Remarkable recent changes in the genetic diversity of the avirulence gene AvrStb6 in global populations of the wheat pathogen Zymoseptoria tritici.

Septoria tritici blotch (STB), caused by the fungus Zymoseptoria tritici, is one of the most economically important diseases of wheat. Recently, both factors of a gene-for-gene interaction between Z. tritici and wheat, the wheat receptor-like kinase Stb6 and the Z. tritici secreted effector protein AvrStb6, have been identified. Previous analyses revealed a high diversity of AvrStb6 haplotypes present in earlier Z. tritici isolate collections, with up to c.18% of analysed isolates possessing the avirulence isoform of AvrStb6 identical to that originally identified in the reference isolate IPO323. With Stb6 present in many commercial wheat cultivars globally, we aimed to assess potential changes in AvrStb6 genetic diversity and the incidence of haplotypes allowing evasion of Stb6-mediated resistance in more recent Z. tritici populations. Here we show, using targeted resequencing of AvrStb6, that this gene is universally present in field isolates sampled from major wheat-growing regions of the world in 2013-2017. However, in contrast to the data from previous AvrStb6 population studies, we report a complete absence of the originally described avirulence isoform of AvrStb6 amongst modern Z. tritici isolates. Moreover, a remarkably small number of haplotypes, each encoding AvrStb6 protein isoforms conditioning virulence on Stb6-containing wheat, were found to predominate among modern Z. tritici isolates. A single virulence isoform of AvrStb6 was found to be particularly abundant throughout the global population. These findings indicate that, despite the ability of Z. tritici to sexually reproduce on resistant hosts, AvrStb6 avirulence haplotypes tend to be eliminated in subsequent populations.

INTRODUCTION

The interactions between plant pathogens and their hosts during infection are highly complex and evolutionarily dynamic. Effectors, molecules including proteins that are produced and secreted by pathogens during infection, constitute a vital part of the repertoire of mechanisms utilized in the successful infection of plant hosts (Lo Presti et al., 2015). Effectors function by altering the metabolism of the host plant to facilitate infection or by suppressing plant immune responses to infection (Djamei et al., 2011; Kleemann et al., 2012; Marshall et al., 2011). However, plants possess a sophisticated innate immune system whose central players are cell surface and intracellular immune receptors including disease resistance (R) genes, which are capable of detecting pathogen effectors and initiating an immune response (Cook et al., 2015; Jones & Dangl, 2006; Kanyuka & Rudd, 2019).

Recognition of a pathogen effector protein by a plant immune receptor thereby introduces an evolutionary pressure on the pathogen to mutate or lose the recognized effector (otherwise known as an avirulence or Avr factor) entirely from its genome. Avr factors may be lost through frameshift or nonsense mutations (Luderer et al., 2002), transposon insertion (Zhang et al., 2015), or repeat‐induced point mutations (RIPs) (Rouxel et al., 2011). Indeed, effectors are often located in genomic regions rich in transposon activity, which drives effector diversity (Dong et al., 2015). Suppression of Avr factor expression, through mutations in the promoter, histone modification (Wang et al., 2020), or posttranscriptional silencing (Qutob et al., 2013), may also restore pathogen virulence. In some cases, point mutations in the Avr gene sequence may allow evasion of recognition whilst maintaining protein function (Blondeau et al., 2015; Plissonneau et al., 2017). This may happen in cases where the Avr factor is important for pathogen fitness. Once mutations in Avr genes have occurred that allow evasion of detection by immune receptors, they often spread rapidly through the pathogen population, particularly when the cognate disease resistance gene is widely used (Cowger et al., 2000; Hovmøller & Justesen, 2007).

Zymoseptoria tritici, the causal agent of Septoria tritici blotch (STB), is one of the most economically important fungal pathogens of wheat, with fungicide control costs exceeding €1 billion per year in Europe alone (Torriani et al., 2015). Z. tritici secretes an array of putative effectors during the infection cycle, among which is the avirulence factor AvrStb6 (Kema et al., 2018; Zhong et al., 2017), recognized in a gene‐for‐gene manner by the corresponding cell surface immune receptor Stb6 (Brading et al., 2002; Saintenac et al., 2018). AvrStb6 has been identified in the genomes of the isolates IPO323 and 1E4 collected from the Netherlands and Switzerland in 1981 and 1999, respectively, and recently cloned (Kema et al., 2018; Zhong et al., 2017). The Stb6 gene has also been recently cloned from wheat landrace Chinese Spring and the UK cultivar (cv.) Cadenza (Saintenac et al., 2018). AvrStb6 and Stb6 encode a small, cysteine‐rich secreted protein and a wall‐associated receptor‐like kinase protein, respectively. A functional haplotype of Stb6 conferring resistance to Z. tritici IPO323 is found in over half of commercial cultivars in Europe (Saintenac et al., 2018) and may have been present in wheat populations globally since early in agricultural history (Chartrain et al., 2005). Consequently, pressure on Z. tritici to lose AvrStb6 variants conditioning avirulence on Stb6‐containing wheat must be considerable.

There are two previous studies in which sequencing of AvrStb6 from populations of Z. tritici has been carried out. In one study, a global population of Z. tritici collected between 1990 and 2001 (Brunner & McDonald, 2018; Zhan et al., 2005) was analysed, whilst the second focused on a largely French population collected in 2009–2010 (Zhong et al., 2017). These studies found a high diversity of AvrStb6 haplotypes and evidence of positive selection driven by point mutations and recombination. The region of Z. tritici chromosome 5 in which AvrStb6 is located was found to be highly dynamic, with extensive transposon activity contributing to AvrStb6 polymorphism (Sánchez‐Vallet et al., 2018).

Whilst the AvrStb6 haplotype distribution in these earlier collections has been well characterized, the AvrStb6 haplotype diversity in more modern Z. tritici populations is unknown. Also unknown is the prevalence of Z. tritici isolates that possess AvrStb6 haplotypes capable of evading Stb6‐mediated defence. The specific polymorphisms that drive the change from avirulence to virulence phenotype in the AvrStb6 protein have not yet been determined, although changes at two amino acid residues (positions 41 and 43) in the AvrStb6 protein have been suggested as being critical for the pathogenicity on wheat cultivars carrying Stb6 (Kema et al., 2018).

In this study, we resequenced the AvrStb6 gene from recent field populations of Z. tritici isolates collected between 2013 and 2017, therefore providing a global snapshot of AvrStb6 haplotype diversity at this time. We show a notable decrease in the haplotype diversity of AvrStb6 compared to previous studies, including a recent worldwide shift towards haplotypes encoding Stb6 resistance‐breaking isoforms. Interestingly, one particular virulence isoform (I02) was found to be the most abundant in several regions of the world examined. This study therefore provides a rare insight into temporal changes in pathogen effector diversity in response to the host‐imposed pressures and highlights the population shifts that can occur when a single corresponding disease resistance gene is deployed extensively.

RESULTS

AvrStb6 haplotype analysis

Primers flanking the AvrStb6 gene were designed and used for PCR amplification of this gene from a collection of 383 Z. tritici isolates sampled from different field‐grown bread wheat (Triticum aestivum) cultivars in 10 countries on five continents. PCR products of expected size were obtained from all analysed samples, indicating AvrStb6 is present in all isolates in the collection. Sequencing of the PCR products revealed a total of 52 AvrStb6 haplotypes (denoted from H01 through to H52; Table S1) coding for 44 protein isoforms (denoted from I01 through to I44; Figure 1). This includes AvrStb6 haplotypes and the corresponding protein isoforms from the reference isolate IPO323 (H01/I01) and another well‐studied isolate IPO88004 (H52/I44), which are avirulent or virulent on Stb6‐containing wheat, respectively.

FIGURE 1

Alignment of the AvrStb6 isoform sequences identified in this study. Protein sequences were aligned using MAFFT v. 7.388. Numbers in parentheses represent the number of isolates identified possessing each isoform. Isoforms I01 and I44 have been identified in the historic Zymoseptoria tritici isolates IPO323 and IPO88004, which are avirulent or virulent on Stb6 wheat, respectively. Amino acids synonymous to the I01 sequence are greyed. Missing residues relative to I01 are represented as dashes. The pink arrow annotates the secretion signal peptide, whilst the green bars represent the exons of the coding gene sequence. The three alignment groupings (I, II, and III) of the isoforms are indicated by vertical blue lines to the right of the sequences, and invariant cysteine residues are highlighted in yellow

Almost all haplotypes were predicted to encode full‐length AvrStb6 protein. The exceptions to this were H17 (only present in four isolates from Turkey) containing a nonsense mutation in the exon 1, and H24 and H46 (each represented by a single isolate also from Turkey) that contained single nucleotide deletions/frameshift mutations located in exon 3 and exon 2, respectively. The majority of recovered AvrStb6 sequences (361 of 381) were 365 bp in length from the ATG start codon to the TGA stop codon and including two introns, with others ranging in lengths from 362 to 366 bp. Most of the nucleotide polymorphisms were in exons 2 and 3, with exon 1 being more conserved, and particularly within the sequence coding for the N‐terminal signal peptide (Figure 1). Eight haplotypes (H09, H10, H12, H21, H22, H23, H37, and H51) represented by 18 isolates contained three types of in‐frame 3‐bp deletions in exon 3. These haplotypes collapse to seven protein isoforms (I08, I09, I10, I11, I17, I40, and I43). All these isoforms apart from I17, represented by two isolates from western Europe, were exclusively found among the Z. tritici isolates from Turkey. The identified AvrStb6 isoforms were composed of 80–82 amino acid residues. From these, residues at 52 positions, including all 12 cysteines, were invariant across the collection (excluding the I42 frameshift mutant) (Figure 1). All other amino acid residues in AvrStb6 were variable, with the highest variation identified at positions 41, 61, and 77. As revealed by the amino acid sequence alignment and haplotype network analysis, AvrStb6 isoforms fell generally into three groups denoted I, II, and III (Figures 1, 2a, and S1). This, in principle, was supported by a phylogenetic analysis, which also identified that group III haplotypes, largely represented by isolates of Turkish origin, are phylogenetically more distant from those in groups I and II (Figure 2b).

FIGURE 2

Analysis of the AvrStb6 haplotypes distribution in the global Zymoseptoria tritici population. (a) Haplotype network for global AvrStb6 sequences obtained from the TCS analysis. Each node represents a separate haplotype identified in this study. The node size represents the number of isolates identified possessing the same haplotype. The nodes are painted in different colours depending on the geographic region(s) from which the isolates of the same node were identified. Connecting lines between nodes denote closely related haplotypes, with numbers in parentheses corresponding to the number of mutations between adjoined haplotypes, but line lengths are not proportional to genetic distance. Black unlabelled nodes represent hypothetical common ancestors of related haplotypes. The three groupings of haplotypes I, II, and III are highlighted by blue ellipses. (b) Phylogenetic tree of the AvrStb6 haplotypes constructed using PhyML. The tree was rooted using the Z. tritici gene Mycgr3G82331 located on chromosome 10, a paralogue of AvrStb6. Numbers on branches indicate bootstrap scores for each branch, whilst numbers in parentheses indicate the number of isolates identified possessing each haplotype

One AvrStb6 isoform predominates among current Z. tritici isolates globally

By far the most prevalent AvrStb6 isoform in the collection was I02, identified in 280 of 381 isolates (73.5%). This isoform was the most common in almost every geographic region studied, including Australia (96.6%), western Europe (91.8%), the United States (82.7%), and Chile (80%) (Table 1). The only exceptions were Argentina and Uruguay, where isoform I03 was most prevalent, and Turkey, where isoform I02 was not identified at all. AvrStb6 isoform I02 was found to be encoded by several haplotypes: H02 (239 isolates), H03 (34 isolates), H06 (6 isolates), H15 (2 isolates), and H33 (1 isolate). However, in the US collection, only a single haplotype (H02) was found to code for I02.

TABLE 1

Prevalence of AvrStb6 isoform I02 in the recent global Zymoseptoria tritici collection vs. in the earlier collections

Region of origin	Collection	Country/state of origin	Collection year(s)	Number of isolates	AvrStb6 I02 (%)
Western Europe	This study	England	2015–2017	135	94.8
		Scotland	2015–2017	14	92.9
		Ireland	2015–2017	8	87.5
		France	2015–2017	12	91.7
		Germany	2015–2017	12	66.7
	Brunner and McDonald (2018)	Switzerland	1999	29	51.7
	Zhong et al. (2017)	France	2009–2010	102	49
North America	This study	USA/Oregon	2016	48	89.6
	Brunner and McDonald (2018)	USA/Oregon	1990	56	0
Australia	This study	Tasmania	2014	58	96.6
	Brunner and McDonald (2018)	New South Wales	2001	27	74.1
South America	This study	Chile	2016	10	90
		Argentina	2016	10	20
		Uruguay	2016	10	30
Mediterranean	This study	Turkey	2013–2016	46	0

The next most common isoform globally was I03, the only other isoform to be found in multiple regions, which was identified in 23 isolates (6% of the isolate collection) in western Europe, South America, and Turkey. All other identified AvrStb6 isoforms were represented by one to eight isolates in the collection. Remarkably, 29 haplotypes and 25 AvrStb6 protein isoforms each were uniquely present in single isolates in the collection, representing well over half of the identified AvrStb6 isoforms (Figure 1).

Turkey is a hotspot of AvrStb6 diversity

Fifty‐seven Z. tritici isolates were sourced from smallholder farmers’ wheat fields in Turkey as part of this study. From this sample, 29 AvrStb6 haplotypes corresponding to 26 protein isoforms were identified. This represents a far higher rate of gene and protein sequence diversity than found in any other region studied here, with more than half of all identified AvrStb6 isoforms being found in Turkey and 25 being unique to this region. The network diagram of AvrStb6 haplotypes (Figure 2a) visualizes this high diversity, with both major branches largely consisting of haplotypes identified exclusively in the Z. tritici population from Turkey. Interestingly, isoform I02—the most prevalent isoform globally—was not observed in the Turkish population. This population therefore represents a notable break in the trend described above.

Stb6 resistance‐breaking isoforms are widespread in the current population

We next aimed to determine whether the various identified AvrStb6 isoforms conferred avirulence or virulence towards wheat genotypes carrying the corresponding resistance gene Stb6. For this, for each of the nine AvrStb6 isoforms that were identified in more than one Z. tritici isolate in the collection (Figure S2) we randomly selected one isolate as representative for use in wheat infection bioassays. One exception was the most common isoform I02, for which two representative isolates, 2NIAB and R16.1 (collected in 2015 and 2016, respectively, from different commercial fields in the UK), were selected. The bioassays involved fungal inoculation of two pairs of wheat genotypes at the seedling stage, with the genotypes in each pair being genetically near identical but distinguished by the presence/absence of Stb6. One pair comprised wheat cv. Cadenza, carrying the resistance‐conferring haplotype 1 of Stb6 (Saintenac et al., 2018), and a mutant of cv. Cadenza that contains no functional Stb6 due to a large CRISPR/Cas9‐induced frameshift deletion in the first exon of this gene. The second pair comprised wheat landrace Chinese Spring (Stb6) and a near‐isogenic line that contains the susceptibility haplotype 3 of Stb6 developed from a cross between Chinese Spring and a susceptible cv. Courtot. The previously characterized avirulent and virulent Z. tritici isolates IPO323 (Brading et al., 2002) and IPO88004 (Saintenac et al., 2018), possessing isoforms I01 and I44, respectively, were used as controls. Of the 10 isolates (representing nine different isoforms) tested, nine induced typical STB disease symptoms and fungal pycnidia on all tested wheat genotypes irrespectively of the presence/absence of Stb6 (Figures 3 and S3). This included the isolate(s) representing I02, the most prevalent AvrStb6 isoform in our collection.

FIGURE 3

Wheat inoculation bioassay. Leaves of differential wheat genotypes, wildtype cv. Cadenza containing Stb6 and a CRISPR/Cas9‐induced deletion mutant (Cadenza ΔStb6) that lacks the functional resistance gene in the same genetic background, were inoculated as young, 3‐week‐old seedlings with a selection of Zymoseptoria tritici isolates possessing different isoforms of AvrStb6. (a) Counts of pycnidiospores washed off the inoculated wheat leaves at 21 days postinoculation (dpi). Asterisks represent isolates with highly significant (**p < 0.005) differences in pycnidiospore counts between the resistant and susceptible genotypes. The most common isoform I02 was represented by the two isolates (labelled 1 and 2) originating from two different years and sampling locations. (b) and (c) Images of inoculated wheat leaves harvested at 21 dpi and incubated for 2 days under c.100% humidity to induce pycnidiation. Scale bar, 10 mm

Other than the avirulent reference isolate IPO323 (I01), only one other isolate could potentially be defined as demonstrating reduced virulence on Stb6 wheat, producing significantly fewer pycnidiospores on resistant wheat lines than those lacking Stb6 (Figure 3a). This isolate possessed the AvrStb6 isoform I13, which shares the highest percentage amino acid identity with the known avirulence isoform I01 (Figure S2). One other isolate, possessing AvrStb6 isoform I03, which was classified as virulent, also induced lower levels of disease on the Stb6‐containing wheat compared to their near‐isogenic genotypes that lack Stb6 as reflected by differences in the observed pycnidiospore counts (Figure 3). However, these differences were significant (at p < 0.01) only for the Z. tritici isolate T1‐A2017.24 possessing the AvrStb6 isoform I13 and the control avirulent isolate IPO323.

Avirulence isoforms of AvrStb6 are common on wheat genotypes irrespective of the Stb6 resistance gene status

We wondered whether overrepresentation of AvrStb6 virulence isoforms in our isolates collection could have been due to most isolates potentially sampled from resistant, Stb6‐containing wheat cultivars. The cultivar information was available for 254 out of 381 sampled Z. tritici isolates, and it was possible to obtain 20 cultivars—sources of 210 isolates—for further study (Table S1). Ten of these cultivars have been characterized in the previous study (Saintenac et al., 2018) and their Stb6 haplotype status and reaction to the avirulent strain IPO323 of Z. tritici are known (Table S1). We then resequenced Stb6 following the procedure described in Saintenac et al. (2018) from the other 10 cultivars (Consort, Dickens, Gallant, Genesis, Kaseberg, KWS Cashel, Lumos, Marston, Solace, and Zulu) to identify their Stb6 haplotype status (Table S1). It was determined that 9/20 of these cultivars possess disease resistance‐conferring haplotype 1 of Stb6 and they were sources of 73 (34.8%) Z. tritici isolates in our collection, whereas the other 11/20 analysed cultivars contained nonfunctional susceptibility haplotypes 3 or 7 of Stb6 and they represented sources of 137 (65.2%) isolates (Table 2). Therefore, it appears that there was no overall bias in the wheat cultivars used in this study towards susceptible or resistant, and that a larger proportion of Z. tritici isolates was sampled from cultivars containing no functional Stb6.

TABLE 2

Stb6 haplotype data for wheat cultivars used in this study

Wheat cultivar	Stb6 haplotype codea	Resistant (R)/susceptible (S) haplotype	No. of isolates collected from the cultivar
Alchemy	7	S	1
Consort	7	S	17
Cordiale	7	S	13
Cougar	1	R	22
Crusoe	7	S	1
Dickens	1	R	22
Evolution	1	R	6
Gallant	7	S	7
Genesis	3	S	10
JB Diego	7	S	2
Kaseburg	7	S	52
KWS Cashel	3	S	21
KWS Santiago	1	R	4
KWS Siskin	1	R	13
Lumos	7	S	8
Marston	1	R	1
Reflection	7	S	5
Solace	1	R	1
Trapez	1	R	2
Zulu	1	R	2
Haplotype totals	1	R	73
	3	S	31
	7	S	106

Stb6 haplotypes are coded as in Saintenac et al. (2018).

AvrStb6 expression is highly variable between Z. tritici isolates

Our main hypothesis is that the breakdown of Stb6‐mediated resistance by the current Z. tritici isolates analysed in this study (Figures 3 and S3) was due to the numerous changes in the AvrStb6 protein sequence identified (Figure 1). However, we also considered an alternative possibility whereby breakdown of Stb6 resistance may have been due to the suppression of AvrStb6 expression during infection. We therefore used a quantitative reverse transcription PCR (RT‐qPCR) to compare the expression levels of AvrStb6 in the 10 selected Z. tritici isolates (representing 10 different most frequent AvrStb6 isoforms) that were virulent on Stb6 wheat versus an avirulent isolate IPO323 during infection of a fully susceptible wheat cv. Taichung 29 that does not possess Stb6 (Ghaffary et al., 2012). As each of the 11 tested isolates has different infection dynamics on this wheat cultivar, the leaf tissues for use in RT‐qPCR were sampled at different time points after fungal inoculation, depending on the isolate, when the first symptoms of the disease become visible. This infection phase was chosen as many Z. tritici effector genes, including AvrStb6, have previously been shown to display maximal expression during transition from the biotrophic to the necrotrophic phase (Rudd et al., 2015; Zhong et al., 2017).

RT‐qPCR analysis revealed a high variability in AvrStb6 expression between isolates (Figure S4). Expression levels for IPO323 (I01) were higher than for many other isolates, apart from the isolate representative of I21, although the statistical significance of these differences was not examined. Importantly, however, although the AvrStb6 expression levels were highly variable between different isolates, the effector gene was expressed by all of them, although in WAI1822, carrying AvrStb6 isoform 14, expression was substantially suppressed (Figure S4). Overall, these findings argue against the mechanism of Stb6 resistance breakdown being due to suppression of AvrStb6 expression.

DISCUSSION

Few studies have analysed the temporal changes in Avr factor prevalence driven by selection pressure from host species in pathogen populations. Often, these have a fairly narrow geographical area coverage. For example, studies into the changes in Avr gene prevalence in the fungal pathogen Leptosphaeria maculans of Brassica crops over time have focused on single countries such as Canada (Fernando et al., 2018) or Australia (Van de Wouw et al., 2018). On the other hand, studies with a global geographical scope often do not consider the impact of sampling time on changes in Avr factor prevalence. Here we analysed the diversity of the avirulence factor AvrStb6 in the recent global Z. tritici populations. Moreover, we took advantage of the two similar published studies that analysed the population diversity of AvrStb6 in earlier Z. tritici collections, sampled in 1990–2001 or 2009–2010 (Brunner & McDonald, 2018; Zhong et al., 2017). Comparing our data that was based on the analysis of isolates sampled in 2013–2017 with that from these two previous studies revealed that large shifts in AvrStb6 haplotype prevalence have taken place in multiple global regions over the relatively short time period between samplings. These include the avirulence isoform of AvrStb6 originally described from the reference isolate IPO323 (collected in the Netherlands in 1981) becoming potentially extinct in the recent Z. tritici populations, and emergence of multiple virulence isoforms with one becoming predominant globally. We hypothesize that these changes in Z. tritici populations may be imposed by a worldwide increase in the deployment of commercial wheat cultivars carrying Stb6 (Chartrain et al., 2005; Saintenac et al., 2018), although comprehensive evidence for this hypothesis is currently unavailable.

The high proportion of AvrStb6 isoforms to haplotypes illustrates the high frequency of nonsynonymous mutations within the effector gene, which in turn suggests a strong selection pressure for the evolution of novel AvrStb6 protein isoforms that cannot be recognized by the corresponding wheat immune receptor Stb6. This phenomenon is well documented for other pathogen effectors, for instance with diversifying selection observed for ToxA, Tox1, and Tox3 of Parastagonospora nodorum (McDonald et al., 2013; Stukenbrock & McDonald, 2007), NIP1 of Rhynchosporium commune (Schürch et al., 2004), and AvrP123 and AvrP4 of the flax rust pathogen Melampsora lini (Barrett et al., 2009). In several fungal pathogens, RIPs have been shown to be responsible for the evolution of virulence due to the diversification of effector gene sequences, including the introduction of multiple premature stop codons (Fudal et al., 2009). Given the transposon‐rich, highly dynamic region of the genome in which AvrStb6 is located (Sánchez‐Vallet et al., 2018), it seems possible that a similar mechanism has contributed to the diversification of this Z. tritici effector. However, an in silico analysis suggests a low frequency of RIPs in the AvrStb6 genomic region (Figure S5). Moreover, we found that the secretion signal peptide region and all the cysteine residues in AvrStb6 were highly conserved (Figure 1), whilst premature stop codon mutations within the gene were extremely rare. The signal peptide of effectors is required for their secretion into the host apoplast, whilst the formation of disulphide bonds between cysteines is thought to help to stabilize the protein in the harsh alkaline pH apoplastic environment. In addition, using RT‐qPCR analysis we showed that AvrStb6 was expressed in a set of Z. tritici isolates representing the most frequently occurring isoforms of this effector during a phase transition from biotrophy to necrotrophy (Figure S4). Whilst efforts to demonstrate a role for AvrStb6 isoforms in virulence have to date proved unsuccessful (Kema et al., 2018), these findings combined suggest an evolutionary pressure to maintain a functional effector, implying the potential contribution of AvrStb6 to fungal virulence or fitness, perhaps under certain, as yet unknown, environmental conditions.

Our study revealed the prevalence of a single AvrStb6 isoform, I02, in the global population of Z. tritici. This isoform was found to be most common in all countries of western Europe from which the isolates were sourced, as well as Australia, the United States, and Chile. It is unknown why this isoform specifically became so prevalent in multiple regions around the world recently, given the ability of isolates carrying other isoforms to also evade Stb6‐mediated resistance (Figure 3). However, the limited diversity of Stb6 haplotypes (n = 3) in the host wheat plants sampled (Table 2) may suggest that AvrStb6 isoform I02 is specifically adapted for successful infection of wheat possessing these Stb6 haplotypes. The virulence isoform I02 may also be the best‐adapted for carrying out its as‐yet unknown effector function. Notably, sequencing of AvrStb6 from Z. tritici isolates sourced from Oregon, United States, in 1990 failed to identify a single occurrence of I02 (Table 1). This is in sharp contrast to our study, which identified prevalence of this isoform, found in 82.7% of the current isolates in the Oregon part of our collection. This population shift, combined with the fact that in all cases I02 was coded by a single haplotype H02, may suggest the introduction and rapid proliferation of this virulence AvrStb6 haplotype throughout the population of Z. tritici in Oregon, as occurred, for example, with the recent accidental importation of Pyricularia oryzae, a wheat blast pathogen, from Brazil to Bangladesh (Islam et al., 2016).

Our data also revealed a profound increase in the prevalence of AvrStb6 isoform I02 in the global recent (2013–2017) Z. tritici population compared to that in the older collections comprising isolates sampled in 1990–2001 (Brunner & McDonald, 2018) or 2009–2010 (Zhong et al., 2017) (Table 1). One exception to this is Australia, where a prevalence of AvrStb6 I02 in the local Z. tritici population sampled in 2001 was already evident (Zhan et al., 2005) (Table 1). Interestingly, in addition to the increasing prevalence of a single isoform I02, the overall diversity of AvrStb6 isoforms seem to have decreased over time. Thus, for example, previous studies of not current isolates by Brunner and McDonald (2018) and Zhong et al. (2017) identified 30 and 18 AvrStb6 isoforms in the populations comprising 142 or 103 isolates, respectively. These numbers are substantially higher than the 42 isoforms identified from 381 current isolates analysed in this study. Such a large shift in the AvrStb6 population diversity that occurred over a relatively short period of time and the increased fixation of AvrStb6 I02 in the population could have been caused by the decreased genetic diversity of wheat due to breeding (Fu & Somers, 2009) and widespread global deployment of Stb6‐containing cultivars by arable farmers (Chartrain et al., 2005; Saintenac et al., 2018). Other studies have shown changes in the Avr factor prevalence, such as an increased incidence of AvrLm4–7 in L. maculans from 47.2% in 2005–2006 (Dilmaghani et al., 2009) to 89.7% in 2010–2011 (Liban et al., 2016) in western Canada, demonstrating the ability of pathogens to adapt rapidly to changing conditions in host populations. However, to our knowledge this is the first time that a shift of such significance has been shown to occur across a global population of a plant pathogen.

The analysis of AvrStb6 haplotypes did not identify I02 as the dominant isoform in Argentina and Uruguay, and this isoform was completely absent from the Turkish Z. tritici population. It is possible that the local wheat cultivars used for sampling Z. tritici isolates for this study in these countries may have unique genetic makeups, although it was not possible to test this hypothesis in this study. The Turkish population is of particular interest, given it has a substantially higher AvrStb6 haplotype diversity compared to those in the other wheat‐growing regions studied here. Whilst the exact reasons for this difference are unknown, we hypothesize that the widespread use of local genetically diverse landraces, instead of genetically more similar commercial wheat cultivars, in Turkey is a driver for matching diversification of Z. tritici genomes. The location of Turkey near or at the hypothesized geographic region of the origin of wheat (Shewry, 2009) and Z. tritici (Stukenbrock et al., 2006) may also be a cause for the high diversity of AvrStb6 haplotypes observed here because origins of species often also correspond to the centres of their genetic diversity.

A recent study (Kema et al., 2018) suggested that the Z. tritici pathogenicity towards wheat cultivars carrying the disease resistance gene Stb6 was associated with changes at the amino acid positions 41 and 43 in the AvrStb6 protein. The new data obtained here support these findings and we confirm that isolates possessing isoforms I01 and I13 having G and D at positions 41 and 43, respectively (Figure S2), were either avirulent or generated significantly fewer pycnidiospores on wheat cultivars possessing Stb6 (Figure 3). However, we also found that the Z. tritici isolates expressing AvrStb6 isoform I03, I17, or I44, having E or V instead of G at position 41 but D at position 43, were fully virulent on Stb6 wheat (Figure S2). This suggests that amino acid changes in AvrStb6 at position 41 alone may be responsible for the Stb6 resistance‐breaking ability of Z. tritici isolates. It would be interesting to test this hypothesis through targeted mutagenesis in follow‐on studies, or indeed through testing of those Turkish isolates also possessing AvrStb6 isoforms with G and D at positions 41 and 43 (Figure 1).

The number of Z. tritici isolates analysed in this study and the failure to identify AvrStb6 isoform I01 in the collection suggest that this isoform has been eliminated from the modern Z. tritici population worldwide. However, the AvrStb6 haplotype network analysis indicates that the true diversity of AvrStb6 may be higher than identified as in some cases multiple mutations were found to separate the neighbouring haplotypes in the network (Figure 2a). A significant amount of undiscovered sequence variation in AvrStb6 may still be present in the global Z. tritici population. However, one of the main conclusions from our study is that the wheat disease resistance gene Stb6 has currently been almost completely overcome due to the predominance of fungal isolates carrying resistance‐breaking AvrStb6 isoforms in the modern Z. tritici populations around the world. This suggests that, despite the ability of Z. tritici to sexually reproduce on resistant hosts (Kema et al., 2018), Avr factors tend to be eliminated in subsequent populations.

EXPERIMENTAL PROCEDURES

Z. tritici isolates collection

The 381 Z. tritici isolates used in this study were collected from naturally infected wheat fields in the UK (n = 150), Ireland (n = 8), France (n = 12), Germany (n = 12), Australia (n = 58), and Turkey (n = 57), and from single fields in Chile (n = 10), Argentina (n = 10), Uruguay (n = 10), and the United States (n = 52). Most of these isolates were collected between 2015 and 2017, with the exception of isolates collected in Australia (2014) and Turkey (2013–2016), and the control isolates IPO323 (avirulent on Stb6 wheat) and IPO88004 (virulent on Stb6 wheat) collected in 1981 and 1988 in the Netherlands and Ethiopia, respectively (Goodwin et al., 2011; Kema & van Silfhout, 1997). Stocks of Z. tritici isolates were stored as blastospore water suspensions in 50% (vol/vol) glycerol at −80 °C.

Sequence, diversity, and phylogenetic analysis of AvrStb6

Genomic DNA extraction was carried out from Z. tritici isolates as previously described (Rudd et al., 2010). Primers for PCR amplification and sequencing of the AvrStb6 gene were designed in the upstream and downstream untranslated regions (UTRs) of the gene using publicly available and own whole‐genome sequencing data for European Z. tritici isolates (Chen et al., 2019, Rothamsted Research, personal communication). Primers avrstb6.f1 and avrstb6.r1 were selected for amplification of AvrStb6 from European isolates, and primers avrstb6.f3 and avrstb6.r1 were used to amplify AvrStb6 from North and South American isolates (Table S2). All amplifications were done using Phusion High‐Fidelity DNA polymerase (New England Biolabs UK Ltd). PCR products were purified using a QIAquick PCR Purification Kit (Qiagen) and Sanger sequenced in house or at Eurofins Genomics UK Ltd. AvrStb6 sequences from the Australian Z. tritici isolates were extracted from the whole‐genome sequencing data (NCBI BioProject accession number PRJNA480739; McDonald et al., 2019).

AvrStb6 haplotype and the corresponding protein isoform sequences were aligned using MAFFT v. 7.388 (Katoh & Standley, 2013) with the phylogenetic tree assembled using PhyML (Guindon & Gascuel, 2003). The AvrStb6 sequence from the Z. tritici isolate IPO323 was used as a reference for MAFFT alignments, whilst a paralogue of AvrStb6 present on chromosome 10 (gene Mycgr3G82331; Brunner & McDonald, 2018) was used to root the phylogenetic tree. A TCS haplotype network (Clement et al., 2000) was created to visualize the diversity of AvrStb6 haplotypes identified in the population using PopArt v. 1.7 (Leigh & Bryant, 2015). Coding regions were annotated and translated to amino acid sequence using Geneious v. 10.2.3 (Biomatters Ltd). RIP analysis of the region of Z. tritici chromosome 5 containing AvrStb6 was carried out using RIPper software (van Wyk et al., 2019).

The AvrStb6 sequence diversity data obtained here was compared to the similar previously published data sets for the historic Z. tritici collections. The AvrStb6 haplotype sequence obtained in Zhong et al. (2017) was kindly provided by Daniel Croll (University of Neuchâtel, Switzerland) and those obtained in Brunner and McDonald (2018) were downloaded from the NCBI PopSet Database (accession number 1,337,388,362).

Generation of wheat genetic material

A wheat Chinese Spring near‐isogenic line that carries the susceptibility haplotype 3 at the Stb6 locus was obtained following five backcrosses starting with F1 Chinese Spring × Courtot. In each generation, plants were assessed for their responses to the avirulent Z. tritici isolate IPO323 and genotyped with simple sequence repeat marker GWM369 to maintain the susceptibility haplotype of Stb6 in the progenies. A single BC5F1 plant heterozygous at the Stb6 locus was self‐pollinated. BC5F2 plants homozygous for the susceptibility haplotype of Stb6 were selfed and constitute the near‐isogenic line used in this study.

The Cadenza ΔStb6 mutant was produced by targeting the first exon of the Stb6 gene with CRISPR/Cas9 at five sites (Figure S6). The sgRNA constructs, carrying two sgRNAs each (annotated sequences are available on figshare), were assembled using the pENTR4‐sgRNA4 vector as previously described (Zhou et al., 2014). All three sgRNA plasmids were co‐delivered along with pCas9‐GFP (Zhang et al., 2019), encoding the wheat codon‐optimized Cas9 fused in frame at the C‐terminus with green fluorescent protein (GFP), and pRRes1.111 (Alotaibi et al., 2018), which directs expression of the bar selectable marker gene into immature wheat embryos of wheat cv. Cadenza using biolistics essentially as described in Sparks and Doherty (2020). CRISPR/Cas9 mutagenized T0 lines were genotyped for indels using the PCR band‐shift assay (Nekrasov et al., 2017). Cas9, sgRNA, and bar transgenes were segregated out in subsequent generations and the homozygous Cadenza ΔStb6 T2 line was used for phenotyping with Z. tritici.

Fungal bioassays

Leaves of 3‐week‐old wheat seedlings were inoculated with Z. tritici blastospore suspensions at 107 spores/ml as previously described (Keon et al., 2007). Wheat cv. Cadenza and the landrace Chinese Spring, both carrying Stb6, were used as resistant controls. A CRISPR/Cas9‐induced Cadenza ΔStb6 mutant and a near‐isogenic line in the Chinese Spring background that carries a susceptibility haplotype at the Stb6 locus were used as susceptible controls. A split‐plot randomized experimental design was used when carrying out the bioassays. Analysis using both sets of cultivars was carried out with five biological replicates per Z. tritici isolate tested. For cv. Cadenza lines, three independent repeats of this bioassay were carried out, whilst a single repeat was carried out using the Chinese Spring near‐isogenic lines.

STB disease symptoms on each inoculated leaf were visually assessed at 21 days postinoculation (dpi) as previously described (Lee et al., 2015). To complement the visual disease assessment, we carried out pycnidiospore wash counts as follows. Approximately 6‐cm long segments were cut from the inoculated leaves and these were exposed to high relative humidity at 15 °C for 48 hr in the dark to induce pycnidiation. Pycnidiospores were washed off the infected leaves by addition of 2 ml of distilled water supplemented with 0.01% Tween 20 followed by vortexing for 30 s. Optical density at 600 nm (OD600) of the harvested pycnidiospore suspensions was measured by spectrophotometer. This was converted to spores/ml by comparing against a standard curve produced by measuring OD600 of control suspensions containing defined pycnidiospore numbers as counted using a haemocytometer.

Wheat Stb6 haplotype analysis

Stb6 haplotype data for 10 wheat cultivars, sources of the Z. tritici isolates used here, were available from the recently published study (Saintenac et al., 2018). For 10 other cultivars, with previously unknown Stb6 status, we extracted genomic DNA from 3‐week‐old wheat plants and resequenced Stb6 exons following previously published methodology (Saintenac et al., 2018). Wheat cultivar information for Z. tritici isolates from Australia, Turkey, and a few from western Europe was not available.

Analysis of AvrStb6 expression during the infection of wheat

Three‐week‐old seedlings of wheat cv. Taichung 29, which is highly susceptible to Z. tritici and contains no known genes/quantitative trait loci for resistance to STB (Ghaffary et al., 2012), were inoculated with Z. tritici spore suspension at 107 spores/ml as previously described (Keon et al., 2007). At the emergence of visible symptoms on the inoculated leaves between 12 and 16 dpi (depending on the particular isolate), the leaves were harvested and snap‐frozen in liquid nitrogen. Total RNA was extracted using TRIzol (Fisher Scientific) following the manufacturer's instructions, and any potential traces of genomic DNA were removed using a TURBO DNA‐free kit (Fisher Scientific). First‐strand cDNA was synthesized from 1 µg of total RNA using SuperScript IV reverse transcriptase (Fisher Scientific). First‐strand cDNA preparations were diluted 1:10 with RNase‐free water and subjected to qPCR using SYBR Green JumpStart Taq ReadyMix (Merck Life Science UK Ltd), with primers specified in Table S2. Three biological and three technical replicates were carried out for each sample, with relative expression of AvrStb6 to the Z. tritici housekeeping gene glucose‐6‐phosphate 1‐dehydrogenase (G6PDH; gene ID Mycgr3G100879) calculated using the 2−ΔΔ t method (Livak & Schmittgen, 2001).

FIGURE S1 Frequency of each of the identified AvrStb6 haplotypes along with their geographic origin

Click here for additional data file.

FIGURE S2 Alignment of AvrStb6 isoforms from Zymoseptoria tritici isolates tested in pathoassays to determine virulence on Stb6‐containing wheat

Click here for additional data file.

FIGURE S3 Plant inoculation bioassay

Click here for additional data file.

FIGURE S4 Expression levels of different AvrStb6 haplotypes during Zymoseptoria tritici infection of wheat

Click here for additional data file.

FIGURE S5 Analysis of repeat‐induced point mutation frequency at the AvrStb6 locus on Zymoseptoria tritici chromosome 5 using RIPper (http://theripper.hawk.rocks)

Click here for additional data file.

FIGURE S6 Alignment of the part of the Stb6 coding sequence (exon 1) with the corresponding region in the CRISPR/Cas9‐induced wheat Cadenza ΔStb6 mutant

Click here for additional data file.

TABLE S1Zymoseptoria tritici isolates used in this study

Click here for additional data file.

TABLE S2 Primers used in this study

Click here for additional data file.