensitivity gene-NE interactions can contribute to the development of SNB. However, their relative effects on disease expression varied from additive to epistatic, and in no case did all three compatible interactions simultaneously contribute significantly to disease development. This observation, along with the results obtained from evaluating the NE gene-knockout mutant isolate and NE gene expression, suggests that the pathogen likely has a target threshold for NE production. It is possible that, although the fungus may harbor numerous NE genes, which ones are expressed and to what levels may depend on the repertoire of NE sensitivity genes present in the host under attack. If the pathogen ‘determines’ that the host harbors multiple NE sensitivity genes that it can exploit because it harbors the corresponding NE genes, then perhaps the pathogen strives to find a balance between the amount of energy it needs to expend to express NE genes and what it needs to do to propagate and complete its lifecycle. In other words, it is probably not efficient for the pathogen to always produce numerous NEs if it can achieve infection and sporulation by producing one or two. Clearly, these host-pathogen interactions are complex and can be affected by multiple factors, and more work related to the interplay between interactions, pathogen NE genes, and host sensitivity genes is required.

From a more applied perspective, the results of this research reiterate the importance of Snn1, Tsn1, and Snn3-B1 in conferring SNB susceptibility in wheat. We therefore continue to recommend that breeders remove the dominant susceptibility alleles from their materials, which can be done efficiently using molecular markers (Faris et al., 2010; Shi et al., 2016a, 2016b). Additionally, recent research (Gao et al., 2015; Phan et al., 2016; Rybak et al., 2017) along with the findings from this study are beginning to unravel the role NE gene expression plays in SNB development and severity along with the complex regulatory mechanisms of these NE genes.

MATERIALS AND METHODS

The RIL Mapping Population

An RIL population composed of 190 lines was developed from a cross between the hexaploid Triticum aestivum (wheat) line Sumai 3 and the Chinese Spring-Triticum turgidum ssp. dicoccoides chromosome 5B disomic substitution line (CS-DIC 5B), which are both landraces from China. Sumai 3 contains the Tsn1 and Snn3-B1 genes, which confer sensitivity to SnToxA and SnTox3, respectively, and CS-DIC 5B has the Snn1 gene and is therefore sensitive to SnTox1. Tsn1 was previously cloned by Faris et al. (2010) and is a present/absent variant that is present in Sumai 3 and null in CS-DIC 5B. Snn1 was previously cloned by Shi et al. (2016b), with Sumai 3 having the recessive allele and CS-DIC 5B having the dominant allele. Snn3-B1 has yet to be cloned. The RILs were developed using the single-seed descent method and were bulked at the F₇ generation with the population designated as CDS.

Infiltrations

The Parastagonospora nodorum genes SnTox1, SnToxA, and SnTox3 have previously been cloned and expressed in Pichia pastoris to produce NE-containing culture filtrates (Friesen et al., 2006; Liu et al., 2009, 2012). SnTox1, SnTox3, and SnToxA were obtained from Pichia pastoris cultures producing these NEs as described by Liu et al. (2009).

Plants were grown, infiltrated, and evaluated 5 d after infiltration as described in Zhang et al. (2011). Reaction types of 2 and 3 were considered sensitive, and 0 and 1 were insensitive. The experiment was performed twice and analyzed using a χ² analysis.

SSR and SNP Analysis

DNA was extracted from leaf tissue as described by Faris et al. (2000) and diluted to ∼200 ng/µL using distilled water. An SSR primer survey using parental DNA (CS-DIC 5B and Sumai3) was used to identify markers that reveal polymorphism between the parents. Markers for the survey were chosen based on previously published locations, which were obtained from the GrainGenes database (http://wheat.pw.usda.gov/GG2/index.shtml). SSR markers located on chromosome arms 1BS, 5BL, and 5BS, within the known vicinity of the Snn1, Tsn1, and Snn3-B1 genes, respectively (Faris et al., 1996; Liu et al., 2004b; Friesen et al., 2008), were considered the highest priority. Three to six additional SSR markers that detect loci on the other wheat chromosomes were selected to assist with assigning linkage groups to chromosomes. The polymorphic SSR markers were selected from the following libraries: WMC (Somers et al., 2004), WMS (marker designation = ‘gwm’; Röder et al., 1998), MAG (Xue et al., 2008), HBG (Torada et al., 2006), CFD (Sourdille et al., 2004), BARC (Song et al., 2005), GDM (Pestsova et al., 2000), HBE (Torada et al., 2006), HBD (Torada et al., 2006), PSP (Bryan et al., 1997), FCP (Reddy et al., 2008; Shi et al., 2015), CFA (Sourdille et al., 2004), and CFB (Sourdille et al., 2004).

DNA fragments were amplified using PCR and the markers chosen above. PCR reactions consisted of 200 ng of template DNA, 1× PCR buffer, 2 mm MgCl₂, 0.2 mm deoxynucleotide triphosphates, 4 pmol of each primer, and 0.5 unit of Taq DNA polymerase, with diluted water added to a final volume of 10 µL. PCR was performed using a GeneAmp PCR system 9700 machine. The PCR cycle was as follows: 94°C for 5 min, cycle 35 times through: 30 s 94°C, 30 s 65–56°C, 90 s 72°C; finishing with one cycle for 7 min at 72°C and cooling to 4°C. PCR products were separated on 6% polyacrylamide gels, stained with GelRed nucleic acid gel stain, and scanned on a Typhoon FLA 9500 variable mode laser scanner (GE Healthcare Life Sciences).

The CDS population was also genotyped using a 9K iSelect Assay BeadChip (Cavanagh et al., 2013). A BeadStation and iScan instrument from Illumina were used for the assay. Clustering data were analyzed using GenomeStudio Polyploid Clustering Module from Illumina, Inc. (2013).

The SSR marker, NE infiltration, and SNP data were combined to develop genetic linkage maps of all 21 chromosomes. The computer software MapDisto version 1.7 (Lorieux, 2012) was used to assemble the linkage maps as described in Faris et al. (2014). The Kosambi mapping function was used to calculate the map distances (Kosambi, 1944).

Inoculations with P. nodorum Isolates

Methods for plant inoculation were as described by Friesen and Faris (2012). Conidia of P. nodorum isolates LDN03Sn4 (Sn4), BBC03Sn5 (Sn5), Sn2000, Sn2000KO6-1, and AuSN15 (SN15) were used to phenotype the population. Sn4, Sn5, and SN15 were previously found to contain SnToxA, SnTox1, and SnTox3 (Friesen et al., 2007; Hane et al., 2007; Faris et al., 2011; Gao et al., 2015; T.L. Friesen, unpublished data). Sn2000 contains SnToxA and SnTox1 (Liu et al., 2004b), and Sn2000KO6-1 contains SnTox1 (Liu et al., 2012; Supplemental Table S6).

Three plants per line were grown in plastic cones that were 3.8 cm in diameter and 21 cm deep (Stuewe and Sons, Inc.). A total of 118 RILs and the parental lines were inoculated in a completely randomized design. The susceptible wheat cultivar Alsen was grown on the outside borders of the racks to reduce any edge effect. Plants were inoculated following the methods in Friesen et al. (2007). Plants were scored at 7 d after inoculation using a 0 to 5 scoring scale described by Liu et al. (2004a) where 0 = highly resistant and 5 = highly susceptible.

Inoculations of each isolate were replicated at least three times with randomization between replicates (Supplemental Table S3). The homogeneity of variances among the replicates was determined by Barlett’s χ² test using the general linear model procedure in SAS (SAS Institute, 2011). The mean separation of the phenotypic means was determined using Fischer’s protected Least Significant Difference (lsd) at an alpha level of 0.05. Phenotypic scores from each replicate were combined to calculate an overall mean if the error of variance was homogenous between replicates.

QTL Analysis

QTL analysis was conducted using the computer software program QGene v 4.3.10 (Joehanes and Nelson, 2008). Composite interval mapping was used to quantify the effects of the Tsn1, Snn1, and Snn3-B1 loci in conferring susceptibility to the various isolates, and also to identify putative novel QTLs associated with resistance. The coefficient of determination (R²) was used to indicate the amount of variation explained by the QTLs and therefore provide an estimate of the contribution of each compatible host-NE interaction in the development of SNB. Critical LOD thresholds at the 0.05 and 0.01 levels of probability were determined using a permutation test with 1000 iterations.

RNA Sequencing

Plants of the RIL CDS37, which has the genotype Snn1/Snn3-B1/Tsn1, were grown and inoculated with Sn4, Sn5, SN15, Sn2000, and Sn2000KO6-1 as described above. Leaf tissue samples for each genotype and isolate combination were collected from the second leaf at 48, 72, and 96 h after inoculation and immediately frozen in liquid nitrogen and stored at −80° until mRNA extraction. Tissue from three samples for each isolate was pooled to make one technical replicate, with a total of two technical replicates used. mRNA was isolated using the Dynabeads mRNA Direct Kit (Life Technologies) following the manufacturer’s protocol. With use of purified mRNA as the input, strand nonspecific RNAseq libraries were prepared with the Illumina TruSeq RNA Sample Preparation v2 following the manufacturer’s recommended protocol. Fragment size distribution of the prepared RNAseq libraries was determined using an Agilent DNA chip on a bioanalyzer (Agilent). Quality and concentrations were determined using KAPA Library Quantification Kit (Roche Molecular Systems) following the Illumina platforms portion of the protocol, and qPCR was run on a Roche LightCycler 480 II. Libraries were sequenced on an Illumina NextSeq 500 at the U.S. Department of Agriculture (USDA)-Agricultural Research Service Small Grains Genotyping Center to produce 150-bp single-end reads.

Bioinformatics and Differential Expression Analysis

Sequencing reads were examined for quality metrics using FastQC (Andrews, 2010). Reads were quality trimmed using trimmomatic (Bolger et al., 2014) to remove adaptor sequences, trim based on quality via a sliding window, remove the leading 10 nucleotides, and discard reads less than 36 bp (ILLUMINACLIP:illumina_sequences_trimming:2:30:10 SLIDINGWINDOW:4:15 HEADCROP:10 MINLEN:36). Quality trimmed RNA sequencing reads were mapped to the LDN03-Sn4 reference genome (Richards et al., 2018) using HISAT2 (Kim et al., 2015; Pertea et al., 2016) specifying a maximum intron length of 3000 bp. Sequence alignment map files were converted to binary alignment map files and subsequently sorted and indexed using SAMtools (Li et al., 2009). Transcripts were assembled using StringTie under default settings (Pertea et al., 2016). Differential gene expression analysis was conducted using the R package DESeq2, specifically comparing the expression of SnTox1 in each isolate at each time point (Love et al., 2014). Expression values were calculated as normalized read counts and used for comparison between isolates and time points. Normalized read counts from the DESeq2 analysis were also extracted for genes encoding predicted secreted proteins (secretome), as well as all genes encoding predicted effectors (effectorome), to determine relative expression of SnTox1 compared with all predicted secreted proteins and effectors at time points with significant differences in expression (J.K. Richards and T.L. Friesen, unpublished data).

De Novo Genome Assembly and Protein Alignments

Illumina short-read sequence of isolate Sn5 was obtained from a related project (J.K. Richards and T.L. Friesen, unpublished data). Sequencing reads were de novo assembled using SPAdes under default settings (Bankevich et al., 2012). Assembled contigs were then used to create a local BLAST database, from which, ToxA, Tox1, and Tox3 sequences were identified via ‘blastn’ (Camacho et al., 2009). In cases where de novo assembled contigs did not contain a complete and contiguous gene sequence, the gene was PCR amplified and sequenced using the Sanger method to provide complete CDS information (Supplemental Table S7). Protein sequences were aligned in Geneious v11.1.5.

For the analysis of polymorphisms within the putative promoter regions, 1000 bp of sequence upstream of the start codons of SnToxA, SnTox1, and SnTox3 was retrieved from the SN15 and Sn4 reference genomes (Syme et al., 2016; Richards et al., 2018) using ‘pyfaidx’ (Shirley et al., 2015). Because no complete genome sequence of isolate Sn5 was available, primers were designed to amplify the putative promoter regions (Supplemental Table S7), and amplicons were sequenced using the Sanger method.

Transcriptional Expression Analysis using RT-qPCR

Plants of the RIL CDS37, which was sensitive to all three NEs because it has the genotype Snn1/Snn3-B1/Tsn1, were grown and inoculated with Sn2000 and Sn2000KO6-1 as described above. Leaf tissue samples for each genotype and isolate combination were collected from the second leaf at 12, 24, 48, 72, and 96 h after inoculation and immediately frozen in liquid nitrogen. RNA extraction and RT-qPCR methods were the same as in Virdi et al. (2016) with the exception of the 10 μL PCR reaction, which contained 1 × SYBR PCR Master Mix (Applied Biosystems), 0.25 µM each primer, and 4 µL of 4-fold diluted complementary DNA. Each experiment was conducted using at least six biological replicates, each consisting of a single inoculated leaf, and at least three technical replicates per biological replicate were performed. Primers for SnToxA were ToxA.RT.F3 (5′-AAC​GCC​AAT​ACA​GTG​CGA​GT-3′) and Tox.cod.1R (5′-GCT​GCA​TTC​TCC​AAT​TTT​CAC​G-3′), for SnTox1 were SnTox1RT1F (5′-CTC​ACG​TTT​GAG​GGC​TTA​GG-3′) and SnTox1RT1R (5′-GGA​TGC​AAT​AGA​GCA​GCA​GA-3′), and for the P. nodorum actin gene were ActinqPCRf (5′-AGT​CGA​AGC​GTG​GTA​TCC​T-3′) and ActinqPCRr (5′-ACT​TGG​GGT​TGA​TGG​GAG-3′). The expression level of the CDS37 sample at 12 h inoculated with Sn2000 was set at 1 as a calibration point. Threshold cycles of the SnToxA, SnTox1, and endogenous actin gene were used to calculate the relative expression levels using the 2^−ΔΔCT method. Statistical analysis was performed between each isolate for each line and time point using t-test to determine whether significantly different at P < 0.05.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers GU259637 for Tsn1in Sumai3, KP085710 for Snn1 in Chinese Spring, and HM191250, MK612041, and MK612040 for SnToxA, SnTox1, and SnTox3, respectively, in Sn5. Whole genome sequences including the NE genes used in this study are present in the National Center for Biotechnology Information database BioProject PRJNA398070 for isolates Sn4 and Sn2000, and BioProject PRJNA13754 for isolate SN15.

Supplemental Data

The following supplemental materials are available:

• Supplemental Figure S1. Histograms of the average lesion-type reactions of the CS-DIC 5B × Sumai 3 recombinant inbred population to various Parastagonospora nodorum isolates.

• Supplemental Figure S2. Histograms of the average lesion-type reaction of the CS-DIC 5B × Sumai 3 recombinant inbred population to Parastagonospora nodorum isolates Sn2000 and Sn2000KO6.

• Supplemental Figure S3. Alignment of the putative SnTox1 promoter region in Parastagonospora nodorum isolates Sn5, SN15, and Sn4.

• Supplemental Figure S4. The amino acid sequence of SnTox1 in Parastagonospora nodorum isolates Sn4, Sn5, and SN15.

• Supplemental Figure S5. Expression of SnToxA in the susceptible line CDS37 (Snn1/Snn3-B1/Tsn1) inoculated with Parastagonospora nodorum isolates Sn4, Sn5, and SN15 at 48, 72, and 96 h post inoculation.

• Supplemental Figure S6. Expression of SnTox3 in the susceptible line CDS37 (Snn1/Snn3-B1/Tsn1) inoculated with Parastagonospora nodorum isolates Sn4, Sn5, and SN15 at 48, 72, and 96 h post inoculation.

• Supplemental Figure S7. Alignment of the putative SnToxA promoter region in Parastagonospora nodorum isolates Sn5, SN15, and Sn4.

• Supplemental Figure S8. The amino acid sequence of SnToxA in Parastagonospora nodorum isolates Sn4, Sn5, and SN15.

• Supplemental Figure S9. Alignment of the putative SnTox3 promoter region in Parastagonospora nodorum isolates Sn5, SN15, and Sn4.

• Supplemental Figure S10. The amino acid sequence of SnTox3 in Parastagonospora nodorum isolates Sn4, Sn5, and SN15.

• Supplemental Table S1. Summary of the genetic linkage maps for each chromosome/genome in the CS-DIC 5B × Sumai 3 population.

• Supplemental Table S2. χ² analysis for sensitivity to the purified NEs SnTox1, SnTox3, and SnToxA, which interact with the wheat sensitivity genes Snn1, Snn3-B1, and Tsn1, respectively, in the CDS population.

• Supplemental Table S3. Bartlett’s χ² analysis for homogeneity amount replicates used in this study.

• Supplemental Table S4. RNA sequencing results for the P. nodorum isolates Sn4, Sn5, and SN15 at in planta time points 48, 72, and 96 h post infection.

• Supplemental Table S5. Summary of the observed interactions in this study for P. nodorum isolates Sn4, Sn5, Sn6, SN15, Sn2000, and Sn2000KO6-1 inoculated onto the CS-DIC 5B × Sumai 3 RI population.

• Supplemental Table S6. Parastagonospora nodorum isolates used in this study along with origin and source of the NE each isolate contains.

• Supplemental Table S7. Primers used in this study for sequencing necrotrophic effector genes.

• Supplemental File S1. Effectorome and secretome of P. nodorum isolate SN15 at 48, 72, and 96 h post inoculation.