# Source paper archive

- Source title: Dissecting rice nonhost resistance to the wheat blast fungus
- Author(s): Shimizu, M., Hiraka, Y., Asuke, S., Adachi, H., Yuki, S., Kanzaki, E., Ito, K., Hirabuchi, A., Miyaji, N., Kajino, T., Minoji, K., Ohtsu, M., Sakai, T., Banfield, M. J., Abe, A., Tosa, Y., Kamoun, S., Terauchi, R.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.14.766612v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.14.766612v1.full.pdf
- PDF SHA-256: 80541802ba1ea9455e193837757fb0c70d01ffeb06ee3fcaddb53d2175ea2bca
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)

## **Dissecting rice nonhost resistance to the wheat blast fungus**


Motoki Shimizu &#91;1*&#93;, Yukie Hiraka &#91;1&#93;, Soichiro Asuke &#91;2&#93;, Hiroaki Adachi &#91;3,4&#93;, Sayaka Yuki &#91;1&#93;, Eiko


Kanzaki &#91;1&#93;, Kazue Ito &#91;1&#93;, Akiko Hirabuchi &#91;1&#93;, Naomi Miyaji &#91;1&#93;, Takuma Kajino &#91;1&#93;, Koki Minoji &#91;4&#93;, Mina


Ohtsu &#91;3&#93;, Toshiyuki Sakai &#91;4&#93;, Mark J Banfield &#91;5&#93;, Akira Abe &#91;1&#93;, Yukio Tosa &#91;2&#93;, Sophien Kamoun &#91;6&#93;, Ryohei


Terauchi &#91;1,4*&#93;


1 Iwate Biotechnology Research Center, Kitakami, Japan


2 Graduate School of Agricultural Science, Kobe University, Kobe, Japan.


3 Laboratory of Plant Immunity, Faculty of Advanced Life Science, Hokkaido University,


Hokkaido, Japan


4 Laboratory of Crop Evolution, Graduate School of Agriculture, Kyoto University, Kyoto, Japan


5 Department of Biochemistry and Metabolism, John Innes Centre, Norwich Research Park,


Norwich, United Kingdom


6 The Sainsbury Laboratory, University of East Anglia, Norwich Research Park, Norwich, United


Kingdom


   - Corresponding author


E-mail: m-shimizu@ibrc.or.jp &amp; terauchi@ibrc.or.jp


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Abstract:** The pandemic wheat blast fungus _Magnaporthe_ _oryzae_ destroys wheat but cannot


infect rice, a form of resistance so uniform that it does not segregate in genetic crosses and cannot


be mapped. We broke this nonhost resistance into single-gene phenotypes by expressing


individual effectors from the wheat-infecting pathotype, one at a time, in a rice-infecting isolate.


Screening 35 effectors across 20 rice cultivars (700 combinations) identified six that trigger


resistance. One, eTO5, is recognized by _Pi-eTO5_, a genetically linked quartet of NLR immune


receptors: SQ1, SQ2, and SQ4 are required for resistance, whereas SQ3 prevents constitutive


defense activation dependent on the other three. eTO5 is conserved across the pandemic B71


lineage spanning South America, Asia and Africa. Effector transfer into a compatible pathogen


should make nonhost resistance genetically tractable in other host–pathogen systems.


**One-Sentence Summary:** One-at-a-time effector transfer turns a nonhost barrier that cannot be


genetically mapped into single-gene resistance phenotypes, yielding known and novel rice


immune receptors.


**Main text:** Plants detect invading pathogens through immune receptors, and many resistance ( _R_ )


genes deployed in crops encode intracellular receptors of the nucleotide-binding leucine-rich


repeat (NLR) class (1). Most cloned _R_ genes have been identified from crop germplasm or its


wild relatives, where resistance segregates and can therefore be genetically mapped. Nonhost


species form a different kind of _R_ gene reservoir. Their NLR receptors have not been under


selection from coevolution with the pathogen in question, and resistance transferred from a


nonhost species to a crop may therefore provide a distinct and potentially durable source of


resistance (2). However, these genes are difficult to access: nonhost resistance is often uniform


across a species, does not segregate in crosses, and cannot be genetically mapped. In addition,


which pathogen effectors with an avirulence activity (AVR) are detected by nonhost _R_ genes


remains poorly known.


The blast fungus _Magnaporthe_ (Syn. _Pyricularia_ ) _oryzae_ causes disease on dozens of wild


and cultivated grasses and is differentiated into host-associated genetic lineages, with occasional


gene flow between them (3, 4). Wheat blast was first reported in Paraná, Brazil, in 1985 and


subsequently spread across South America. In 2016 and 2018, the disease emerged in Bangladesh


and Zambia, respectively, following separate introductions of the pandemic B71 clonal lineage


from South America (5–9). Only a limited number of wheat resistance genes effective against


wheat blast have been described (10, 11). Adaptation of _M. oryzae_ to wheat has also involved loss


of AVR effectors: functional loss of _PWT3_, for example, enabled escape from _Rwt3_ -mediated


resistance (12). Rice, in contrast, is uniformly highly resistant to the wheat-infecting pathotype.


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


We sought a way to genetically dissect rice nonhost resistance to the wheat-infecting


pathotype of _M. oryzae_ . We reasoned that the extensive NLR diversity of rice germplasm could


include receptors capable of detecting wheat blast effectors. Because resistance to the intact


pathogen is likely to involve multiple recognition events, we screened rice with effector genes


rather than with the pathogen itself. Individual candidate effectors from the wheat-infecting


pathotype were expressed, one at a time, in a rice-infecting isolate that is otherwise fully virulent


on rice. In this way, each transformant isolates one component of the nonhost barrier as a


genetically tractable resistance phenotype. The screen recovered effector recognition mediated by


_Pii_ and _Pia_, resistance loci previously characterized against the rice-infecting pathotype, showing


that the barrier to the wheat-infecting pathotype rests in part on the resistance repertoire rice


already maintains against its own pathogen. It also identified _eTO5_, an effector without a


detectable orthologue in the rice-infecting pathotype, as recognized by a subset of cultivars. We


then used _eTO5_ to genetically isolate its cognate resistance locus, _Pi-eTO5_, and found that _eTO5_


is conserved across the pandemic B71 lineage of the wheat blast fungus.


**Host specificity between rice and wheat blast enables systematic discovery of AVR effectors.**


To examine host specificity, we inoculated leaves of three rice cultivars, Hitomebore, Sasanishiki


and Nipponbare, with the rice-infecting _M. oryzae_ pathotype _Oryza_ (MO _)_ isolate Ao92-06-2 and


the wheat-infecting pathotype _Triticum_ (MT) isolate Br48. The MO isolate Ao92-06-2 produced


expanding lesions on all three rice cultivars, whereas the MT isolate Br48 failed to infect any of


them. These results confirmed the contrasting host specificity of the two pathotypes on rice ( **Fig** .


**1A; Fig. S1** ).


We reasoned that this host specificity could reflect the collective recognition of MT-specific


effectors as avirulence (AVR) factors by rice immune receptors and could therefore be exploited


as a discovery platform to systematically identify AVR genes. We first sought to define the


repertoire of candidate effector genes associated with the wheat blast pathogen.


We generated a _de novo_ genome assembly of Br48 ( **Data S1** ) and performed RNA


sequencing during barley infection to identify genes expressed _in planta_ . We identified 114


candidate cytoplasmic effector genes that met four criteria: they encoded predicted secreted


proteins, were highly expressed at 24 hours post inoculation (TPM&gt;100), were classified as


cytoplasmic or cytoplasmic/apoplastic effectors by EffectorP (13), and encoded proteins shorter


than 150 amino acids **(Fig. S2** ).


We then assessed the presence of these genes in nine MO isolates ( **Data S2** ). Thirty-five of


the 114 candidates were present in Br48 but showed minimal sequence coverage (≤30%) in all


nine tested MO isolates and were therefore operationally defined as absent from MO ( **Fig. 1B;**


**Fig. S3** ). We designated these MT-specific genes _eTOs_, for effectors unique to MT versus MO


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


( **Data S3** ).


**Functional screen of MT-specific effectors reveals multiple AVR activities in rice.**


To functionally identify MT-specific effector genes ( _eTO_ s) that trigger resistance in rice, we


developed a systematic functional screening pipeline ( **Fig. S4** ). We generated a library of


transgenic _M. oryzae_ Ao92-06-2 strains, each expressing one of the 35 _eTO_ genes under the


control of the _pex22_ ( _AVR-Pia_ ) promoter, which drives effector expression during host infection


(14).


Each transformant was inoculated onto 20 rice cultivars that were susceptible to the


parental Ao92-06-2 isolate ( **Data S4** ). Resistance specifically induced by an _eTO_ -expressing


transformant was interpreted as AVR activity recognized by the corresponding rice cultivar.


Screening all 35 transformants across the 20 rice cultivars generated 700 host-pathogen


combinations. Six _eTOs_ ( _eTO1_, _eTO2_, _eTO5_, _eTO16_, _eTO24_, and _eTO31_ ) triggered resistance in


at least one cultivar ( **Fig. 2; Fig. S5; Fig. S6** ). Notably, _eTO2_ triggered resistance in all 20 rice


cultivars tested, showing that it was broadly recognized across the panel and identifying it as a


candidate determinant of the inability of MT to infect rice.


The screen independently recovered two previously characterized AVR–R gene


relationships without prior knowledge of their identities. eTO1 and eTO24 showed amino acid


sequence similarity to the previously characterized MO AVR effectors AVR-Pii and AVR1-CO39,


respectively ( **Fig. S7** ), which are recognized by the rice resistance loci _Pii_ (15) and _Pia_ (16). To


test whether _eTO1_ -triggered resistance requires _Pii_, we inoculated Hitomebore and two


independent _pii_ mutant lines, Hit5948 and Hit5882, with Ao92-06-2 expressing _eTO1._


Hitomebore was resistant, whereas both _pii_ mutants were susceptible ( **Fig. 2C; Fig. S8A** ),


demonstrating that _eTO1_ -triggered resistance requires _Pii_ . Similarly, Sasanishiki was resistant to


Ao92-06-2 expressing _eTO24,_ whereas two independent _rga4_ mutant lines, Sas2127 and Sas1943,


were susceptible ( **Fig. 2C; Fig. S8B** ), demonstrating that _eTO24_ -triggered resistance requires _Pia._


These genetic results establish that the screen recovers genetically defined AVR–R relationships.


They also show that two of the six recognition events are mediated by _R_ genes that rice maintains


against the rice-infecting pathotype, indicating that the barrier to MT is assembled in part from


the standing anti-MO resistance repertoire rather than from machinery dedicated to nonhost


recognition. Because _Pii_ and _Pia_ segregate among cultivars, however, neither can by itself


account for resistance that is uniform across the species. We therefore focused on _eTO5_, whose


recognition is not attributable to a known rice blast resistance gene, to identify its cognate locus.


_**Pi-eTO5**_ **comprises a genetically linked NLR quartet**


Using the pipeline shown in **Fig. S4B**, we sought to identify the rice resistance locus responsible


4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


for _eTO5_ -triggered resistance, hereafter designated _Pi-eTO5_ ( **Fig. 3** ). We crossed the three


resistant rice cultivars, Nortai, WRC43 and Sesia with the susceptible cultivar Hitomebore.


Recombinant inbred line (RIL) populations derived from Nortai × Hitomebore and Sesia ×


Hitomebore and an F 2 population derived from WRC43 × Hitomebore were subjected to QTL

seq (17). Using the Nortai genome as the reference ( **Data S5** ), the three analyses identified an


overlapping interval on chromosome 11 associated with _eTO5_ -triggered resistance: 24.7–29.6 Mb


in Nortai, 24.4–28.4 Mb in WRC43, and 19.9–27.8 Mb in Sesia, with a common overlap of 24.7–


27.8 Mb ( **Fig. 3A, Fig. S9** ). Subsequent fine mapping using recombinant progeny derived from


Nortai × Hitomebore delimited _Pi-eTO5_ to an approximately 0.5-Mb interval containing four


adjacent complete NLR genes, _SQ1_, _SQ2_, _SQ3_ and _SQ4_ ( **Fig. 3B; Fig. S10; Data S6** ). High gene


coverage of all four NLR genes was observed in Nortai, Sesia, and WRC43, the three cultivars


that exhibited _eTO5_ -triggered resistance, whereas the _SQ1_ – _SQ4_ haplotype was absent from the


susceptible cultivar Hitomebore ( **Fig. S10C** ).


Target gene knockout analyses revealed that disruption of _SQ1_, _SQ2_ and _SQ4_ abolished


resistance to Ao92-06-2 expressing _eTO5_ in both the Nortai and WRC43 backgrounds ( **Fig. 3C;**


**Fig. S11** ), demonstrating that all three genes are required for _eTO5_ -triggered resistance. By


contrast, homozygous disruption of _SQ3_ caused severe dwarfism in both the Nortai and WRC43


backgrounds, whereas wild-type and _SQ3/sq3_ heterozygous siblings displayed normal growth


( **Fig. 3D; Fig. S12** ). Because _sq3_ homozygotes are dwarfed, _eTO5_ -triggered resistance could not


be scored in this background. Additional disruption of _SQ1,_ _SQ2_, or _SQ4_ suppressed the dwarf


phenotype of _sq3_ plants ( **Fig. 3E; Fig. S13** ), demonstrating that the dwarf phenotype caused by


loss of _SQ3_ depends on functional _SQ1_, _SQ2_, and _SQ4_ . RNA-seq analysis of _sq3/sq3_ seedlings


before visible growth defects revealed broad transcriptional reprogramming, including increased


expression of defense-associated genes and enrichment of stress-response pathways ( **Fig. S14;**


**Data S7–S8** ). Together with the genetic suppression of _sq3_ dwarfism by disruption of _SQ1_, _SQ2_,


or _SQ4_, these results support a role for _SQ3_ in restraining immune activity.


Phylogenetic analysis of the NB-ARC domain placed SQ1, SQ2, SQ3, and SQ4 in distinct


clades, arguing against their origin through a recent local duplication event ( **Fig. 3F; Data S9** ).


–
Genes corresponding to _SQ1_ _SQ4_ were also identified across the japonica and indica groups of


cultivated rice and in _O. rufipogon_ and other wild _Oryza_ species, indicating that putative complete


SQ1–SQ4 haplotypes occur in a subset of cultivated and wild _Oryza_ accessions ( **Fig. S15; Data**


**S10–S13** ).


Together, these results identify _Pi-eTO5_ as a genetically linked NLR quartet with


specialized functions. _SQ1_, _SQ2_, and _SQ4_ are required for _eTO5_ -triggered resistance, whereas


_SQ3_ is required to prevent constitutive defense activation that depends on _SQ1, SQ2,_ and _SQ4,_


thereby avoiding associated growth defects.


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Evolutionary conservation and molecular associations of eTO5**


eTO5 is broadly conserved across the B71 pandemic clonal lineage, with increased copy


number in several geographically dispersed derivatives ( **Fig. 4A** ; **Data S14–S15** ). eTO5 and


eTO1, both members of the ZiF effector family (18) ( **Fig. S16A** ), show a similar copy-number


increase in isolates from the geographically expanded derivatives of this lineage ( **Fig. 4A; Fig.**


**S16B** ). This parallel copy-number increase suggests that copy-number expansion can occur in


multiple structurally related effectors rather than being unique to eTO5. In contrast, eTO16 and


eTO31 are consistently absent from Bangladeshi isolates within the pandemic lineage ( **Fig. 4A** ),


indicating that both copy-number expansion and gene loss contribute to effector repertoire


variation within the pandemic lineage.


To characterize the functional features of the _Pi-eTO5_ -linked NLRs, we analyzed the


protein sequences of SQ1, SQ2, SQ3, and SQ4. SQ4 uniquely contains a MADA motif (19) ( **Fig.**


**S17** ), a conserved N-terminal motif associated with NLR-mediated cell-death activity. We next


examined whether eTO5 associates with the Pi-eTO5 NLR proteins using co

immunoprecipitation (co-IP) assays following transient expression of the proteins in _Nicotiana_


_benthamiana_ . eTO5 co-immunoprecipitated with SQ2 and SQ3, but not with SQ1 or SQ4 ( **Fig.**


**4B; Fig. S18** ). This selective association pattern, together with the distinct distribution of NLR

associated motifs among SQ1–SQ4, is consistent with functional specialization among the four


_Pi-eTO5_ -linked NLRs.


**Discussion:** _Magnaporthe oryzae_ exhibits strong host specificity to different grass species. Here


we identified six wheat blast effector genes that trigger resistance in rice. Notably, _eTO2_ is the


only effector recognized by all rice cultivars tested, suggesting that recognition of _eTO2_ may


contribute broadly to the host-specific barrier against MT in rice. The absence of a susceptible


genotype among the cultivars tested prevented genetic mapping of the corresponding resistance


determinant. This broad recognition could reflect one or more widely distributed rice immune


receptors, although its genetic basis remains unresolved. Together, these results reveal


overlapping layers of effector recognition within rice nonhost resistance to MT. Although


recognition of individual effectors such as eTO5 segregates among rice genotypes, the pathogen


appears to remain unable to infect because additional recognition specificities mask this


segregation. Thus, effector-by-effector screening exposes genetic variation in resistance that is


hidden when nonhost resistance is examined only at the whole-pathogen level. Uniformity at the


species level is therefore emergent from overlapping specificities rather than a property of any


single receptor.


Unlike _eTO2_, wheat blast effectors including _eTO1, eTO5, eTO16, eTO24_ and _eTO31_ were


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


recognized by a subset of rice cultivars. Of these, _eTO1-_ and _eTO24-_ triggered resistance were


genetically dependent on _Pii_ and _Pia_, respectively. Focusing on _eTO5_, we isolated the cognate


resistance locus using our genetic pipeline. These results support the idea that recognition of


multiple MT effectors contributes substantially to the nonhost resistance of rice. The contrast with


wheat is informative. Under selection from wheat, MT lost _PWT3_ and escaped _Rwt3_ (12). Rice


imposes no reciprocal selection, and MT accordingly retains effectors such as _eTO1_ and _eTO24_


that rice R genes still detect.


_Pi-eTO5_ comprises four tightly linked NLR genes with genetically distinct contributions


to immunity and growth regulation. Disruption of _SQ1_, _SQ2_, or _SQ4_ abolished _eTO5_ -triggered


resistance, whereas loss of _SQ3_ caused severe dwarfism and constitutive defense-associated


transcriptional activation. The dwarf phenotype of _sq3/sq3_ plants was suppressed by additional


disruption of _SQ1_, _SQ2_, or _SQ4_, indicating that _SQ3_ is required to restrain immune activity


dependent on the other three NLRs. Consistent with functional specialization within the quartet,


SQ4 uniquely contains a MADA motif, whereas eTO5 co-immunoprecipitated with SQ2 and SQ3.


These observations support a model in which the four linked NLRs perform distinct but


coordinated functions, although the molecular sequence of effector recognition and immune


activation remains to be established.


_eTO5_ and _eTO1_ both show copy-number increases within the B71 clonal lineage ( **Fig. 4A;**


**Data S14–S16** ), with elevated copy-number estimates observed in geographically dispersed


derivatives of the lineage. In contrast, _eTO16_ and _eTO31_ are absent from the Bangladeshi isolates


examined, illustrating that both copy-number expansion and gene loss contribute to variation in


the effector repertoire of the pandemic lineage. We summarize these patterns in a model in which


the effector repertoire varies during geographic spread of the lineage, while several _eTOs_


recognized by rice resistance genes remain present in currently circulating isolates ( **Fig. 4C** ). The


retention of _eTO5_ across the sampled B71 pandemic isolates provides a rationale for testing _Pi-_


_eTO5_ as a source of wheat blast resistance. Whether all four NLRs are required in a heterologous


background, and whether a four-gene cassette can be deployed in wheat, remain to be determined.


More broadly, this functional screening approach provides a strategy for genetically


accessing nonhost resistance that would otherwise be difficult to map. By transferring individual


effectors into a related pathogen that can infect the nonhost species, complex resistance to the


intact pathogen can be decomposed into genetically tractable recognition phenotypes. Applied


across diverse host–pathogen systems, this strategy could facilitate systematic discovery of AVR–


R relationships and uncover resistance genes with potential for deployment in crops. Thus,


effector-driven screening converts genetically inaccessible nonhost resistance into tractable


phenotypes for resistance-gene discovery.


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Acknowledgements**


This study was supported by JSPS KAKENHI grants 23K20042 and 24H00010 (to R.T.) and


25K01992 (to M.S.). Some computations were performed on the supercomputer at the National


Institute of Genetics, Research Organization of Information and Systems. We thank K. Oikawa,


S. Nagata, S. Takei, and T. Fujiwara for technical assistance. We thank the National BioResource


Project at the National Institute of Genetics, Japan, for providing seeds of the wild Oryza


accessions. We also thank the Genebank of the National Agriculture and Food Research


Organization, Japan, for providing seeds of accessions from the Japanese Rice Core Collection


and the World Rice Core Collection (WRC1, WRC18, and WRC43).Created in BioRender.


SHIMIZU, M. (2026) https://BioRender.com/3j4650t


**REFERENCE**


1. J. Kourelis, R. A. L. van der Hoorn, Defended to the nines: 25 years of resistance gene


cloning identifies nine mechanisms for _R_ protein function. _Plant Cell_ **30**, 285–299 (2018).


2. S. Lee, V. M. Whitaker, S. F. Hutton, Potential applications of non-host resistance for crop


improvement. _Front. Plant Sci._ **7**, 997 (2016).


3. B. C. Couch, I. Fudal, M.-H. Lebrun, D. Tharreau, B. Valent, P. Van Kim, J.-L. Nottéghem,


L. M. Kohn, Origins of host-specific populations of the blast pathogen _Magnaporthe oryzae_


in crop domestication with subsequent expansion of pandemic clones on rice and weeds of


rice. _Genetics_ **170**, 613–630 (2005).


4. P. Gladieux, B. Condon, S. Ravel, D. Soanes, J. L. N. Maciel, A. Nhani Jr., L. Chen, R.


Terauchi, M.-H. Lebrun, D. Tharreau, T. Mitchell, K. F. Pedley, B. Valent, N. J. Talbot, M.


Farman, E. Fournier, Gene flow between divergent cereal- and grass-specific lineages of the


rice blast fungus _Magnaporthe oryzae_ . _mBio_ **9**, e01219-17 (2018).


5. S. Igarashi, C. M. Utiamada, L. C. Igarashi, A. H. Kazuma, R. S. Lopes, Pyricularia em trigo.


1. Ocorrência de _Pyricularia_ sp. no estado do Paraná. _Fitopatol. Bras._ **11**, 351–352 (1986).


6. M. T. Islam, D. Croll, P. Gladieux, D. M. Soanes, A. Persoons, P. Bhattacharjee, M. S.


Hossain, D. R. Gupta, M. M. Rahman, M. G. Mahboob, N. Cook, M. U. Salam, M. Z. Surovy,


V. B. Sancho, J. L. N. Maciel, A. Nhani Júnior, V. L. Castroagudín, J. T. de Assis Reges, P.


C. Ceresini, S. Ravel, R. Kellner, E. Fournier, D. Tharreau, M.-H. Lebrun, B. A. McDonald,


T. Stitt, D. Swan, N. J. Talbot, D. G. O. Saunders, J. Win, S. Kamoun, Emergence of wheat


blast in Bangladesh was caused by a South American lineage of _Magnaporthe oryzae_ . _BMC_


_Biol._ **14**, 84 (2016).


7. P. K. Malaker, N. C. D. Barma, T. P. Tiwari, W. J. Collis, E. Duveiller, P. K. Singh, A. K.


Joshi, R. P. Singh, H.-J. Braun, G. L. Peterson, K. F. Pedley, M. L. Farman, B. Valent, First


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


report of wheat blast caused by _Magnaporthe oryzae_ pathotype _triticum_ in Bangladesh. _Plant_


_Dis._ **100**, 2330 (2016).


8. B. Tembo, R. M. Mulenga, S. Sichilima, K. K. M’siska, M. Mwale, P. C. Chikoti, P. K. Singh,


X. He, K. F. Pedley, G. L. Peterson, R. P. Singh, H. J. Braun, Detection and characterization


of fungus ( _Magnaporthe oryzae_ pathotype _Triticum_ ) causing wheat blast disease on rain-fed


grown wheat ( _Triticum aestivum_ L.) in Zambia. _PLOS ONE_ **15**, e0238724 (2020).


9. S. M. Latorre, V. M. Were, A. J. Foster, T. Langner, A. Malmgren, A. Harant, S. Asuke, S.


Reyes-Avila, D. R. Gupta, C. Jensen, W. Ma, N. U. Mahmud, M. S. Mehebub, R. M. Mulenga,


A. N. M. Muzahid, S. K. Paul, S. M. F. Rabby, A. A. M. Rahat, L. Ryder, R.-K. Shrestha, S.


Sichilima, D. M. Soanes, P. K. Singh, A. R. Bentley, D. G. O. Saunders, Y. Tosa, D. Croll,


K. H. Lamour, T. Islam, B. Tembo, J. Win, N. J. Talbot, H. A. Burbano, S. Kamoun, Genomic


surveillance uncovers a pandemic clonal lineage of the wheat blast fungus. _PLOS Biol._ **21**,


e3002052 (2023).


10. C. D. Cruz, G. L. Peterson, W. W. Bockus, P. Kankanala, J. Dubcovsky, K. W. Jordan, E.


Akhunov, F. Chumley, F. D. Baldelomar, B. Valent, The 2NS translocation from _Aegilops_


_ventricosa_ confers resistance to the _Triticum_ pathotype of _Magnaporthe oryzae_ . _Crop Sci._


**56**, 990–1000 (2016).


11. V. L. Anh, N. T. Anh, A. G. Tagle, T. T. P. Vy, Y. Inoue, S. Takumi, I. Chuma, Y. Tosa, _Rmg8_,


a new gene for resistance to _Triticum_ isolates of _Pyricularia oryzae_ in hexaploid wheat.


_Phytopathology_ **105**, 1568–1572 (2015).


12. Y. Inoue, T. T. P. Vy, K. Yoshida, H. Asano, C. Mitsuoka, S. Asuke, V. L. Anh, C. J. R.


Cumagun, I. Chuma, R. Terauchi, K. Kato, T. Mitchell, B. Valent, M. Farman, Y. Tosa,


Evolution of the wheat blast fungus through functional losses in a host specificity


determinant. _Science_ **357**, 80–83 (2017).


13. J. Sperschneider, P. N. Dodds, EffectorP 3.0: Prediction of apoplastic and cytoplasmic


effectors in fungi and oomycetes. _Mol. Plant-Microbe Interact._ **35**, 146–156 (2022).


14. K. Yoshida, H. Saitoh, S. Fujisawa, H. Kanzaki, H. Matsumura, K. Yoshida, Y. Tosa, I.


Chuma, Y. Takano, J. Win, S. Kamoun, R. Terauchi, Association genetics reveals three novel


avirulence genes from the rice blast fungal pathogen _Magnaporthe oryzae_ . _Plant Cell_ **21**,


1573–1591 (2009).


15. H. Takagi, A. Uemura, H. Yaegashi, M. Tamiru, A. Abe, C. Mitsuoka, H. Utsushi, S. Natsume,


H. Kanzaki, H. Matsumura, H. Saitoh, K. Yoshida, L. M. Cano, S. Kamoun, R. Terauchi,


MutMap-Gap: whole-genome resequencing of mutant F2 progeny bulk combined with de


novo assembly of gap regions identifies the rice blast resistance gene _Pii_ . _New Phytol._ **200**,


276–283 (2013).


16. Y. Okuyama, H. Kanzaki, A. Abe, K. Yoshida, M. Tamiru, H. Saitoh, T. Fujibe, H.


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


Matsumura, M. Shenton, D. C. Galam, J. Undan, A. Ito, T. Sone, R. Terauchi, A multifaceted


genomics approach allows the isolation of the rice _Pia_ -blast resistance gene consisting of


two adjacent NBS-LRR protein genes. _Plant J._ **66**, 467–479 (2011).


17. H. Takagi, A. Abe, K. Yoshida, S. Kosugi, S. Natsume, C. Mitsuoka, A. Uemura, H. Utsushi,


M. Tamiru, S. Takuno, H. Innan, L. M. Cano, S. Kamoun, R. Terauchi, QTL-seq: rapid


mapping of quantitative trait loci in rice by whole genome resequencing of DNA from two


bulked populations. _Plant J._ **74**, 174–183 (2013).


18. J. C. De la Concepcion, T. Langner, K. Fujisaki, X. Yan, V. Were, A. H. C. Lam, I. Saado, H.


J. Brabham, J. Win, K. Yoshida, N. J. Talbot, R. Terauchi, S. Kamoun, M. J. Banfield, Zinc

finger (ZiF) fold secreted effectors form a functionally diverse family across lineages of the


blast fungus Magnaporthe oryzae. PLoS Pathog. 20, e1012277 (2024).


19. H. Adachi, M. P. Contreras, A. Harant, C.-H. Wu, L. Derevnina, T. Sakai, C. Duggan, E.


Moratto, T. O. Bozkurt, A. Maqbool, J. Win, S. Kamoun, An N-terminal motif in NLR


immune receptors is functionally conserved across distantly related plant species. _eLife_ **8**,


e49956 (2019).


10


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Fig.1. Host specificity between rice- and wheat-infecting** _**M. oryzae**_ **pathogens enables**


**systematic identification of MT-specific candidate effectors.**


**(A)** Host-specificity of _Magnaporthe oryzae_ pathotype _Oryza_ (MO) and pathotype _Triticum_ (MT).


The rice-infecting MO isolate Ao92-06-2 produced expanding lesions on three rice cultivars


(Hitomebore, Sasanishiki, and Nipponbare), whereas the wheat-infecting MT isolate Br48 failed


to produce expanding lesions on any of them. Representative images from punch-inoculation


assays at 7 days post-inoculation (dpi) are shown. Scale bar, 1cm.


**(B)** Comparative genomic analysis of candidate cytoplasmic effector genes in MT and MO


isolates. The heatmap shows the gene coverage for 114 candidate effector genes identified in the


MT isolate Br48 across Br48 and nine MO isolates. Candidates were selected based on four


criteria: a predicted signal peptide, expression of TPM &gt; 100 at 24 hours post-inoculation,


classification as a cytoplasmic or cytoplasmic/apoplastic effector by EffectorP, and a predictied


protein length of &lt;150 amino acids. Thirty-five genes present in Br48 but with ≤ 30% gene


coverage in each of the nine MO isolates were operationally defined as absent from MO and


designated _eTOs,_ for effector genes unique to MT versus MO. Gene order along the horizontal


axis and isolate order along the vertical axis were determined by hierarchical clustering using


Euclidean distance and average linkage.


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Fig. 2. Functional screen of MT-specific effector genes identifies multiple AVR activities in**


**rice.**


**(A)** Representative punch-inoculation phenotypes of five rice cultivars inoculated with the


parental MO isolate Ao92-06-2 or Ao92-06-2 transformants expressing individual MT-specific


_eTO_ genes under the control of the _pex22_ promoter from _AVR-Pia_ . Leaves were photographed at


10 days post-inoculation (dpi). Six _eTOs_ ( _eTO1, eTO2, eTO5, eTO16, eTO24_, and _eTO31_ )


triggered resistance in at least one rice cultivar. Quantitative lesion area measurements are shown


in **Fig. S6** . Scale bar, 1 cm. **(B)** Compatibility matrix showing the responses of 20 rice cultivars


to the parental Ao92-06-2 isolate and the six resistance-inducing _eTO_ -expressing transformants.


Filled blue circles indicate incompatible interactions (resistance), whereas open circles indicate


compatible interactions (susceptibility). _eTO2_ triggered resistance in all 20 cultivars tested,


whereas the other five _eTOs_ showed cultivar-specific recognition patterns. **(C)** Genetic validation


of _eTO1-_ and _eTO24_ -triggered resistance. Hitomebore was resistant to Ao92-06-2 expressing


_eTO1_, whereas two independent _pii_ mutant lines, Hit5948 and Hit5882, were susceptible,


demonstrating that _eTO1_ -triggered resistance requires _Pii_ . Similarly, Sasanishiki was resistant to


Ao92-06-2 expressing _eTO24_, whereas two independent _rga4_ mutant lines, Sas2127 and Sas1943,


were susceptible, demonstrating that _eTO24_ -triggered resistance requires _Pia_ . Scale bars, 0.5 cm.


12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Fig. 3. A genetically linked NLR quartet underlies** _**Pi-eTO5**_ **-mediated resistance.**


**(A)** QTL-seq analyses of the Nortai × Hitomebore, WRC43 × Hitomebore, and Sesia ×


Hitomebore populations, using the Nortai genome as the reference, identified an overlapping


interval between 24.7 and 27.8 Mb on chromosome 11 that was associated with _eTO5_ -triggered


resistance. The plot shows the ΔSNP-index for the Nortai x Hitomebore population. The red line


indicates the ΔSNP-index smoothed using a sliding-window approach, and the green lines


indicate the 99% confidence limits. The mapping intervals obtained from the three populations


are shown below. **(B)** Fine mapping of _Pi-eTO5_ using recombinant progeny derived from the


Nortai × Hitomebore cross. Recombinants #91 and #24 delimited _Pi-eTO5_ to an approximately


0.53-Mb interval between M2 and M3. This interval contains four adjacent NLR genes, _SQ1_, _SQ2_,


_SQ3_, and _SQ4_ . **(C)** Requirement of _SQ1_, _SQ2_, and _SQ4_ for _eTO5_ -triggered resistance.


Representative punch-inoculation phenotypes of wild-type and CRISPR-Cas9 knockout lines in


the Nortai and WRC43 backgrounds are shown at 10 days post-inoculation. Disruption of _SQ1_,


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


_SQ2_, or _SQ4_ abolished resistance to Ao92-06-2 expressing _eTO5_ . Images shown in (C) are


representative images from the corresponding experiments shown in Fig. S11. Scale bar, 1 cm. **(D)**


Growth phenotypes of segregating T 2 progeny derived from T 1 plants heterozygous for an _SQ3_


mutation in the Nortai and WRC43 backgrounds. Homozygous _sq3/sq3_ plants displayed severe


dwarfism, whereas wild-type ( _SQ3/SQ3_ ) and heterozygous ( _SQ3/sq3_ ) siblings showed normal


growth. Scale bars, 10 cm. **(E)** Genetic suppression of the _sq3_ dwarf phenotype. Additional


disruption of _SQ1_, _SQ2_, or _SQ4_ suppressed the dwarf phenotype in the _sq3_ background, showing


that the _sq3_ dwarf phenotype requires functional _SQ1_, _SQ2_, and _SQ4_ . Scale bar, 10 cm. **(F)**


Maximum-likelihood phylogeny based on the NB-ARC domains of rice NLR proteins, including


SQ1–SQ4. SQ1, SQ2, SQ3, and SQ4 fall into distinct phylogenetic clades, arguing against their


origin through a recent local duplication event.


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


**Figure 4. Phylogenetic distribution of** _**eTO**_ **copy-number variation and functional**


**association of eTO5 with Pi-eTO5 NLRs.**


**(A)** Maximum-likelihood phylogenetic tree of 65 _M. oryzae_ pathotype _Triticum_ (MT) isolates


based on genome-wide SNPs, midpoint-rooted and annotated with copy-number (CN) estimates


for six eTO loci ( _eTO1_, _eTO2_, _eTO5_, _eTO16_, _eTO24_, and _eTO31_ ). Geographic origin (Brazil,


Zambia, Bangladesh, and Bolivia) and collection period are indicated alongside the tree. Heatmap


colors represent estimated copy number: red indicates loci classified as absent (mean CDS depth


&lt; 5×), white indicates loci within the isolate-specific single-copy range, and the blue gradient


indicates increasing CN estimates above the isolate-specific empirical BUSCO P95 threshold. CN


estimates and BUSCO normalization statistics are provided in **Data S15** and S **16**, respectively.


The scale bar indicates 0.25 substitutions per site. **(B)** Co-immunoprecipitation assays examining


association of eTO5 with Pi-eTO5-linked NLR proteins. 4×Myc-tagged eTO5 was co-expressed


with 3×FLAG-tagged SQ1, SQ2, SQ3, or SQ4 in Nicotiana benthamiana. FLAG-tagged proteins


were immunoprecipitated using anti-FLAG agarose, and input and immunoprecipitated fractions


were analyzed by immunoblotting with anti-FLAG and anti-Myc antibodies. Pikp-1–3×FLAG


and AVR-PikD–4×Myc were included as controls. eTO5 co-immunoprecipitated with SQ2 and


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766612; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.14.766612)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc/4.0/)


SQ3 but was not detected with SQ1 or SQ4 under these conditions. Asterisks indicate nonspecific


bands. **(C)** Schematic model illustrating variation in the _eTO_ repertoire during geographic


expansion of the MT lineage and the potential use of rice resistance genes recognizing MT


effectors for deployment in wheat. _eTO5a_ denotes a diverged _eTO5_ copy present in later isolates


( **Fig. S16C** ).


16


## Supplementary Materials for

# **Dissecting rice nonhost resistance to the wheat blast fungus**

Motoki Shimizu*, Yukie Hiraka, Soichiro Asuke, Hiroaki Adachi, Sayaka Yuki, Eiko Kanzaki,

Kazue Ito, Akiko Hirabuchi, Naomi Miyaji, Takuma Kajino, Koki Minoji, Mina Ohtsu,
Toshiyuki Sakai, Mark J Banfield, Akira Abe, Yukio Tosa, Sophien Kamoun, Ryohei Terauchi


Corresponding authors: Motoki Shimizu, m-shimizu@ibrc.or.jp &amp;

Ryohei Terauchi, terauchi@ibrc.or.jp

**The PDF file includes:**


Materials and Methods
Fig. S1 to S18
References

**Other Supplementary Materials for this manuscript include the following:**


Data S1 to S17


1


**Materials and Methods**

**Fungal isolates and culture conditions**
The _M. oryzae_ isolates Ao92-06-2 and Br48 were maintained on potato sucrose agar (PSA). For
experiments, isolates were transferred to oatmeal agar and cultured at 25°C for 7 days. Aerial
mycelia were then removed by scraping to induce conidiation, and the cultures were incubated
under near-ultraviolet light for 2 days.

**Rice inoculation and lesion quantification**
Rice leaf-blade punch inoculation assays were performed using the indicated _M. oryzae_ strains.
Conidial suspension adjusted to 3×10⁵ conidia mL⁻¹ was applied to punched sites on leaves of rice
plants approximately 1 month after sowing. Prior to inoculation, wild-type plants were grown in a
naturally lit greenhouse under ambient temperature conditions, whereas transgenic plants were
grown in a naturally lit P1 containment greenhouse maintained at approximately 28°C. Inoculated
plants were maintained in a dew chamber at 27°C in darkness for 24 h and then transferred to a
growth chamber under a 16-h light/8-h dark photoperiod at 27°C. Lesion areas were quantified
using ImageJ software (1).

**Genome assembly of Br48**
Publicly available Illumina paired-end reads of the _M. oryzae_ pathotype _Triticum_ isolate Br48
(DDBJ Sequence Read Archive accession DRX054196; 2) were downloaded and quality-filtered
using FaQCs v2.10 (3). Filtered reads were assembled _de novo_ using MaSuRCA (4) with default
parameters. Assembly completeness was assessed using BUSCO v5.5.0 in genome mode with the
sordariomycetes_odb10 Data set (5). The assembly contained 97.1% complete BUSCOs (96.8%
single-copy and 0.3% duplicated), 0.9% fragmented BUSCOs, and 2.0% missing BUSCOs among
3,817 ortholog groups. Of these, 3,696 BUSCO genes classified as complete and single-copy were
subsequently used for depth normalization in the effector copy-number analysis ( **Fig. 4A** ).
Assembly metrics are summarized in **Data S1** .

**Identification of MT effector candidates**
Seeds of the barley cultivar Nigrate were sown in soil, and 1-week-old seedlings were inoculated
with _M. oryzae_ pathotype _Triticum_ isolate Br48. Leaf tissues were collected at 24 hpi from multiple
plants and bulked, and total RNA was extracted using the SV Total RNA Isolation System
(Promega). RNA-seq libraries were prepared using the NEBNext Ultra II Directional RNA Library
Prep Kit for Illumina (New England Biolabs) and sequenced to generate paired-end reads (DDBJ
Sequence Read Archive accession DRR594671). RNA-seq reads were quality-filtered using
FaQCs v2.10 (3) and aligned to the assembled Br48 genome using HISAT2 v2.2.1 (6) with the -dta option. Genome-guided transcript assembly and transcript abundance estimation were
performed using StringTie v2.2.1 (7), and transcript sequences were extracted using gffread
v0.12.7 (8). Transcripts shorter than 200 bp were excluded, resulting in 7,774 transcript sequences.
Open reading frames were predicted using TransDecoder v5.5.0 (9), and exact duplicate protein
sequences were removed using CD-HIT v4.8.1 (10) with a sequence identity threshold of 1.0.
Candidate cytoplasmic effector genes were selected using the pipeline shown in **Fig. S2** using the
following criteria: (i) the encoded protein contained a signal peptide predicted using SignalP v4.1
(11); (ii) transcript abundance exceeded 100 TPM at 24 hpi; (iii) the protein was classified as a
cytoplasmic or dual apoplastic/cytoplasmic effector by EffectorP v3.0 (12); and (iv) the predicted


2


protein was shorter than 150 amino acids. In total, 114 candidate cytoplasmic effector genes were
identified.

**Selection of** _**eTO**_ **genes**
To identify MT-specific effector candidates, publicly available whole-genome sequencing reads
from Br48 and nine MO isolates ( **Data S2** ) were quality-filtered using FaQCs v2.10 (3) and aligned
to the Br48 genome using Bowtie2 v2.3.5.1 (13). Alignments were sorted and indexed using
SAMtools v1.19.2 (14). For each of the 114 candidate effector genes, sequence coverage was
calculated using mosdepth v0.3.6 (15) as the proportion of nucleotide positions supported by a
read depth of at least 3× ( **Fig. S3** ). Genes were operationally defined as absent from MO when
sequence coverage was ≤30% in all nine MO isolates. This analysis identified 35 MT-specific
candidate effector genes, designated _eTO1_ to _eTO35_ (effector unique to MT versus MO; **Data S**
**3** ).

**Generation of** _**eTO**_ **-expressing MO transformants**
The coding sequences of _eTO1_ to _eTO35_, except _eTO29_, were amplified by PCR from cDNA
prepared from barley leaves infected with the MT isolate Br48 using the primers listed in **Data**
**S17** . Because _eTO29_ could not be amplified from cDNA, a synthetic coding sequence
corresponding to the predicted Br48 _eTO29_ sequence was obtained from Azenta Life Sciences.
The amplified or synthesized coding sequences were cloned downstream of the _pex22_ ( _AVR-Pia_ )
promoter in the pCB1531- _pex22p_ vector described previously (16), either by digestion with
BamHI and XbaI followed by ligation or by In-Fusion cloning (Takara Bio) into the BamHI/XbaI
-linearized vector. The method used for each construct is indicated in **Data S17** . The resulting
constructs were individually introduced into the MO isolate Ao92-06-2 by polyethylene glycol–
mediated transformation of fungal protoplasts as described previously (17). Transformants were
selected on bialaphos-containing medium, and the presence of each construct was confirmed by
PCR. Two independent transformants for each _eTO_ gene were used in the primary screen.

**Effector screening across rice cultivars**
The resulting Ao92-06-2 transformants expressing individual _eTO_ genes were inoculated onto 20
rice cultivars susceptible to the parental isolate Ao92-06-2, yielding 700 cultivar– _eTO_
combinations ( **Fig. S4** ; **Data S4** ). Cultivar– _eTO_ combinations scored as resistant in the primary
screen were retested in a secondary screen to confirm the phenotype. An interaction was scored as
resistant (R) when no lesion expansion beyond the inoculation site was observed and as susceptible
(S) when any lesion expansion was detected.

**Genetic validation of** _**eTO1**_ **- and** _**eTO24**_ **-triggered resistance**
To test whether _eTO1_ -triggered resistance requires _Pii_, Hitomebore and two independent _pii_
mutant lines, Hit5948 (18) and Hit5882 (19), were inoculated with Ao92-06-2 expressing _eTO1_ .
Similarly, to test whether _eTO24_ -triggered resistance requires _Pia_, Sasanishiki and two
independent _rga4_ mutant lines, Sas2127 and Sas1943 (20), were inoculated with Ao92-06-2
expressing _eTO24_ . Inoculations were performed using the punch inoculation procedure described
above.

**Rice genome sequencing and** _**de novo**_ **assembly**


3


High-molecular-weight genomic DNA was extracted from leaves of the rice cultivars Nortai,
WRC43, and Sesia using the NucleoBond HMW DNA kit (Macherey-Nagel) according to the
manufacturer's instructions. Libraries were prepared using the Ligation Sequencing Kit V14
(SQK-LSK114; Oxford Nanopore Technologies) and sequenced on a PromethION 2 Solo using
R10.4.1 flow cells. Raw signal data were base-called using Dorado v0.9.6 with the
dna_r10.4.1_e8.2_400bps_sup@v5.0.0 model and a minimum Q-score of 12. Reads from each
cultivar were assembled _de novo_ using hifiasm v0.25.0-r726 in ONT mode (21). Assemblies were
generated with duplication purging disabled (-l 0), an estimated genome size of 400 Mb, and the -rl-cut and --sc-cut parameters set to 10,000 and 12, respectively. Primary contig graphs were
converted to FASTA format using gfatools. Contigs were ordered and oriented into
pseudochromosomes using RagTag v2.1.0 (22), with IRGSP-1.0 (Nipponbare) as the reference
genome. Assembly completeness was assessed using BUSCO v5.5.0 (5) with the poales_odb10
lineage data set. Assembly statistics are provided in **Data S5** .

**QTL-seq analysis**
Three mapping populations were analyzed: 50 Nortai × Hitomebore recombinant inbred lines
(RILs), 83 WRC43 × Hitomebore F₂ individuals, and 38 Sesia × Hitomebore RILs. Resistance to
Ao92-06-2:: _pex22p_ : _eTO5_ was evaluated by punch inoculation, and lesion areas were measured
using ImageJ at 7–9 days post-inoculation. Individuals or lines with the smallest and largest lesion
areas were selected for the resistant bulk (RB) and susceptible bulk (SB), respectively. Each bulk
contained 20 individuals or lines for the Nortai × Hitomebore and WRC43 × Hitomebore
populations and 15 RILs for the Sesia × Hitomebore population. Genomic DNA was extracted
using the NucleoSpin Plant kit (Macherey-Nagel), pooled according to phenotype, and used to
prepare libraries with the Collibri ES DNA Library Prep Kit for Illumina (Thermo Fisher
Scientific). Paired-end sequencing was performed on an Illumina HiSeq X platform. Reads were
quality-filtered using fastp (23) and aligned to the Nortai reference genome using Bowtie2 (13).
BAM files were analyzed using the QTL-seq pipeline (18, 24). SNP-index and ΔSNP-index values
were calculated for each population, and genomic regions exceeding the 99% confidence limits
were considered candidate QTL regions.

**Fine mapping of** _**Pi-eTO5**_
CAPS markers were designed from sequence polymorphisms between Nortai and Hitomebore
using V-primer (25). PCR products were digested with XbaI or BamHI to distinguish the parental
alleles. Marker sequences, primer pairs, and restriction enzymes are provided in **Data S17** . From
a population of 250 Nortai × Hitomebore RILs, lines carrying recombination events between
markers M1 and M4 were selected and genotyped with additional markers within this interval.
Recombinant lines were inoculated with Ao92-06-2 and Ao92-06-2:: _pex22p_ : _eTO5_ . Recombinant
haplotypes and _eTO5_ -triggered resistance phenotypes were used to delimit the _Pi-eTO5_ locus.

**Transcriptome-guided gene annotation and identification of candidate NLR genes**
Total RNA was extracted from leaves of the rice cultivar Nortai using the SV Total RNA Isolation
System (Promega). Strand-specific RNA-seq libraries were prepared using the NEBNext Ultra II
Directional RNA Library Prep Kit for Illumina (New England Biolabs) and sequenced on an
Illumina HiSeq X platform to generate paired-end reads. Reads were adapter-trimmed and qualityfiltered using fastp (23), aligned to the Nortai genome assembly using HISAT2 (6), and subjected
to genome-guided transcript assembly using StringTie (7). Transcript sequences were extracted


4


using gffread (8), and open reading frames were predicted using TransDecoder (9). Predicted
proteins were analyzed using NLRtracker (26). Complete CNLs containing coiled-coil (CC),
nucleotide-binding NB-ARC, and leucine-rich repeat (LRR) domains were retained as candidate
NLRs. whereas degenerate NLRs lacking an intact NB-ARC domain were excluded. Four
complete CNLs were identified within the _Pi-eTO5_ candidate interval: _Nortai.21965.1_ ( _SQ1_ ),
_Nortai.21966.1_ ( _SQ2_ ), _Nortai.21967.1_ ( _SQ3_ ), and _Nortai.21968.1_ ( _SQ4_ ). Nortai.21962.5, a
degenerate NLR lacking an intact NB-ARC domain, was excluded. Genes supported by leaf RNAseq and annotated within this interval are listed in **Data S6** . The presence or absence of the _SQ1_ –
_SQ4_ haplotype was concordant with the eTO5-response phenotype: lines carrying _SQ1_ – _SQ4_ were
resistant, whereas lines lacking the haplotype were susceptible ( **Fig. S10B** ).

**Gene coverage analysis of** _**SQ1**_ **–** _**SQ4**_ **across rice cultivars**
To examine the distribution of _SQ1_, _SQ2_, _SQ3_, and _SQ4_ across the 20 rice cultivars used in the
functional screen, whole-genome sequencing reads were analyzed. Publicly available reads for the
Japanese Rice Core Collection (JRC) and World Rice Core Collection (WRC) accessions were
obtained from Tanaka et al. (27, 28), whereas reads for Hitomebore, Moukoto, Nortai, Sasanishiki,
Sesia, and WRC43 were generated in this study. Reads were adapter-trimmed and quality-filtered
using fastp (23), aligned to the Nortai reference genome using Bowtie2 v2.3.5.1 (13), and
processed using SAMtools v1.19.2 (14). For each cultivar, gene coverage of _SQ1, SQ2, SQ3,_ and
_SQ4_ was calculated using mosdepth v0.3.6 (15) as the percentage of nucleotide positions within
each annotated gene region supported by a read depth of at least 3×. Coverage values were
visualized as a heatmap. Cultivar names, sequence accession numbers, and data sources are
provided in **Data S4** . Responses to Ao92-06-2:: _pex22p_ : _eTO5_ were determined by the functional
screening described above and are shown below the heatmap in **Fig. S10C** .

**Generation of rice knockout mutants using the CRISPR-Cas9 system**
Rice knockout mutants of _SQ1, SQ2, SQ3_, and _SQ4_ were generated using the CRISPR-Cas9
system described previously (29). Guide RNA target sequences were designed using
CRISPRdirect (30). Complementary oligonucleotides corresponding to each target sequence were
annealed and cloned into the pU6::ccdB::gRNA vector at the BbsI site. The resulting gRNA
cassette containing the OsU6 promoter was subsequently transferred into the
pZH::gYSA::MMCas9 binary vector using AscI and PacI. Constructs were introduced into
_Agrobacterium tumefaciens_ strain EHA101 and transformed into the rice cultivars Nortai and
WRC43 as described previously (20). Regenerated T 0 plants were genotyped by PCR and Sanger
sequencing of the targeted regions. Independent mutant alleles carrying mutations predicted to
disrupt protein function were selected for subsequent analyses. T 1 plants or segregating T 2 progeny
were used for phenotypic analyses, as indicated in the corresponding figures. For _SQ3_,
heterozygous T 1 plants were self-pollinated, and the resulting T 2 progeny were genotyped by PCR–
restriction fragment length polymorphism analysis using AcuI or EcoNI to distinguish wild-type,
heterozygous, and homozygous mutant siblings. Mutations in the other edited genes were
determined by Sanger sequencing. Guide RNA target sequences and primers used for genotyping
are provided in **Data S17** .

**Generation and genotyping of double mutants**
Double-knockout plants were generated by introducing CRISPR-Cas9 constructs targeting _SQ1,_
_SQ2_, or _SQ4_ into calli induced from segregating T 2 seeds obtained from an _SQ3/sq3_ T 1 plant. After


5


regeneration, plants were genotyped at the _SQ3_ locus and the additionally targeted _SQ_ gene.
Among regenerated plants homozygous for the _sq3_ mutation, none carried a wild-type or
monoallelic mutant genotype (WT/mt1) at the additionally targeted locus; all carried either
homozygous (mt1/mt1) or biallelic (mt1/mt2) mutations. These plants were used as doubleknockout lines for growth analyses. Guide RNA target sequences and primers used for genotyping
are provided in **Data S17** .

**RNA sequencing and differential gene expression analysis of segregating** _**SQ3/sq3**_ **progeny**
Segregating T 2 progeny derived from a self-pollinated T 1 plant heterozygous for the _SQ3_ mutation
were grown on Murashige and Skoog medium under sterile conditions. At 7 days after germination,
seedlings were genotyped, and five individual _SQ3/SQ3_ and five individual _sq3/sq3_ seedlings were
selected before visible growth defects appeared. Each seedling was analyzed independently as a
biological replicate. Total RNA was extracted from whole seedlings using the SV Total RNA
Isolation System (Promega). Strand-specific RNA-seq libraries were prepared using the NEBNext
Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs) and sequenced on
an Illumina NovaSeq platform to generate paired-end reads. Reads were adapter-trimmed and

–
quality-filtered using fastp (23). Because the _SQ1_ _SQ4_ locus is absent from the public IRGSP-1.0
Nipponbare reference genome, the corresponding Nortai genomic region was appended to the

–
reference together with gene models for _SQ1_ _SQ4_ . Reads were aligned to this composite reference
genome using HISAT2 (6), retaining only properly paired, uniquely mapped read pairs (MAPQ ≥
20). Gene-level read counts were obtained using featureCounts (31). Differential expression was
analyzed a quasi-likelihood F-test in edgeR (32), with genotype ( _sq3/sq3_ versus _SQ3/SQ3_ ) as the
sole factor. Genes with a false-discovery rate (FDR) &lt; 0.05 and |log2-fold change| ≥ 1 were
considered differentially expressed ( **Data S7** ). For principal component analysis, raw gene counts
were normalized by variance-stabilized using DESeq2 (33), and PCA was performed on the
transformed values. For visualization of individual gene expression, transcript abundance was
expressed as transcripts per million (TPM), calculated from gene-level read counts using the exonunion gene length reported by featureCounts, and shown as log2 (TPM + 1).
Functional enrichment analysis of genes significantly upregulated in each genotype was performed
using clusterProfiler v4.10.0 (34) with a custom Gene Ontology Biological Process (GO BP)
annotation derived from the RAP-DB IRGSP-1.0 representative gene annotation. The background
set was restricted to retained after low-count filtering in the edgeR analysis. GO BP terms with
Benjamini-Hochberg (BH)-adjusted P value &lt; 0.05 were considered significantly enriched ( **Data**
**S8** ). Genes upregulated in _sq3/sq3_ within significantly enriched defense-related GO terms were
visualized as heat maps of VST-normalized, row-scaled expression values. Immune-related genes
belonging to NB-LRR, WRKY, pathogenesis-related, and receptor-like kinase families were
identified from RAP-DB annotation using InterPro domain and description keywords and
visualized in the same manner.

**Phylogenetic analysis of rice NLR proteins**
Predicted protein sequences ≥800 amino acids in length from the _Oryza sativa_ ssp. _japonica_ cv.
Nipponbare reference annotation were obtained from the Rice Annotation Project Database (RAPDB) (35). NLR proteins were identified using NLRtracker (26). Degenerate NLRs lacking an intact
NB-ARC domain were excluded, and genes sharing a duplicate gene identifier were consolidated
into single entries. The predicted protein sequences of SQ1, SQ2, SQ3, and SQ4, together with the
paired NLRs Pii-1 (AB820896) and Pii-2 (LC190730) (18, 19), Pikm-1 (AB462324) and Pikm-2


6


(AB462325) (36), Pit-1 (AB379816) and Pit-2 (AB379819) (37), and RGA4 (AB604622) and
RGA5 (AB604627) (20), were added to the Nipponbare NLR data set. NB-ARC domain sequences
were extracted from each NLR protein and aligned using MUSCLE v5.2 (38). A maximumlikelihood phylogenetic tree was inferred using RAxML-NG v1.2.2 under the JTT+G4+F amino
acid substitution model with 1,000 nonparametric bootstrap replicates (39, 40). The resulting tree
was visualized and annotated using iTOL v6 (41). Genes and NB-ARC domain sequences used in
this analysis are listed in **Data S9** .

**Conservation analysis of** _**SQ1**_ **–** _**SQ4**_ **across** _**Oryza**_ **genomes**
To assess the conservation of _SQ1, SQ2, SQ3_, and _SQ4_ across cultivated and wild rice, we first
analyzed Nanopore-based genome assemblies of _O. sativa_ _japonica_ and _indica_ and the wild
progenitor _O. rufipogon_ (42–44). The predicted Nortai amino acid sequences of SQ1, SQ2, SQ3,
and SQ4 were used as queries against each genome assembly using Exonerate v2.4.0 in
protein2genome mode (45), and the highest-scoring alignment was retained for each gene and
accession. Relative similarity to Nortai was calculated by normalizing the Exonerate bit score for
each accession to that obtained by aligning the corresponding Nortai sequence to the Nortai
genome (relative score = Bᵢ,ⱼ / Bᵢ,Nortai) ( **Data S10** ). Accessions with relative scores ≥ 0.6 for all
four genes were conservatively classified as carrying a putative complete SQ1–SQ4 haplotype.
Such haplotypes were detected in 20 of 143 _O. sativa_ accessions and 15 of 109 _O. rufipogon_
accessions, whereas most other accessions showed only low alignment scores, consistent with
presence/absence-type variation of the SQ1–SQ4 region. We next extended the analysis to a
broader panel of wild _Oryza_ species using short-read-based genome assemblies (46). Putative
complete SQ1–SQ4 haplotypes were likewise detected in only 12 of 108 accessions, which were
largely restricted to _O. glumaepatula_ ( **Data S10** ). Because these assemblies were generated from
short reads, however, physical linkage of the four genes could not be confirmed. To examine the
locus structure directly, we generated Nanopore _de novo_ genome assemblies for two representative
wild accessions, W2149 ( _O. glumaepatula_ ) and W1577 ( _O. punctata_ ). Genomic DNA was
sequenced using Oxford Nanopore R9.4.1 flow cell chemistry. Reads were base-called using
Dorado v0.9.0 with the dna_r9.4.1_e8_sup@v3.6 super-accuracy model and a minimum q-score
of 8, error-corrected with Ratatosk (47) using newly generated Illumina short reads, and assembled
_de novo_ using hifiasm v0.25.0-r726 in ONT mode ( **Data S11** ). For accessions classified as carrying
putative complete SQ1–SQ4 haplotypes, amino acid sequences of SQ1–SQ4 were manually
determined from CDS-based alignments and used for haplotype comparison ( **Data S12** **and S13** ).
To examine the evolutionary relationships of SQ1, SQ2, SQ3, and SQ4 alleles, amino acid
sequences were aligned independently for each gene using MAFFT v7 with the –auto option (48).
Maximum-likelihood phylogenetic trees were inferred using IQ-TREE v2 with automatic model
selection (-m MFP) (49, 50), with branch support assessed using 1,000 ultrafast bootstrap
replicates and 1,000 SH-aLRT replicates (-bb 1000 -alrt 1000).

**Conservation analysis of AVR–active** _**eTO**_ **genes across MT isolates**
Sixty-five _M. oryzae_ pathotype _Triticum_ (MT) isolates were characterized ( **Data S14** ). Wholegenome short-read sequencing data were obtained from public databases (2, 51–56). Qualitytrimmed paired-end data were mapped to the Br48 reference genome using BWA MEM (57). PCR
duplicates were marked using Picard MarkDuplicates, and BAM files were indexed with
SAMtools (14).


7


A set of 3,696 complete single-copy BUSCO genes identified using the sordariomycetes_odb10
data set (5) was used for depth normalization. Mean read depth across BUSCO loci and the coding
sequences of six AVR-active _eTO_ genes ( _eTO1, eTO2, eTO5, eTO16, eTO24_, and _eTO31_ ) was
calculated using mosdepth (15), retaining reads with mapping quality ≥20. For each isolate,
normalized copy-number (CN) was calculated as the mean depth of the _eTO_ coding sequence
divided by the median depth of the 3,696 BUSCO genes. An eTO locus with a mean CDS depth
&lt;5× was classified as absent. For loci with mean CDS depth ≥5×, an isolate-specific single-copy
threshold was defined as the empirical 95th percentile of normalized BUSCO depths (BUSCO P95
depth/BUSCO median depth). Loci with normalized CN values above this threshold were
classified as amplified, whereas those at or below the threshold were classified as single-copy.
Because depth was measured against the Br48 reference assembly, CN values reflect the combined
depth of each _eTO_ locus and of any closely related sequences that are present in an isolate but not
represented in the reference. CN estimates are provided in **Data S15**, and BUSCO depth statistics
and isolate-specific thresholds are provided in **Data S16** .
SNPs were called against the Br48 reference genome using bcftools (14). Filtering: Variants were
retained if they were biallelic SNPs with QUAL ≥30, depth ≥10, and ≤10% missing data. The
filtered VCF was converted to PHYLIP format using vcf2phylip (58). Maximum-likelihood
phylogenetic trees were inferred using IQ-TREE 2 (59) under the GTR+ASC model (60), with
branch support assessed using 1,000 ultrafast bootstrap replicates (61) and SH-aLRT (62). Trees
were visualized and annotated using iTOL (41) with isolate metadata and copy-number data from
**Data S15** .

_**De novo**_ **assembly of MT isolates and assembly-based detection of eTO5-, eTO5a-, and eTO1-**
**related sequences**
To obtain assembly-based evidence for the presence of multiple eTO5- and eTO1-related
sequences in representative MT isolates ( **Fig. S16B** ), publicly available Oxford Nanopore long
reads were assembled _de novo_ for selected isolates from Bangladesh and Zambia. Nanopore reads
for the Bangladeshi isolates were obtained from BioProject PRJEB27137 (MinION; 63), and those
for the Zambian isolates from BioProject PRJEB53707 (PromethION; The John Innes Centre,
2022). Matched Illumina paired-end reads used for error correction were obtained from BioProject
PRJEB22035 for the Bangladeshi isolates (55) and PRJEB43836 for the Zambian isolates (56).
Illumina reads were adapter-trimmed and quality-filtered using fastp v1.3.6 (23). Nanopore reads
were hybrid error-corrected with the matched Illumina reads using Ratatosk v0.9.0 (47) in correct
mode. Corrected reads were assembled _de novo_ using hifiasm v0.25.0-r726 (21) in ONT mode
with duplication purging disabled, and the primary contig graph was converted to FASTA format
using gfatools v0.4-r214.
The amino acid sequences of eTO5, eTO5a, and eTO1 were searched against each assembly using
tblastn v2.12.0+ (64) with default parameters. Because eTO5 and eTO5a differ at only 7 of 96
amino acid positions and therefore cross-detect each other, only alignments spanning the full query
length with 100% amino acid identity were retained, allowing each retained hit to be assigned
unambiguously to one of the three query sequences. The positions of retained hits were plotted
along the corresponding contigs using a custom Python script based on matplotlib v3.6.3.

_**Nicotiana benthamiana**_ **growth conditions**
Wild-type _Nicotiana benthamiana_ plants were grown in a growth chamber at 23-25°C under longday conditions (16-h light/8-h dark).


8


**Co-immunoprecipitation assay**
To test associations between the Pi-eTO5-linked NLRs and eTO5, NLR (SQ1, SQ2, SQ3, SQ4,
and Pikp-1) and effector (eTO5 and AVR-PikD) genes were synthesized with codon optimization
for _Nicotiana benthamiana_ (Integrated DNA Technologies). The synthesized genes were
assembled into binary vector pJK268c (26) using the Golden Gate system with the MoClo Tool
Kit (65) and MoClo Plant Parts Kit (66). The promoter module pICH51288 (2x35S; Addgene no.
50269) and terminator module pKH1001, containing synthesized _Arabidopsis thaliana HEAT_
_SHOCK PROTEIN 18.2_ (At5g59720) terminator, were used for all constructs. NLRs and effectors
were tagged using 3×FLAG (pICSL50007; Addgene no. 50308) and 4×Myc (pICSL30009;
Addgene no. 50301), respectively.
Transient gene expression in _N. benthamiana_ was performed by agroinfiltration as described
previously (67). Briefly, leaves of four-week-old _N. benthamiana_ plants were infiltrated with
_Agrobacterium tumefaciens_ strain GV3101 carrying the indicated binary constructs using
needleless syringes. _Agrobacterium_ cultures were resuspended in infiltration buffer containing 10
mM MES-NaOH (pH 5.6), 10 mM MgCl 2, and 150 µM acetosyringone and adjusted to an OD 600
of 0.5. Cultures carrying the NLR and effector constructs were mixed at a 1:1 ratio before
infiltration.
Two infiltrated leaves were harvested 3 days after infiltration, frozen in liquid nitrogen, and ground
to a fine powder. Total proteins were extracted with two volumes of ice-cold GTEN-based
extraction buffer relative to tissue fresh weight &#91;25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM
EDTA, 10% (v/v) glycerol, 10 mM DTT, 0.3% (v/v) NP-40, 2% (w/v) polyvinylpolypyrrolidone
(PVPP), and protease inhibitor cocktail (Merck, P9599)&#93;. Extracts were clarified by centrifugation
at 10,000 × _g_ for 10 min at 4°C, and an aliquot of each clarified protein extract was used as the
input sample. FLAG-tagged NLR proteins were immunoprecipitated by incubating the clarified
extracts with 20 µL of anti-FLAG agarose beads (Merck, A2220) for 1 h at 4°C. Beads were
washed five times with GTEN buffer supplemented with 0.3% (v/v) NP-40. Bound proteins were
eluted by incubation with FLAG peptide at a final concentration of 150 µg mL⁻¹ for 30 min at 4°C.
Immunoprecipitated proteins and input samples were separated by SDS-PAGE and transferred to
PVDF membranes using a Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were
blocked with 3% (w/v) skim milk in TBS-T &#91;50 mM Tris-HCl (pH 7.6), 150 mM NaCl, and 0.05%
(v/v) Tween 20&#93; for 1 h at room temperature and subsequently incubated with Horseradish
peroxidase (HRP)-conjugated anti-FLAG (Merck, A8592) or anti-Myc (Santa Cruz Biotechnology,
SC-40 HRP) antibodies at a 1:5,000 dilution for 1 h at room temperature. After washing with TBST, chemiluminescent signals were detected using SuperSignal West Dura Extended Duration
Substrate (Thermo Scientific) and a Luminograph II EM imaging system (ATTO). Protein loading
and transfer were assessed by Ponceau S staining. The co-immunoprecipitation experiment was
performed independently three times with similar results.


9


**Fig. S1. Host specificity of MO and MT isolates on rice.**
Lesion areas were measured on leaves of three _Oryza sativa_ cultivars (Hitomebore, Sasanishiki,
and Nipponbare) inoculated with the rice-infecting MO isolate Ao92-06-2 or the wheat-infecting
MT isolate Br48. Each dot represents an individual plant. Horizontal lines indicate the median,
and error bars indicate the interquartile range. Differences between Ao92-06-2 and Br48 were
assessed separately for each cultivar using a two-tailed Mann–Whitney U test. _P_ values are
indicated above the comparisons.


10


**Fig. S2. Pipeline for identification of candidate cytoplasmic effector genes from** _**M. oryzae**_
**pathotype** _**Triticum**_ **isolate Br48.**
RNA was extracted from leaves of barley cultivar Nigrate at 24 hours post-inoculation with the
MT isolate Br48. RNA-seq reads were processed through the following steps: (1) quality control
using FaQCs v2.10; (2) alignment to the Br48 genome using HISAT2 v2.2.1; (3) genome-guided
transcriptome assembly using StringTie v2.2.1, followed by transcript sequence extraction using
gffread v0.12.7, yielding 7,774 transcripts longer than 200 bp; (4) open reading frame prediction
using TransDecoder v5.5.0 and removal of exact duplicate protein sequences using CD-HIT
v4.8.1; and (5) selection of candidates encoding SignalP-predicted secreted proteins, showing
expression of TPM &gt; 100 at 24 hpi, classified as cytoplasmic or cytoplasmic/apoplastic effectors
by EffectorP v3.0, and encoding proteins shorter than 150 amino acids. This pipeline identified
114 candidate cytoplasmic effector genes.


11


**Fig. S3. Pipeline for estimating candidate effector gene coverage across** _**M. oryzae**_ **isolates.**
Whole-genome sequencing reads from the MT isolate Br48 and nine MO isolates were qualityfiltered using FaQCs v2.10 and aligned to the Br48 reference genome using Bowtie2 v2.3.5.1. The
resulting alignments were processed into sorted and indexed BAM files. A BED file containing
the genomic coordinates of the 114 candidate effector genes in the Br48 genome was used for
coverage analysis with mosdepth v0.3.6. For each candidate effector gene, gene coverage was
calculated as the percentage of nucleotide positions supported by a read depth of ≥3×.


12


**Fig. S4. Pipeline for systematic identification of MT-specific AVR genes and their cognate**
**rice resistance loci.**
**(A)** Functional screen of MT-specific effector genes. Individual _eTO_ genes are introduced into the
rice-infecting MO isolate Ao92-06-2, and the resulting transformants are inoculated onto multiple
rice lines. An _eTO_ that specifically triggers resistance in a rice line is identified as a candidate
AVR gene. In the example shown, _eTO-D_ triggers resistance in Line-A but not in Line-B. R and
S indicate resistant and susceptible responses, respectively. **(B)** Mapping of the resistance locus
cognate to the identified AVR gene. The resistant Line-A is crossed with the susceptible Line-B,
and F 2 progeny or recombinant inbred lines (RILs) are inoculated with the Ao92-06-2 transformant
expressing _eTO-D_ . Resistant and susceptible progeny are pooled separately to generate R-bulks
and S-bulks, which are subjected to whole-genome sequencing and QTL-seq analysis. Differences
in allele frequencies between the two bulks are used to identify the resistance locus responsible for
_eTO-D_ -triggered resistance.


13


**Fig. S5. Complete compatibility matrix for functional screening of 35 MT-specific effector**
**genes across 20 rice cultivars.**
The parental MO isolate Ao92-06-2 and 35 Ao92-06-2 transformants, each expressing an
individual _eTO_ gene under the control of the _pex22_ promoter, were inoculated onto 20 rice cultivars
that were susceptible to the parental isolate. Columns indicate rice cultivars ( **Data S4** ), and rows
indicate the parental isolate or transformants expressing individual _eTO_ genes. Filled blue circles
indicate incompatible interactions (resistance), whereas open circles indicate compatible
interactions (susceptibility). The parental Ao92-06-2 isolate was compatible with all cultivars
tested. Six _eTOs_ ( _eTO1, eTO2, eTO5, eTO16, eTO24_, and _eTO31_ ) triggered resistance in at least
one cultivar.


14


15


**Fig. S7. Amino acid sequence alignments eTO1 with AVR-Pii and eTO24 with AVR1-CO39.**
(A) Pairwise amino acid sequence alignment of eTO1 and AVR-Pii. (B) Pairwise amino acid
sequence alignment of eTO24 and AVR1-CO39. Identical amino acid residues are highlighted in

gray.


16


**Fig. S8.** _**Pii**_ **and** _**Pia**_ **are required for** _**eTO1-**_ **and** _**eTO24**_ **-triggered resistance, respectively.**
**(A)** Lesion areas in Hitomebore, which carries the paired NLR genes _Pii-1_ and _Pii-2_, and two
independent _pii_ mutant lines, Hit5948 and Hit5882, following inoculation with Ao92-06-2
expressing _eTO1_ under the control of the pex22 promoter. **(B)** Lesion areas in Sasanishiki, which
carries the paired NLR genes _RGA4_ and _RGA5_ at the _Pia_ locus, and two independent _rga4_ mutant
lines, Sas2127 and Sas1943, following inoculation with Ao92-06-2 expressing _eTO24_ under the
control of the pex22 promoter. Each dot represents an individual plant. Horizontal lines indicate
the median, and error bars indicate the interquartile range. Plus and minus symbols indicate the
presence or absence of _Pii-1_ and _Pii-2_ in (A) and _RGA4_ and _RGA5_ in (B). Statistical significance
was assessed using a Kruskal–Wallis test followed by Dunn’s multiple-comparison test. _P_ values
are indicated above the comparisons.


17


18


19


20


**Fig. S9. Independent QTL-seq analyses identify an overlapping** _**Pi-eTO5**_ **interval on**
**chromosome 11.**
Genome-wide ΔSNP-index profiles obtained from recombinant inbred line (RIL) populations
derived from **(A)** Nortai × Hitomebore and **(C)** Sesia × Hitomebore, and from **(B)** an F 2 population
derived from WRC43 × Hitomebore, are shown. QTL-seq analyses were performed using the
Nortai genome as the reference. Blue dots indicate ΔSNP-index values for individual SNPs. The
red line indicates the sliding-window-smoothed ΔSNP-index. Green and orange lines indicate the
95% and 99% confidence limits, respectively. The three populations identified intervals of 24.7–
29.6 Mb, 24.4–28.4 Mb, and 19.9–27.8 Mb on chromosome 11, respectively, defining a common
interval of 24.7–27.8 Mb associated with _eTO5-_ triggered resistance.


21


**Fig. S10. Fine mapping and cultivar distribution of the** _**Pi-eTO5**_ **NLR quartet.**
**(A)** Fine mapping of _Pi-eTO5_ using recombinant progeny derived from the Nortai×Hitomebore
cross. Recombinant haplotypes between markers M1 and M4 are shown together with
representative punch-inoculation phenotypes and lesion areas following inoculation with Ao9206-2 or Ao92-06-2 expressing _eTO5_ . Horizontal lines in the right graph indicate the median, and
error bars indicate the interquartile range. Differences between Ao92-06-2 and Ao92-06-2
expressing _eTO5_ were assessed within each rice line using two-way ANOVA followed by Šídák’s
multiple-comparison test. _P_ values are indicated; ns, not significant. The interval delimited by

–
recombinants #91 and #24 spans approximately 0.53 Mb and contains the _SQ1_ _SQ4_ gene cluster.

–
Scale bar, 1 cm **. (B)** PCR-based genotyping of marker M2 and _SQ1_ _SQ4_ in the parental lines and
recombinant progeny shown in **(A)** . **(C)** Gene coverage of _SQ1_, _SQ2_, _SQ3_, and _SQ4_ across the 20
rice cultivars used for functional screen. Gene coverage was calculated as the percentage of
nucleotide positions supported by a read depth of ≥ 3×. The compatibility response to Ao92-06-2
expressing _eTO5_ is shown below. Filled blue circles indicate incompatible interactions, whereas
open circles indicate compatible interactions. High gene coverage of all four NLR genes was
observed in Nortai, Sesia, and WRC43, the three cultivars that exhibited _eTO5_ -triggered resistance.


22


**Fig. S10. (continued)**



23


**resistance.**
**(A, C, E)** CRISPR-Cas9 target sites and mutations introduced into _SQ1_, _SQ2_, and _SQ4_,
respectively. Gene structures and target positions are shown above the aligned wild-type and
mutant sequences. Protospacer sequences are boxed, and protospacer-adjacent motifs are shown
in red. Insertions are indicated by triangles, and deletions by dashes. The predicted consequences
of each mutation, including premature termination and the resulting protein length, are indicated
on the right. Homo, homozygous; Heteroallelic, carrying two distinct mutant alleles; T 1 and T 2
indicate plant generations. **(B, D, F)** Lesion areas and representative punch-inoculation phenotypes
of wild-type and mutant lines for _SQ1_, _SQ2_, and _SQ4_, respectively, in the (a) Nortai and (b)
WRC43 backgrounds. Plants were inoculated with the parental Ao92-06-2 isolate or Ao92-06-2
expressing _eTO5_ . Each dot represents an individual plant. Horizontal lines indicate the median,
and error bars indicate the interquartile range. Statistical significance was assessed using a twotailed Mann–Whitney U test. P values are indicated above the comparisons. Scale bars, 1 cm.


24


**Fig. S11. (continued)**



25


**Fig. S11. (continued)**



26


**Fig. S11. (continued)**



27


**Fig. S11. (continued)**



28


**Fig. S11. (continued)**



29


**Fig. S12. Independent disruption of** _**SQ3**_ **causes severe dwarfism, whereas disruption of** _**SQ1**_ **,**
_**SQ2**_ **, or** _**SQ4**_ **does not affect plant growth.**
**(A)** CRISPR-Cas9 target sites and mutations introduced into _SQ3_ . The gene structure and target
positions are shown above the aligned wild-type and mutant sequences. Protospacer sequences are
boxed, and protospacer-adjacent motifs are shown in red. Insertions are indicated by triangles and
deletions by dashes. **(B)** Growth phenotypes of segregating T 2 progeny derived from independent
T 1 plants heterozygous for _SQ3_ mutations in the (a) Nortai and (b-d) WRC43 backgrounds.
Representative plants and plant-height measurements are shown for _SQ3/SQ3_ (+/+), _SQ3/sq3_ (+/−),
and _sq3/sq3_ (−/−) progeny. Homozygous _sq3_ plants displayed severe dwarfism, whereas wild-type
and heterozygous siblings showed normal growth. Statistical significance was assessed separately
for each edited line using one-way ANOVA followed by Tukey’s multiple-comparison test. _P_
values are indicated above the comparisons; ns, not significant. Scale bars, 10 cm. **(C)**
Representative growth phenotypes and plant-height measurements of _SQ1_, _SQ2_, and _SQ4_
knockout lines in the (a) Nortai and (b) WRC43 backgrounds. The Nortai lines were _sq1_ (#5-1-4),
_sq2_ (#12-1-1), and _sq4_ (#15-1-3), and the WRC43 lines were _sq1_ (#20-3-2), _sq2_ (#32-1-9), and
_sq4_ (#40-1-3). Statistical significance was assessed separately for each genetic background using
one-way ANOVA followed by Dunnett’s multiple-comparison test, with each knockout line
compared with the corresponding wild type. No significant differences in plant height were
detected. ns, not significant. Scale bars, 10 cm.


30


**Fig. S12. (continued)**



31


**Fig. S12. (continued)**



32


**Fig. S13. Additional disruption of** _**SQ1**_ **,** _**SQ2**_ **, or** _**SQ4**_ **suppresses the severe dwarf phenotype**
**of** _**sq3**_ **plants.**
**(A)** Strategy used to generate double-knockout lines in the _sq3/sq3_ background. Segregating T 2
seeds derived from the T 1 _SQ3/sq3_ line #3-1-2 were used for callus induction. CRISPR-Cas9
constructs targeting _SQ1_, _SQ2_, or _SQ4_ were introduced into the calli, followed by plant
regeneration and genotyping. The _SQ3_ genotype and target-site sequences of line #3-1-2 are shown
below. **(B)** Mutations introduced into _SQ1_, _SQ2_, or _SQ4_ in the _sq3/sq3_ background. Wild-type and
edited target-site sequences are shown. Protospacer sequences are boxed, and protospacer-adjacent
motifs are shown in red. Insertions are indicated by triangles and deletions by dashes. The
predicted consequences of each mutation, including premature termination and the resulting
protein length, are indicated on the right. **(C)** Growth phenotypes and plant-height measurements
of _sq3/sq3_ single mutants and double mutants additionally disrupted for _SQ1_, _SQ2_, or _SQ4_ . The
upper panels correspond to the initial experiment shown in **Fig. 3E**, whereas the lower panels show
an independent biological replicate using additional edited lines. The _sq3/sq3_ plants displayed
severe dwarfism, whereas additional disruption of _SQ1_, _SQ2_, or _SQ4_ restored plant growth in both
experiments. Each dot represents an individual plant. Horizontal lines indicate the mean, and error
bars indicate the standard deviation. Statistical significance was assessed separately for each
experiment using one-way ANOVA followed by Dunnett’s multiple-comparison test, with the _sq3_
single mutant as the reference group. _P_ values are indicated above the comparisons. Scale bars, 10

cm.


33


34


**Fig. S13. (continued)**



35


**Fig. S14. Loss of** _**SQ3**_ **activates defense-related transcription before visible growth defects**

**appear.**
**(A)** Experimental design for transcriptome analysis. Segregating T 2 progeny derived from an
_SQ3/sq3_ heterozygous plant were grown on Murashige and Skoog (MS) medium under sterile
conditions. After 1 week, seedlings were genotyped, and five _SQ3/SQ3_ and five _sq3/sq3_ seedlings
were selected for RNA extraction and RNA sequencing before visible growth defects appeared.
**(B)** Principal component analysis of RNA-seq samples based on normalized gene-expression
profiles. _SQ3/SQ3_ and _sq3/sq3_ samples separated primarily along PC1, which explained 77.4% of
the total variance. **(C)** Volcano plot of differential gene expression in _sq3/sq3_ relative to _SQ3/SQ3_
seedlings. Red and cyan dots indicate genes significantly upregulated in _SQ3/SQ3_ and _sq3/sq3_
seedlings, respectively; gray dots indicate genes that were not significantly differentially expressed.
Differentially expressed genes were defined by |log2 fold change| ≥ 1 and FDR &lt; 0.05 (dashed


36


lines). **(D)** Numbers of genes significantly upregulated in _SQ3/SQ3_ and _sq3/sq3_ seedlings. **(E)**
Gene Ontology (Biological Process) enrichment analysis of genes significantly upregulated in
_sq3/sq3_ (left) and _SQ3/SQ3_ (right) seedlings. Dot size indicates the number of genes annotated to
each term (Count); color indicates the BH-adjusted _P_ value. Only GO BP terms with an adjusted
_P_ value &lt; 0.05 are shown. **(F)** Heat map of differentially expressed genes belonging to significantly
enriched defense-related GO BP terms : chitin catabolic process (Chitinase) and response to
oxidative stress/hydrogen peroxide catabolic process (Peroxidase/oxidative stress). Gene
membership reflects the enrichment result shown in **(E)** . Color represents row-scaled (z-score)
VST-normalized expression; each column represents one biological replicate. **(G)** Heat map of
immune-related genes showing higher expression in _sq3/sq3_ than in _SQ3/SQ3_, irrespective of
statistical significance: NLR receptors, pathogenesis-related proteins, receptor-like kinases, and
WRKY transcription factors. Genes were identified by keyword search of RAP-DB InterPro and
Description. Color represents row-scaled (z-score) VST-normalized expression; each column
represents one biological replicate. **(H)** Expression levels of _SQ1_, _SQ2_, _SQ3_, and _SQ4_ in _SQ3/SQ3_
and _sq3/sq3_ seedlings. Expression values are shown as log 2 (TPM + 1). Each dot represents an
individual biological replicate, and horizontal lines indicate the mean.


37


**Fig. S14. (continued)**



38


**Fig. S14. (continued)**



39


**Fig. S14. (continued)**



40


**Fig. S14. (continued)**



41


42


**Fig. S15. Conservation of** _**SQ1–SQ4**_ **across cultivated and wild** _**Oryza**_ **genomes.**
**(A)** Relative Exonerate score (normalized to Nortai) for SQ1–SQ4 in _O. sativa_ accessions from
the _japonica_, _aus_, _basmati_, and _indica_ groups. **(B)** Relative Exonerate score for SQ1–SQ4 in _O._
_rufipogon_ accessions from previously defined genetic groups _Or-Ia_, _Or-Ib_, _Or-II_, _Or-IIIa_, and _Or-_
_IIIb_ (44). **(C)** Relative Exonerate score for SQ1–SQ4 in other wild _Oryza_ species ( _O. barthii_, _O._
_glumaepatula_, _O. meridionalis_, _O. punctata_ &#91;BB&#93;, _O. officinalis_ &#91;CC&#93;, _O. australiensis_ &#91;EE&#93;, _O._
_brachyantha_ &#91;FF&#93;, _O. granulata_ &#91;GG&#93;). In (A)–(C), darker colors indicate higher relative similarity
to the corresponding Nortai protein sequence. Accessions with a relative score ≥ 0.6 for all four
genes were conservatively classified as carrying a putative complete SQ1–SQ4 haplotype and are
indicated in blue. Among the remaining accessions, haplotypes are numbered in order of
decreasing frequency, with ties broken by increasing amino acid divergence from Nortai. **(D)**
Maximum-likelihood phylogenetic trees of (a) SQ1, (b) SQ2, (c) SQ3, and (d) SQ4 amino acid
sequences across the accessions in (A)–(C). Sequences were aligned with MAFFT v7 (--auto) and
the tree was inferred with IQ-TREE v2 using ModelFinder (-m MFP), with branch support
assessed by 1,000 ultrafast bootstrap replicates and 1,000 SH-aLRT replicates. ( **E** ) Gene
organization of the SQ1–SQ4 region and amino acid identity (%) of each SQ protein relative to
Nortai in II32 ( _O. sativa indica_ ), W2149 ( _O. glumaepatula_ ), and W1577 ( _O. punctata_ ). Arrows
indicate gene orientation. Gene positions in W2149 and W1577 were confirmed by Nanopore
sequencing and _de novo_ assembly (see Methods).


43


44


45


46


47


48


**Fig. S16. Structural conservation and genomic copy-number variation of ZiF-family**
**effectors.**
**(A)** Structural and sequence conservation among the ZiF-family effectors AVR-Pii, eTO1, and
eTO5. Ribbon representations of AVR-Pii (orange; PDB ID: 7PP2) and AlphaFold 3-predicted
structures of eTO1 (green) and eTO5 (cyan) are shown. Zinc ions are depicted as gray spheres,
and the side chains of cysteine and histidine residues coordinating the zinc ion are shown as sticks.
Amino acid sequence alignment of AVR-Pii, eTO1, and eTO5 highlights conservation of residues
forming the zinc-binding motif. **(B)** Assembly-based evidence for multiple eTO5/eTO1-family
copies in representative isolates from Zambia and Bangladesh. Positions of _eTO5_, _eTO5a_, and
_eTO1_ sequence hits are shown on assembled contigs of the Zambian isolates ZMW20-16 and
ZMW18-11 and the Bangladeshi isolate BTJP4-1. Contig lengths are indicated in parentheses.
Corresponding read-depth-based copy-number (CN) estimates for _eTO5_ and _eTO1_ are shown for
each isolate. **(C)** Sequence and functional comparison of eTO5 and eTO5a. Amino acid sequence
alignment of eTO5 and eTO5a shows the sequence differences between the two proteins, with
identical residues indicated by asterisks. Representative leaves and lesion-area measurements are
shown following inoculation with Ao92-06-2, Ao92-06-2:: _eTO5_, and Ao92-06-2:: _eTO5a_ .


49


**Fig. S16. (continued)**



50


**Fig. S16. (continued)**



51


**Fig. S17. Domain architecture of Pi-eTO5 NLR proteins.**
Schematic representation of the predicted domain structures of SQ1, SQ2, SQ3, and SQ4,
indicating conserved motifs associated with NLR function. All four proteins contain the P-loop
and MHD motifs characteristic of NLR proteins. SQ4 uniquely contains a MADA motif, a
conserved N-terminal motif associated with NLR-mediated cell-death activity. The distinct
distribution of NLR-associated motifs among SQ1–SQ4 is consistent with functional
specialization within the Pi-eTO5 NLR quartet.


52


**Fig. S18. Reproducibility of co-immunoprecipitation assays between eTO5 and Pi-eTO5-**
**linked NLR proteins.**
Co-immunoprecipitation assays were performed in _Nicotiana benthamiana_ transiently expressing
3×FLAG-tagged SQ1, SQ2, SQ3, or SQ4 together with 4×Myc-tagged eTO5. Pikp-1–3×FLAG
and AVR-PikD–4×Myc were included as controls. FLAG-tagged NLR proteins were
immunoprecipitated with anti-FLAG agarose, and input and immunoprecipitated fractions were
analyzed by immunoblotting with anti-FLAG and anti-Myc antibodies. Ponceau S staining is
shown as a protein-loading and transfer control. (A–C) Three independent experiments performed
using the same experimental conditions. The experiments reproducibly detected association of
eTO5 with SQ2 and SQ3, whereas association with SQ1 or SQ4 was not detected under these
conditions.


53


**Fig. S18. (continued)**



54


**Fig. S18. (continued)**



55


**Data S1.**
Summary statistics for the genome assembly of _M. oryzae_ pathotype _Triticum_ isolate Br48.

**Data S2.**
One MT isolate and nine MO isolates were used for comparative genomic analyses.

**Data S3.**
A list of 35 MT-specific candidate effector genes identified in _M. oryzae_ pathotype _Triticum_ isolate
Br48.

**Data S4.**
Rice cultivars used for functional screening of MT-specific effector genes and genome sequence
analysis.

**Data S5.**
Summary statistics for the Nanopore-based _de novo_ genome assemblies of rice cultivars Nortai,
WRC43, and Sesia.

**Data S6.**
Genes within the _Pi-eTO5_ candidate interval identified by transcriptome-guided annotation of the
Nortai genome.

**Data S7.**
Genes differentially expressed between _sq3/sq3_ and _SQ3/SQ3_ seedlings.

**Data S8.**
Gene Ontology Biological Process (GO BP) over-representation analysis of genes differentially
expressed between _sq3/sq3_ and _SQ3/SQ3_ seedlings.

**Data S9.**
NB-ARC domain sequences used for phylogenetic analysis of rice NLR proteins (Fig. 3).

**Data S10.**
Conservation of SQ1-SQ4 across cultivated and wild _Oryza_ genomes, assessed by Exonerate
protein2genome alignment.

**Data S11.**
Exonerate alignment scores and haplotype classification of SQ1–SQ4 across _O. sativa_ and _O._
_rufipogon_ accessions.

**Data S12.**
Representative amino acid sequences for each SQ1–SQ4 haplotype defined in Data S11.

**Data S13.**


56


Summary statistics for the Nanopore-based _de novo_ genome assemblies of the wild _Oryza_
accessions W2149 ( _O. glumaepatula_ ) and W1577 ( _O. punctata_ ).

**Data S14.**
Metadata for 65 _M. oryzae_ pathotype _Triticum_ (MT) isolates used for SNP-based phylogenetic
analysis and copy-number estimation ( **Fig. 4A** ).

**Data S15.**
Copy-number estimates of _eTO1, eTO2, eTO5, eTO16, eTO24,_ and _eTO31_ across 65 _M. oryzae_
pathotype _Triticum_ (MT) isolates.

**Data S16.**
Per-isolate BUSCO depth statistics used for copy-number normalization and single-copy
classification thresholds.

**Data S17** .
Primer sequences in this study used.

**References**
1. C. A. Schneider, W. S. Rasband, K. W. Eliceiri, NIH Image to ImageJ: 25 years of image
analysis. Nat. Methods 9, 671–675 (2012).
2. K. Yoshida et al., Host specialization of the blast fungus _Magnaporthe oryzae_ is associated
with dynamic gain and loss of genes linked to transposable elements. BMC Genomics 17, 370
(2016).
3. C. C. Lo, P. S. G. Chain, Rapid evaluation and quality control of next generation sequencing
data with FaQCs. BMC Bioinformatics 15, 366 (2014).
4. A. V. Zimin et al., The MaSuRCA genome assembler. Bioinformatics 29, 2669–2677 (2013).
5. M. Manni et al., BUSCO update: Novel and streamlined workflows along with broader and
deeper phylogenetic coverage for scoring of eukaryotic, prokaryotic, and viral genomes. Mol.
Biol. Evol. 38, 4647–4654 (2021).
6. D. Kim, J. M. Paggi, C. Park, C. Bennett, S. L. Salzberg, Graph-based genome alignment and
genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37, 907–915 (2019).
7. M. Pertea et al., StringTie enables improved reconstruction of a transcriptome from RNA-seq
reads. Nat. Biotechnol. 33, 290–295 (2015).
8. G. Pertea, M. Pertea, GFF utilities: GffRead and GffCompare. F1000Research 9, 304 (2020).
9. B. J. Haas et al., _De novo_ transcript sequence reconstruction from RNA-seq using the Trinity
platform for reference generation and analysis. Nat. Protoc. 8, 1494–1512 (2013).
10. L. Fu, B. Niu, Z. Zhu, S. Wu, W. Li, CD-HIT: Accelerated for clustering the next-generation
sequencing data. Bioinformatics 28, 3150–3152 (2012).
11. T. N. Petersen, S. Brunak, G. von Heijne, H. Nielsen, SignalP 4.0: Discriminating signal
peptides from transmembrane regions. Nat. Methods 8, 785–786 (2011).
12. J. Sperschneider, P. N. Dodds, EffectorP 3.0: Prediction of apoplastic and cytoplasmic
effectors in fungi and oomycetes. Mol. Plant-Microbe Interact. 35, 146–156 (2022).
13. B. Langmead, S. L. Salzberg, Fast gapped-read alignment with Bowtie 2. Nat. Methods 9,
357–359 (2012).


57


14. P. Danecek et al., Twelve years of SAMtools and BCFtools. GigaScience 10, giab008 (2021).
15. B. S. Pedersen, A. R. Quinlan, Mosdepth: Quick coverage calculation for genomes and
exomes. Bioinformatics 34, 867–868 (2018).
16. K. Yoshida et al., Association genetics reveals three novel avirulence genes from the rice blast
fungal pathogen _Magnaporthe oryzae_ . Plant Cell 21, 1573–1591 (2009).
17. J. A. Sweigard, F. Chumley, A. Carroll, L. Farrall, B. Valent, A series of vectors for fungal
transformation. Fungal Genetics Reports 44, 52–53 (1997).
18. H. Takagi et al., QTL-seq: Rapid mapping of quantitative trait loci in rice by whole-genome
resequencing of DNA from two bulked populations. Plant J. 74, 174–183 (2013).
19. H. Takagi et al., Rice blast resistance gene Pii is controlled by a pair of NBS-LRR genes Pii1 and Pii-2. bioRxiv, 227132 (2017).
20. Y. Okuyama et al., A multifaceted genomics approach allows the isolation of the rice Piablast resistance gene consisting of two adjacent NBS-LRR protein genes. Plant J. 66, 467–
479 (2011).
21. H. Cheng et al., Efficient near-telomere-to-telomere assembly of nanopore simplex reads.
Nature 655, 166–173 (2026).
22. M. Alonge et al., Automated assembly scaffolding using RagTag elevates a new tomato
system for high-throughput genome editing. Genome Biol. 23, 258 (2022).
23. S. Chen, Y. Zhou, Y. Chen, J. Gu, fastp: An ultra-fast all-in-one FASTQ preprocessor.
Bioinformatics 34, i884–i890 (2018).
24. Y. Sugihara et al., High-performance pipeline for MutMap and QTL-seq. PeerJ 10, e13170
(2022).
25. S. Natsume et al., V-primer: Software for the efficient design of genome-wide InDel and SNP
markers from multi-sample variant call format (VCF) genotyping data. Breed. Sci. 73, 415–
420 (2023).
26. J. Kourelis, T. Sakai, H. Adachi, S. Kamoun, RefPlantNLR is a comprehensive collection of
experimentally validated plant disease resistance proteins from the NLR family. PLoS Biol.
19, e3001124 (2021).
27. N. Tanaka et al., Investigation of the genetic diversity of a rice core collection of Japanese
landraces using whole-genome sequencing. Plant Cell Physiol. 61, 2087–2096 (2020a).
28. N. Tanaka et al., Whole-genome sequencing of the NARO World Rice Core Collection
(WRC) as the basis for diversity and association studies. Plant Cell Physiol. 61, 922–932
(2020b).
29. M. Mikami, S. Toki, M. Endo, Comparison of CRISPR/Cas9 expression constructs for
efficient targeted mutagenesis in rice. Plant Mol. Biol. 88, 561–572 (2015).
30. Y. Naito, K. Hino, H. Bono, K. Ui-Tei, CRISPRdirect: Software for designing CRISPR/Cas
guide RNA with reduced off-target sites. Bioinformatics 31, 1120–1123 (2015).
31. Y. Liao, G. K. Smyth, W. Shi, featureCounts: An efficient general purpose program for
assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).
32. D. J. McCarthy, Y. Chen, G. K. Smyth, Differential expression analysis of multifactor RNASeq experiments with respect to biological variation. Nucleic Acids Res. 40, 4288–4297
(2012).
33. M. I. Love, W. Huber, S. Anders, Moderated estimation of fold change and dispersion for
RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
34. T. Wu et al., clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. The
Innovation 2, 100141 (2021).


58


35. H. Sakai et al., Rice Annotation Project Database (RAP-DB): An integrative and interactive
database for rice genomics. Plant Cell Physiol. 54, e6 (2013).
36. I. Ashikawa et al., Two adjacent nucleotide-binding site–leucine-rich repeat class genes are
required to confer Pikm-specific rice blast resistance. Genetics 180, 2267–2276 (2008).
37. K. Hayashi, H. Yoshida, Refunctionalization of the ancient rice blast disease resistance gene
Pit by the recruitment of a retrotransposon as a promoter. Plant J. 57, 413–425 (2009).
38. R. C. Edgar, Muscle5: High-accuracy alignment ensembles enable unbiased assessments of
sequence homology and phylogeny. Nature Commun. 13, 6968 (2022).
39. D. T. Jones, W. R. Taylor, J. M. Thornton, The rapid generation of mutation data matrices
from protein sequences. Comput. Appl. Biosci. 8, 275–282 (1992).
40. A. M. Kozlov, D. Darriba, T. Flouri, B. Morel, A. Stamatakis, RAxML-NG: A fast, scalable
and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics 35,
4453–4455 (2019).
41. I. Letunic, P. Bork, Interactive Tree of Life (iTOL) v6: Recent updates to the phylogenetic
tree display and annotation tool. Nucleic Acids Res. 52, W78–W82 (2024).
42. P. Qin et al., Pan-genome analysis of 33 genetically diverse rice accessions reveals hidden
genomic variations. Cell 184, 3542–3558.e16 (2021).
43. F. Zhang et al., Long-read sequencing of 111 rice genomes reveals significantly larger pangenomes. Genome Res. 32, 853–863 (2022).
44. D. Guo et al., A pangenome reference of wild and cultivated rice. Nature 642, 662–671 (2025).
45. G. S. C. Slater, E. Birney, Automated generation of heuristics for biological sequence
comparison. BMC Bioinformatics 6, 31 (2005).
46. M. Shimizu et al., A genetically linked pair of NLR immune receptors shows contrasting
patterns of evolution. Proc. Natl. Acad. Sci. U.S.A. 119, e2116896119 (2022).
47. G. Holley et al., Ratatosk: Hybrid error correction of long reads enables accurate variant
calling and assembly. Genome Biol. 22, 28 (2021).
48. K. Katoh, D. M. Standley, MAFFT multiple sequence alignment software version 7:
Improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).
49. L.-T. Nguyen, H. A. Schmidt, A. von Haeseler, B. Q. Minh, IQ-TREE: A fast and effective
stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32,
268–274 (2015).
50. S. Kalyaanamoorthy, B. Q. Minh, T. K. F. Wong, A. von Haeseler, L. S. Jermiin,
ModelFinder: Fast model selection for accurate phylogenetic estimates. Nat. Methods 14,
587–589 (2017).
51. M. T. Islam et al., Emergence of wheat blast in Bangladesh was caused by a South American
lineage of _Magnaporthe oryzae_ . BMC Biol. 14, 84 (2016).
52. Z. Peng et al., Effector gene reshuffling involves dispensable mini-chromosomes in the wheat
blast fungus. PLoS Genet. 15, e1008272 (2019).
53. S. Asuke, A. Horie, K. Komatsu, R. Mori, T. T. P. Vy, Y. Inoue, Y. Jiang, Y. Tatematsu, M.
Shimizu, Y. Tosa, Loss of PWT7, Located on a Supernumerary Chromosome, Is Associated
with Parasitic Specialization of _Pyricularia oryzae_ on Wheat. Mol. Plant-Microbe Interact.
36, 716–725 (2023).
54. Y. Inoue et al., Evolution of the wheat blast fungus through functional losses in a host
specificity determinant. Science 357, 80–83 (2017).
55. V. Were et al., Genome sequences of sixty _Magnaporthe oryzae_ isolates from multiple host
plant species. Zenodo (2021); https://doi.org/10.5281/zenodo.4627042.


59


56. B. Tembo et al., Whole genome shotgun sequences of _Magnaporthe oryzae_ wheat blast
isolates from Zambia. Zenodo (2021); https://doi.org/10.5281/zenodo.4637174.
57. H. Li, R. Durbin, Fast and accurate short read alignment with Burrows–Wheeler transform.
Bioinformatics 25, 1754–1760 (2009).
58. E. M. Ortiz, vcf2phylip v2.0: Convert a VCF matrix into several matrix formats for
phylogenetic analysis. Zenodo (2019).
59. B. Q. Minh et al., IQ-TREE 2: New models and efficient methods for phylogenetic inference
in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).
60. P. O. Lewis, A likelihood approach to estimating phylogeny from discrete morphological
character data. Syst. Biol. 50, 913–925 (2001).
61. D. T. Hoang, O. Chernomor, A. von Haeseler, B. Q. Minh, L. S. Vinh, UFBoot2: Improving
the ultrafast bootstrap approximation. Mol. Biol. Evol. 35, 518–522 (2018).
62. S. Guindon, J.-F. Dufayard, V. Lefort, M. Anisimova, W. Hordijk, O. Gascuel, New
algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the
performance of PhyML 3.0. Syst. Biol. 59, 307–321 (2010).
63. J. Win, E. Chanclud, C. S. Reyes-Avila, T. Langner, T. Islam, S. Kamoun, Nanopore
sequencing of genomic DNA from _Magnaporthe oryzae_ isolates from different hosts. Zenodo
(2019); https://doi.org/10.5281/zenodo.2564950.
64. C. Camacho et al., BLAST+: Architecture and applications. BMC Bioinformatics 10, 421
(2009).
65. E. Weber, C. Engler, R. Gruetzner, S. Werner, S. Marillonnet, A modular cloning system for
standardized assembly of multigene constructs. PLoS ONE 6, e16765 (2011).
66. C. Engler et al., A Golden Gate modular cloning toolbox for plants. ACS Synth. Biol. 3, 839–
843 (2014).
67. K. Honda, T. Fujiwara, H. Adachi, Blue Native-PAGE Assay of NLR Protein Complexes
Combined with Genetic Complementation in Nicotiana benthamiana. In: Plant Immunity.
Methods in Molecular Biology, vol. 3012, pp. 129–141 (2026).


60


