Functional diversification gave rise to allelic specialization in a rice

Cooperation between receptors from the NLR superfamily is important for intracellular activation of immune responses. NLRs can function in pairs that, upon pathogen recognition, trigger hypersensitive cell death and stop pathogen invasion. Natural selection drives specialization of host immune receptors towards an optimal response, whilst keeping a tight regulation of immunity in the absence of pathogens. However, the molecular basis of co-adaptation and specialization between paired NLRs remains largely unknown. Here, we describe functional specialization in alleles of the rice NLR pair Pik that confers resistance to strains of the blast fungus Magnaporthe oryzae harbouring AVR-Pik effectors. We revealed that matching pairs of allelic Pik NLRs mount effective immune responses whereas mismatched pairs lead to autoimmune phenotypes, a hallmark of hybrid necrosis in both natural and domesticated plant populations. We further showed that allelic specialization is largely underpinned by a single amino acid polymorphism that determines preferential association between matching pairs of Pik NLRs. These results provide a framework for how functionally linked immune receptors undergo co-adaptation to provide an effective and regulated immune response against pathogens. Understanding the molecular constraints that shape paired NLR evolution has implications beyond plant immunity given that hybrid necrosis can drive reproductive isolation.


Summary
Cooperation between receptors from the NLR superfamily is important for 26 intracellular activation of immune responses. NLRs can function in pairs that, upon 27 pathogen recognition, trigger hypersensitive cell death and stop pathogen invasion. 28 Natural selection drives specialization of host immune receptors towards an optimal 29 response, whilst keeping a tight regulation of immunity in the absence of pathogens. 30 However, the molecular basis of co-adaptation and specialization between paired 31 NLRs remains largely unknown. Here, we describe functional specialization in 32 alleles of the rice NLR pair Pik that confers resistance to strains of the blast fungus 33 Magnaporthe oryzae harbouring AVR-Pik effectors. We revealed that matching pairs 34 of allelic Pik NLRs mount effective immune responses whereas mismatched pairs 35 lead to autoimmune phenotypes, a hallmark of hybrid necrosis in both natural and 36 domesticated plant populations. We further showed that allelic specialization is 37 largely underpinned by a single amino acid polymorphism that determines 38 preferential association between matching pairs of Pik NLRs. These results provide a 39 framework for how functionally linked immune receptors undergo co-adaptation to 40 provide an effective and regulated immune response against pathogens. 41 Understanding the molecular constraints that shape paired NLR evolution has Introduction 45 Pathogens use an array of molecules, termed effectors, to successfully colonize hosts 46 (Win et al., 2012). Intracellular detection of effectors relies on immune receptors from 47 single amino acid polymorphism in the helper NLR Pik-2 that underpins both allelic 135 specialization and immune homeostasis. This finding allowed to reconstruct the 136 evolutionary history of this coevolution. Altogether, these results demonstrate that 137 NLR pairs can undergo co-adaptation and functional specialization, offering a 138 molecular framework to understand how they evolve to respond to pathogen 139 effectors while maintaining a tight regulation of immune responses.

141
A coevolved Pik NLR pair is required for efficient cell death response to AVR-Pik 142 effectors in N. benthamiana. 143 Two of the most studied Pik alleles, Pikp (cv. K60) and Pikm (cv. Tsuyuake), fall into  (Kiyosawa, 1978) while Pikp originated in the Indica cultivar Pusur in Pakistan 147 (Kiyosawa, 1969). Thus, we hypothesised that these alleles have been exposed to 148 differential selection pressures during domestication of elite cultivars and have 149 undergone distinct evolutionary trajectories. 150 To test for sensor/helper specificity in allelic Pik pairs, we co-expressed the sensor  159 These results indicate that Pikm-2 is required for the full Pikm mediated cell death  165 To dissect the basis of the differential cell death phenotypes displayed by Pikp-2 and 166 Pikm-2 in response to the AVR-Pik effectors, we used site-directed mutagenesis to    Pik autoactivity is linked to immune signalling. 221 We sought to gain knowledge on the constitutive cell death mediated by Pikm-2 and 222 understand the link with NLR activation. To this end, we mutated Pikm-2 in the 223 conserved P-loop and MHD motifs and tested their ability to trigger constitutive cell 224 death responses in the absence of the AVR-PikD effector.

225
The P-loop motif is conserved in NLR proteins and mediates nucleotide binding

261
Additionally, we confirmed that constitutive cell death triggered by Pik-2 262 Asp230Glu is also dependent on the P-loop and MHD motifs, confirming that this 263 mutation leads to immune activation ( Figure 5 -Figure supplement 2). 264 Interestingly, cell death responses were reduced but not completely abolished when   The Glu230 amino acid polymorphism has evolved in modern rice. 281 Having identified a determinant of Pik NLR pair specialization and compatibility as 282 a single amino acid polymorphism, we aimed to gain an evolutionary perspective of 283 the specialization process of Pik-2. For this, we combined the Pik-2 coding sequences 284 from rice cultivars described above with the Pik-2 orthologs from wild Asian and    Sensor/helper hetero-pairing alters protein accumulation in Pik NLRs. 321 We aimed to obtain mechanistic understanding of Pik NLR pair coevolution and 322 autoactivation. For this, we investigated whether accumulation of sensor Pik-1 or 323 helper Pik-2 proteins is altered in the presence of the coevolved or mismatched pair.    We co-expressed C-terminally tagged Pikp-1 or Pikm-1 with either C-terminally 344 tagged Pikp-2 or Pikm-2 in N. benthamiana. Following total protein extraction, we 345 performed co-immunoprecipitation to test for differences in NLR association ( Figure   346 7B). Pikp-1 and Pikm-1 were also co-infiltrated with the rice NLR Pia-2 (the sensor 347 NLR, also known as RGA5, of the immune receptor pair Pia) as a negative control.

348
Both Pikp-2 and Pikm-2 could be detected after immunoprecipitation of either Pikp-  Sensor and helper Pik NLRs preferentially associate with their coevolved pair. 373 To gain a deeper knowledge of Pik pair association, we investigated whether allelic 374 Pik NLRs display any preference in association to their coevolved NLR pair. As both 375 autoactive and non-autoactive pairs associate, we designed an NLR competition 376 assay with a cell death readout to test for preferential association between allelic  383 To test this, we transiently co-expressed both Pikp-1 and Pikm-2 NLRs in N. 384 benthamiana using a fixed concentration (OD 600 0.4) of Agrobacterium tumefaciens to 385 deliver each construct. We also co-delivered increasing concentrations of Pikp-2 386 (spanning an OD 600 of 0-0.6) and scored the cell death phenotype (Figure 9). 387 Interestingly, Pikp-2 acted as a suppressor of autoimmune phenotypes triggered by preference to signal through coevolved Pikp-2 rather than Pikm-2. 392 We also replicated this experiment co-infiltrating a fixed concentration of Pikp-1 and

415
Altogether, these data reveal that coevolved Pik NLRs display preference in 416 association over non-coevolved NLRs. This represents another example of NLR pair 417 co-adaptation. These differences may underpin the observed cell death phenotypes 418 in response to effectors and in autoimmunity.

419
Pik helper/sensor association preference is underpinned by Pik-2 polymorphism. 420 To shed light on the basis of the preferential binding between Pikp-1 and Pikp-2, we   432 To investigate if the preferential sensor/helper association is related to the activation 433 of the helper NLR Pik-2, we tested whether the constitutive cell death mediated by 434 Pikp-2 Asp230Glu could be supressed by mutants that render Pikp-2 inactive.  The work presented here highlights sensor/helper coevolution in an allelic rice NLR 449 pair and the basis of their functional diversification towards differential effector 450 recognition specificities (Figure 12). We discovered that a single amino acid 451 polymorphism underpins specialization of the helper Pik-2 NLR to its corresponding  We found mismatched allelic NLR pairs can lead to constitutive cell death. We 504 further narrowed down this phenotype to a single Asp to Glu polymorphism, which 505 is the same polymorphism that underpins an extended cell death response to AVR-506 Pik effectors. Introducing this Asp230Glu polymorphism in Pikp-2 led to an increase 507 of cell death in response to AVR-Pik effectors as well as to autoimmune phenotypes.

508
As these amino acids have very similar properties it is intriguing how a fairly minor 509 difference can underpin such a major phenotype. The mechanistic basis of this 510 autoactivation phenotype remains obscure, but it is possible that the larger amino   Codon-based alignment was generated using MUSCLE 3.8.425 (Edgar, 2004). The 580 alignment positions with more than 40% data missing were removed using 581 QKphylogeny (https://github.com/matthewmoscou/QKphylogeny). The 582 maximum likelihood tree was calculated from a 3,066-nt-long alignment using 1000 583 bootstrap method (Felsenstein, 1985) and GTRGAMMA substitution model (Tavaré,584 1986) as implemented in RAxML v8.2.11 (Stamatakis, 2014). Best-scoring tree was          (Felsenstein, 1985) and GTRGAMMA substitution model (Tavaré, 1986). Best-941 scoring ML tree was manually rooted using the Pik-2 sequence from Leersia perreri as