Epiphytic proliferation of Zymoseptoria tritici isolates on resistant wheat leaves

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


42
Zymoseptoria tritici is an ascomycete fungus that causes the wheat disease Septoria Tritici Blotch 43 (STB). STB is the most important pathogen of temperate-grown wheat, causing yield losses of up to 44 £200M per year in the UK alone and necessitating the extensive -and expensive -use of fungicides 45 to protect even resistant wheat cultivars . Resistance to STB is often 46 quantitative, although a number of resistance genes, termed stb genes, are also known (Saintenac et 47 al., 2018;Mekonnen et al., 2021). It has been reported that the defining feature of interactions 48 between avirulent isolates and resistant wheat cultivars is that fungal growth is arrested at the point 49 of stomatal penetration (Kema et al., 1996;Battache et al., 2022) due to the accumulation of 50 hydrogen peroxide (Shetty et al., 2003), cell wall strengthening, metabolic changes and production 51 of apoplastic defences such as PR-proteins (Rudd et al., 2015;Yang et al., 2015), or promotion of 52 stomatal closure (Battache et al., 2022).

53
Prior to penetration, however, the fungus may colonise the leaf surface. Epiphytic colonisation of 54 the leaf has been previously documented in interactions between virulent isolates and susceptible 55 wheat (Fones et al., 2017;Haueisen et al., 2019;Fantozzi et al., 2021). Our previous work showed 56 that the reference isolate, Z. tritici IPO323, germinates asynchronously to form hyphae which grow 57 randomly across the leaf surface, penetrating stomata as they are encountered. Rare stomatal 58 penetration events were observed as soon as 1 dpi, but significant ingress into the leaf was not 59 observed until around ten days after inoculation, once sufficiently large hyphal networks had 60 developed for growing tips to encounter stomata frequently (Fones et

68
This epiphytic growth has so far not been investigated in detail in interactions where the Z. tritici 69 isolate is unable or has limited ability to penetrate stomata and colonise the host apoplast. The 70 prevention of stomatal penetration is of fundamental importance in wheat defences against Z. tritici. 71 This, however, does not preclude the possibility that fungal germination and epiphytic growth prior 72 to penetration might occur on resistant wheat cultivars. The only study, to our knowledge, which has 73 explicitly compared pre-penetration growth in isolates of contrasting virulence was that of Siah et al. 74 (2010). These authors noted that pre-penetration growth in a weakly pathogenic isolate was similar 75 to that of a fully pathogenic isolate (Siah et al., 2010 Primer sequences are given in Table S1. 156

157
For the measurement of superoxide, harvested leaves were submerged in 0.1% (w/v) nitroblue 158 tetrazolium (NBT) (Love et al., 2005) overnight, before clearing by boiling in methanol. Cleared, 159 stained leaves were scanned at high resolution and images were analysed in HSB colour space in 160 ImageJ to measure the percentage of the total leaf area that had been stained blue by NBT. For each 161 isolate and time point, a minimum of three leaves (3-6) were scanned and analysed, and the 162 experiment was repeated three times independently. Supplemental Info S1.

Sequencing and genomic analyses
173 Library preparation, sequencing and de novo assembly was performed by Exeter Sequencing Service. 174 Genomic DNA was extracted 5-day old Z. tritici grown on YPD agar using a standard phenol-175 chloroform extraction procedure. DNA was and quantified by Qubit assay (Thermo Fisher Scientific Galaxie, but American isolates induce necrosis without producing pycnidia..

193
In order to determine the virulence of the isolates IPO323, IPO94269, IPO97001, T5, T23 and T39 on  194 Galaxie, infections were carried out and both symptom development and pycnidiation measured 195 ( at 21 dpi, highly significant differences were detected (ANOVA, p < 0.0001). These differences 199 appear to be due to the much more extensive necrosis seen in the European strains ( Fig 1C) IPO97001. There were differences in pycnidiation among the isolates (ANOVA, p < 0.0001; Fig 1D). 208 European isolatesdiffered from each other and from the uninoculated control, , whereas none of the 209 American isolates differed significantly from the uninoculated control (Tukey's simultaneous  210 comparisons; for detailed statistical results, see Supplementary Info S1). For simplicity, we adopt 211 the shorthand 'fully virulent' ('FV') and 'Necrosis-inducing with rare pycnidiation' ('NIRP')' to 212 describe European vs American isolates in the rest of this report, although of course it must be 213 noted that these appellations apply only to their interactions with Galaxie. The American isolates are 214 likely to be more able to complete their lifecycle and produce pycnidia on certain other wheat 215 cultivars, while it is understood that the European isolates are avirulent on some wheat cultivars. For 216 instance, IPO323 is avirulent on cultivars carrying stb6. Therefore, we emphasise that the FV and 217 NIRP phenotypes described here are specific to these particular isolate-cultivar interactions.
Differences between the two groups of isolates are likely to be attributable to the geographical 219 separation of these groups. We sequenced the genomes of the six isolates and identified SNPs on 220 the 13 core chromosomes (see Supplemental Info S2). Pairwise comparisons showed a minimum of 221 142,000 SNPs (T5/T39), while the maximum was 257,000 (IPO97001/T39). A pseudosequence tree 222 derived from these SNPs (Fig. S1) shows that, as expected, European (fully virulent) isolates cluster 223 together, as do American (NIRP) isolates. between the isolates, as are differences in the number of cells in yeast-like vs hyphal growth forms. 232 In particular, isolate T23 shows extensive hyphal growth on the leaf surface, while T39 shows a 233 proliferation of yeast like cells, from clumps at 7 dpi to extensive leaf surface coverage by 14 dpi. By 234 contrast, the fully virulent isolates show internal colonisation by 14 dpi, with comparatively little 235 increase in surface coverage between 7 and 24 dpi. This is illustrated in more detail in Fig. 3

243
In order to quantify these differences in leaf colonisation and fungal growth, multiple CSLM images 244 were obtained of each isolate on or in the leaf at multiple time points after inoculation. These 245 images were analysed to determine how many individual fungi contained one or more cells in the 246 hyphal state (Fig. 4A) or remained wholly on the leaf surface, with no stomatal ingress or 247 colonisation of internal leaf spaces (Fig 4B). Germination of cells to form hyphae occurred at 248 different rates for different isolates, with differences apparent by day 3 (ANOVA, p = 0.04525). 249 However, all isolates showed the ability to form hyphae. The two groups of isolates (FV vs NIRP) 250 could not be distinguished based on the rate of hyphal formation on at any of the time points 251 analysed (t-tests comparing means for isolate groups: 3 dpi, p = 0.269; 7 dpi, p = 0.963; 10 dpi, p = 252 0.626). Unlike hyphal formation, however, leaf penetration separated the two groups of isolates (t-253 tests comparing means for isolate groups: 10 dpi, p = 0.0339; 12 dpi, p = 0.0496). In fact, no internal 254 hyphae were observed for any NIRP isolate throughout the time course (Fig 4B). isolates already elicit slightly higher ROS production than the FV isolates (ANOVA, p = 0.00282). This 279 difference no longer apparent by 2 dpi (t-test, p = 0.2257) and is reversed, but remains non-280 significant, at 7 dpi (t-test, p = 0.183). This reversal, with the reduction of ROS response elicited by 281 avirulent isolates to less than that seen in response to the fully virulent isolates, is significant at 14 282 dpi (t-test, p = 0.00085). This pattern of response at 14 dpi was also seen in expression of the β-1,3-283 glucanase gene, whose expression is associated with fungal leaf penetration and the resulting wheat 284 defence response ( Figure S2; see Supplementary Info S2 for methods).

287
In this work, we have shown that there are interactions between certain wheat cultivars and isolates 288 of the fungal wheat pathogen Zymoseptoria tritici in which the Z. tritici isolate is able to survive and 289 reproduce on the leaf surface when despite being unable to colonise the apoplastic spaces inside the 290 leaf. Given that the NIRP phenotype is seen in American isolates which are more closely related to 291 each other than to isolates from Europe, it is possible that the NIRP phenotype reflects poor 292 adaptation of American isolates to the European wheat Galaxie. However, such poor adaptation to 293 available hosts might also be expected in the field when, for example, a new elite cultivar is 294 introduced. It will be important to determine whether the NIRP phenotype is also seen in any field 295 isolates on the wheat varieties from which they were isolated.

296
Epiphytic proliferation is common to fully virulent and NIRP interactions.

297
Confocal and scanning electron microscopy were used to visualise three fully virulent and three NIRP 298 isolates on and in the leaves of Galaxie wheat (Figs1-2). As can be seen in Fig. 2, epiphytic fungal 299 proliferation was clearly visible in every isolate at both 7 and 14 dpi. This is similar to the findings of 300 Siah et al., 2010, who showed that both a fully-and a weakly virulent isolate grew similarly on the 301 leaf surface prior to penetration. However, by comparing multiple Z. tritici isolates, we show that 302 the form and extent of epiphytic growth in NIRP isolates are as variable as for fully virulent isolates. 303 In particular, the isolate T39 showed extensive blastosporulation (microcycle conidiation/budding) 304 on the leaf surface, which persisted throughout the infection period (Fig. 3). In contrast, the 305 reference isolate IPO323 was less visible on the leaf surface from 10 dpi onwards, by which time it 306 had penetrated, and begun to colonise, the apoplast. Quantitative analysis showed that the fully 307 virulent isolates could not be distinguished from the NIRP isolates by growth form. Indeed, the 308 percentage of cells in hyphal vs yeast like forms at each timepoint differed as much within as 309 between isolate groups (Fig. 4A). Differences in the proportion of hyphal vs yeast-like individuals 310 may arise from differences in the rate at which initial blastospore inoculum germinates to produce 311 cells in the hyphal growth form, but also in the rate at which blastosporulation occurs on the leaf surface to form new yeast-like cells. Both IPO323 and T23 rapidly germinate and reach around 90% 313 hyphal growth form by 10 dpi (Fig 4A), in line with previously published data for IPO323 (Fones et al.,314 2017; Fantozzi et al., 2021). Other isolates show a maximum of 30-60% hyphal cells (Fig 4A) in the 315 same time period, reflecting either reduced germination (T5, IPO97001) or increased  316  blastosporulation (IPO94269, T5, T39), as can be seen in figures 2 and 3.

317
Ability to penetrate stomata and colonise the apoplast is essential for pycnidiation, but not for 318 increases in fungal biomass.

319
Unlike germination, leaf penetration provides a clear separation between the fully virulent and NIRP 320 isolates. No NIRP strain was able to colonise the apoplast, with all observed fungal cells being wholly 321 on the leaf surface at all time points analysed ( Figure 4B). This is in line with previous findings that all isolates with the host is similar, and might be expected to provoke similar defence responses. At 334 14 dpi, however, the NBT response was higher towards the fully virulent isolates, supporting the 335 observation that only these isolates had penetrated the leaves. The greater superoxide response to 336 NIRP isolates at 1 dpi might indicate early detection of these isolates. We speculate that this 337 response to NIRP isolates may play a role in inducing the defences that both cause the necrosis 338 induction seen in response to NIRP isolates and prevent them from successfully invading the leaves. 339 This possibility should be investigated further. qPCR of Z. tritici specific DNA in washed (Fones et al.,340 2017) and unwashed leaf samples supported the finding from microscopy that NIRP isolates were 341 not, or were barely, detectable inside the leaf at either 7 or 14 dpi (Fig 5). However, in unwashed 342 leaf samples, up to 4x as much Z. tritici DNA was found for NIRP than for fully virulent isolates (Fig.  343 5B). This counter-intuitive result indicates that NIRP isolates are not only surviving on the leaf 344 surface but proliferating. Further, this increase in fungal biomass on the leaf surface is not slow and 345 limited, but in fact outstrips the increase in biomass of virulent isolates at 14 dpi. At this time, the 346 fully virulent isolates have entered the apoplast and are beginning to colonise the mesophyll tissues, 347 but the switch to necrotrophic growth is yet to occur (

377
NIRP isolates rapidly trigger wheat defence signalling while on the leaf surface.

378
Nitroblue tetrazolium (NBT) staining experiments (Fig 6) indicate that the wheat plant responds to 379 the presence Z. tritici on the leaf surface rapidly by the production of superoxide. At 1 dpi, this 380 response is higher for NIRP than for fully virulent isolates, with this discrepancy reducing in size so 381 that it is not significant at 2 or 7 dpi, and then reversing. By 14 dpi, the response to fully virulent 382 isolates is significantly higher than for NIRP isolates, though the overall response is much lower than 383 that seen initially (Fig 6). It is likely that the response to fully virulent isolates reflects penetration 384 and the early stages of apoplast colonisation, which are known to induce defence responses (Keon et 385 al., 2007). Corroborating this, we see an upregulation of the β-1,3-glucanase gene in response to 386 fully virulent isolates, but not to NIRP isolates at 14 dpi ( Fig S1). This defence gene has previously 387 been reported to show induction following apoplastic colonisation by Z. tritici (Shetty et  for the different defence responses include i) differences in effector production; ii) differences in 398 toxin production; iii) differences in avirulence gene complement and iv) differences in early 399 penetration attempt rates. There are a relatively small number of well-characterised effector 400 proteins in Z. tritici, and most of those that whose functions are understood have their highest 401 expression at around 10-14 dpi, during the transition from biotrophic to necrotrophic growth 402 (Marshall et    internal from day 10 in IPO323 but remain on the surface in T39. Background red colour comes from 615 chlorophyll autofluorescence in the underlying mesophyll cells until leaves became necrotic (15 dpi Figure S1: Pseudosequence tree of isolates used in this study. Bases at SNP sites were used to 665 produce a pseudosequence for each strain, that was used to infer phylogenetic relationship 666 Numbers at nodes represent bootstrap results. Only SNPs with a minimum read depth of 10 and 95% 667 identity were included.
668 Figure S2: Wheat defence responses against fully virulent and NIRP isolates. Leaves of Galaxie