A built-in brake for plant immunity: Resistance (R) protein–mediated immunity is fine-tuned by R gene analogs Authors: Jiahui Nie, Lang Liu, Yetong Qi, Chengyu Zhu, Zeng Liu, Xinya Wu, Jing Zhou, Jie Wu, Meng Xu, Hongjun Li, Xinyuan Sun, Mingshuo Fan, Ruimin Yu, Huishan Qiu, Qingguo Sun, Shengxuan Liu, Paul R. J. Birch, Zhendong Tian Published: 2026 ABSTRACT Potato late blight caused by Phytophthora infestans threatens global food security. Unlike the fact that most race-specific resistance to P. infestans (Rpi) genes is rapidly overcome by evolving pathogen populations, the R8 gene from Solanum demissum has conferred durable, quantitative resistance for nearly a century despite encoding a canonical coiled-coil–type nucleotide-binding, leucine-rich repeat (NLR) protein. We identified specific R8 gene analogs that act as negative modulators of R8-mediated immunity. These R8 gene analogs lack independent resistance functions but perturb the recognition of avirulence effector Avr8 by R8 and interfere with oligomerization and plasma membrane association of NRC2, a helper NLR essential for the R8 signaling pathway. Stable overexpression of such RGAs effectively compromises R8-mediated resistance in both potato and Nicotiana benthamiana. We propose that this endogenous regulation balances the intensity of R8-mediated immunity, likely reducing selection pressure on the pathogen population and prolonging resistance durability. Our findings reveal a regulatory layer where genetically linked RGAs control resistance (R) protein–mediated immunity, providing a conceptual framework for durable resistance breeding. INTRODUCTION Plants have evolved robust defense mechanisms to combat pathogen infection. Pathogen-associated molecular pattern–triggered immunity is activated by cell surface–localized pattern-recognition receptors upon the recognition of conserved pathogen-associated molecular patterns. In contrast, effector-triggered immunity is mediated by intracellular nucleotide-binding, leucine-rich repeat (NLR) receptors that detect specific pathogen effectors to initiate a stronger and more sustained immune response (1, 2). NLRs share a conserved architecture comprising a central nucleotide-binding (NB-ARC) domain and C-terminal leucine-rich repeats (LRRs), typically associated with variable N-terminal domains. On the basis of their N-terminal domains and the evolutionary history of the NB-ARC domain, NLRs are classified into three major groups: TIR (Toll/interleukin-1 receptor)–type NLRs, coiled-coil (CC)–type NLRs, and RPW8 CC domain–containing NLRs (3). In addition, some NLRs harbor noncanonical integrated domains that mimic pathogen targets and are activated upon interaction with the effectors (4, 5). Normally, NLRs are divided into two functional groups: direct sensor NLRs and indirect sensor NLRs (6). Direct sensor NLRs function as singleton receptors, capable of both effector recognition and immune activation (7, 8). In contrast, indirect sensor NLRs work in a pair consisting of a “sensor” and a “helper” NLR that are genetically linked (7). Most plant disease-resistance (R) genes encode NLR proteins that monitor pathogen infection. Meanwhile, qualitative resistance mediated by race-specific R genes often exerts intense selection pressure, driving the emergence of virulent variants that overcome such R genes (9, 10). This “boom-and-bust” breeding pattern, observed across major crop diseases, including wheat rust, rice blast, and potato late blight, fundamentally limits sustainable agricultural production (11). In contrast, quantitative resistance partially suppresses pathogen infection, slowing disease progression and imposing minimal selection pressure, which favors long-term durability (10). Late blight, caused by the oomycete Phytophthora infestans, is a notorious disease that caused the Irish potato famine in the 1840s, resulting in an estimated 1 million deaths (12). It remains the most serious biotic constraint, leading to severe losses in potato yield and quality globally. Great efforts have focused on introgression of Rpi genes (resistance genes against P. infestans) from wild relatives into cultivated potato for more than 180 years (13). Although more than 70 Rpi genes have been identified, the rapidly evolving P. infestans populations have overcome most of them, particularly those from the wild species Solanum demissum such as R1, R2, R3a, and R3b (13, 14). Fortunately, some Rpi genes, including RB/Rpi-blb1, Rpiblb2, and Rpivnt1.1, exhibit a broader resistance spectrum (15–18). Nonetheless, some emerging virulent lineages of P. infestans can defeat them, but they still function across the global potato production area (13). Among the S. demissum–derived genes, R8 is a notable exception. R8 encodes a CC-NLR protein that recognizes the avirulence protein Avr8 from P. infestans and has maintained its efficacy in the field for nearly a century (19). R8 is found in durably resistant varieties from Europe, US, and China (19, 20). Despite being a typical NLR, unlike other race-specific R genes, R8 provides durable resistance with quantitative characteristics in the field (21, 22). In our breeding program, we observed that resistance levels varied substantially among F1 progeny carrying R8, ranging from highly resistant to fully susceptible. This phenomenon was also observed across multiple breeding populations derived from parental lines with distinct genetic backgrounds (19–21). Potato R8-mediated resistance exhibits quantitative and durable characteristics. The variable resistance levels cannot be adequately explained by a binary “presence-or-absence” model; rather, it is likely regulated by complex, yet-to-be-characterized intragenic mechanisms. Plant NLR receptor genes are typically organized in genomic clusters, normally with several R gene analogs (RGAs), which facilitate rapid R gene evolution by sequence exchanges via recombinatorial mispairing. This process may generate high haplotypic diversity, novel R gene sequences that may encode altered specificities or have altered expression patterns, or generate pseudogenized genes (23, 24). In this study, we demonstrate that certain R8 gene homolog analogs (RGAs), likely generated by random unequal crossing over, nucleotide mutation, etc., within R8 loci from the different haplotype on the different chromosomes, modulate R8 function by interfering with its association with Avr8, thereby fine-tuning R8 activity. We propose that this endogenous regulation balances the intensity of R8-mediated resistance, which is presumed to exert less selection pressure on the pathogens and thus prolong the duration of its resistance. This study highlights that R8 and its RGAs coordinately regulate late blight resistance. Our findings reveal a regulatory layer controlling R-mediated resistance and provide a conceptual framework for deploying resistance genes with extended durability in plant breeding. RESULTS Potato F1 progenies harboring the R8 gene exhibit differential resistance to P. infestans The potato late blight resistance gene R8, which confers broad-spectrum and durable resistance, has been extensively used in breeding programs (19, 21, 22). To use R8 in our breeding efforts, several crosses were generated using one or both parents containing the R8 gene (table S1). Initially, R8-specific molecular markers were used to screen F1 progeny (fig. S1), and those clones displaying the positive R8-marker signal underwent late blight resistance evaluation via detached leaf inoculation assays. The inoculation results revealed that R8 marker–positive clones from three distinct crosses exhibited varied resistance levels, ranging from highly resistant to fully susceptible (fig. S2 and table S2). These findings demonstrate that, even in those clones potentially containing R8, resistance levels are highly diverse, reflecting the quantitative nature of R8-mediated resistance. Such phenotypic variation was also evident in field conditions (19, 21). Diverse RGAs of R8 exhibit distinct functional profiles The observation that certain R8 marker–positive clones showed only moderate resistance or even susceptibility prompted us to investigate whether these clones contained the functional R8 gene. Three representative progenies (10-20, highly resistant; 10-58, resistant; 10-65, susceptible) from the same cross and one control breeding line JX-1 (containing R8, highly resistant) (fig. S2C) were selected for R8 coding sequence (CDS) amplification from cDNA. The resulting polymerase chain reaction (PCR) products were cloned into a plant expression vector and transformed into Escherichia coli. From each representative potato progeny, 20 positive E. coli clones were randomly selected, and 35 clones containing inserts of the expected length (∼3738 base pairs) were subjected to full-length sequencing. Sequence alignment identified multiple identical clones, which were subsequently categorized into 11 distinct groups across the four potato lines (table S3 and fig. S3). All sequence information is provided in the Supplementary Materials (data S8). In each material, identical R8 CDS sequences were identified. Two sequences from the potato progeny 10-20 each contain one nucleotide synonymous mutation, and they encode identical R8 proteins. Other sequences encoded diverse putative proteins that are different from R8, while some harbored premature stop codons, resulting in truncated proteins (table S3). Sequences encoding putative proteins different from R8 are hereafter referred to as RGAs. Traditionally, most RGAs are dismissed as nonfunctional pseudogenes or silent genomic elements with negligible contribution to plant immunity (25, 26). To test whether those RGAs recognize the avirulence protein Avr8, representative RGAs from each potato line were selected for functional characterization via co-infiltration with Avr8 in Nicotiana benthamiana. Agrobacterium clones harboring the authentic R8 CDS (including 10-20-21 and 10-20-26) consistently induced a robust hypersensitive response (HR). In contrast, other RGAs elicited either a weakened HR (e.g., JX1-16 and JX1-17) or no visible symptoms (e.g., 10-58-8 and 10-65-1) (Fig. 1A). Diverse RGAs of R8 exhibit distinct functions. (A) Recognition of Avr8 by R8 and RGAs. RGAs and the functional R8 were transiently coexpressed with Avr8 in wild-type (WT) N. benthamiana leaves. HR phenotypes were captured 5 days post-agroinfiltration under UV light. 1: EV-GFP + Avr8 (negative control); 2: clone R8 + Avr8; 3 to 5: clone RGAs + Avr8. Scale bars represent 1 cm. White circles indicate the agroinfiltration sites. (B and C) Modulation of R8-mediated resistance by specific RGAs. Left panel images: RGA 10-65-1 and RGA 10-58-11 exhibited no resistance function against P. infestans. GFP-R8 provided effective protection. EV-GFP, negative control. Right panel images: Coexpression of RGA 10-58-11 with R8 did not interfere with R8-mediated disease resistance. However, RGA 10-65-1 attenuated the R8 resistance function, particularly with an Agrobacterium culture at an OD600 ratio of 1:3 (R8:RGA 10-65-1). Leaves were challenged with P. infestans 24 hours post-agroinfiltration. Data are presented as box plots; different letters above box plots indicate statistically significant differences determined by Tukey’s multiple range test (one-way ANOVA; P < 0.001; three biological replicates). (D) RGA 10-65-1 suppresses R8/Avr8–triggered HR. Left image: HR phenotypes resulting from the coexpression of R8/Avr8 or Pto/AvrPto with GFP-10-65-1, 10-58-11, or R9a. The right box plot shows the HR index (one-way ANOVA; P < 0.0001; ns indicates not significant; n = 32; three biological replicates). (E) Western blot analysis confirming that the coexpression of GFP-10-65-1, GFP-10-58-11, or R9a-GFP did not alter the protein accumulation levels of R8 or Avr8. PS, Ponceau S. In box plots of (B) and (C), center lines show the medians, box limits indicate the 25th and 75th percentiles, whiskers extend to 1.5 times the interquartile range from the 25th and 75th percentiles, the top lines represent maxima, and the bottom lines represent minima. Further P. infestans inoculation assays demonstrated that these RGAs did not confer resistance (Fig. 1, B and C, and fig. S4, A to F). We then examined whether these RGAs interfere with the R8 function. While most RGAs (such as 10-58-11) had no inhibitory effect, specific RGAs, notably 10-65-1, significantly compromised R8-mediated resistance when co-infiltrated with Agrobacterium culture at an optical density at 600 nm (OD600) ratio of 1:3 (R8:RGA) (Fig. 1C). Similarly, RGAs 10-65-2 and 10-58-2 perturbed the R8 function in a dose-dependent manner (fig. S4, A and B). Western blot analysis confirmed that R8 protein accumulation was not affected by the coexpression of these RGAs, and all proteins were stably expressed (fig. S5). To further characterize this interference, we investigated the effect of RGA 10-65-1 on R8/Avr8–triggered HR. RGA 10-58-11, RGA 10-65-1, and R9a (another late blight resistance gene) were transiently coexpressed with either R8/Avr8 or the control pair Pto/AvrPto. Neither R9a, RGA 10-65-1, nor RGA 10-58-11 affected Pto/AvrPto–mediated HR. Similarly, R9a and RGA 10-58-11 did not interfere with R8/Avr8–triggered HR. However, RGA 10-65-1 (along with 10-65-2 and 10-58-2) specifically suppresses R8/Avr8–triggered HR (Fig. 1D and fig. S6A). Western blots confirmed that all the target proteins were stably expressed (Fig. 1E and fig. S6B). Together, these findings revealed that RGAs of R8 exhibit distinct function (table S4). Some RGAs certainly disturb R8/Avr8 recognition and R8-mediated resistance in N. benthamiana. Stable expression of RGA 10-65-1 attenuates R8-mediated resistant function To verify whether specific RGAs could compromise the R8 function, plant overexpression (OE) constructs targeting RGA 10-65-1 (a suppressive analog) and RGA 10-58-11 (a nonsuppressive analog) were generated under the control of the 35S promoter. Initially, the 35S::RFP-10-65-1 construct was transformed into the N. benthamiana “R8-23” transgenic line, which stably expressed R8 and exhibited strong resistance to the P. infestans isolate 88069 (fig. S7). Three independent OE-RFP-10-65-1 lines with varying expression levels were selected for inoculation (Fig. 2A). Compared with the control “R8-23” line, all three RGA 10-65-1–overexpressing lines showed significantly enhanced pathogen colonization. Notably, disease lesion size increased progressively in parallel with the increasing expression level of RGA 10-65-1 (Fig. 2, B and C). R8-mediated resistance is suppressed in transgenic potato and N. benthamiana plants overexpressing RGA 10-65-1. (A) Relative expression levels of RGA 10-65-1 in OE-10-65-1 transgenic N. benthamiana lines. Expression was determined using RFP tag–specific primers. (B) Representative leaf images showing disease lesions on “R8-23” N. benthamiana lines overexpressing RGA 10-65-1 following P. infestans inoculation. Leaf images were captured under UV light. Scale bars, 1 cm. (C) Quantitative analysis of lesion areas showing a significant increase in OE-10-65-1 lines compared to the “R8-23” control. Data are presented as box plots (one-way ANOVA; ****P < 0.0001; three biological replicates). (D and G) Representative leaf phenotypes showing disease lesions on OE-RGA 10-65-1/10-58-11 transgenic potato lines following inoculation with the P. infestans isolate HB0914-2. (E and H) Significant lesion expansion in RGA 10-65-1 OE lines, but not in RGA 10-58-11 OE lines, relative to the “98100” (WT) control. Data are presented as box plots (one-way ANOVA; *P < 0.05 and ****P < 0.0001; three biological replicates). (F) Bar graph showing negligible sporangia on the leaves of RGA 10-58-11 OE plants. (I) Sporangia quantification showing significantly higher sporangial density in RGA 10-65-1 OE plants (one-way ANOVA; *P < 0.05 and ****P < 0.0001; three biological replicates, each containing three leaves). Data are presented as the means ± SEM. White circles indicate lesion areas. Significant differences were determined by Tukey’s multiple-range test. In box plots of (C), (E), and (H), center lines show the medians, box limits indicate the 25th and 75th percentiles, whiskers extend to 1.5 times the interquartile range from the 25th and 75th percentiles, the top lines represent maxima, and the bottom lines represent minima. To further validate the above observation, the constructs 35S::RFP-10-65-1 and 35S::RFP-10-58-11 were transformed into a haploid potato line “98100,” which carries the R8 gene and displays a highly resistant phenotype. Stable OE-RFP-10-65-1 and OE-RFP-10-58-11 potato lines were obtained (fig. S8, A and B). Transcript levels were verified by quantitative real-time PCR (qRT-PCR) (fig. S8, C and D). Following inoculation with P. infestans isolates HB0914-2 and UK3928A, the resistance level of the RGA 10-58-11 OE lines remained comparable to the wild-type “98100” (Fig. 2, D to F, and fig. S8, E to G). In contrast, RGA 10-65-1 OE lines exhibited significantly larger lesions and higher spore counts, with the degree of resistance loss positively correlating with RGA 10-65-1 expression levels (Fig. 2, G to I, and fig. S8, H to J). Collectively, these results demonstrate that stable expression of suppressive RGA 10-65-1 effectively antagonizes the R8 function in both potato and N. benthamiana. R8/Avr8 recognition activates NRC2 oligomerization and plasma membrane translocation Upon recognizing the P. infestans effector Avr8, R8 induces HR, a form of programmed cell death. To characterize the R8/Avr8 interaction while avoiding confounding cell death, we transiently expressed these components in nrc2/3/4 knockout N. benthamiana (27). A coimmunoprecipitation (co-IP) assay demonstrated that HA (hemagglutinin)-R8 coimmunoprecipitated with RFP (red fluorescent protein)-Avr8, but not with control RFP-Avr2 (another P. infestans avirulence protein), or EV (empty vector)-RFP. These results confirms the interaction between R8 and Avr8 (Fig. 3A and fig. S9). The direct interaction was further confirmed by luciferase (LUC) complementation assays, where coexpression of cLuc-R8 and Avr8-nLuc produced a strong luminescence signal, similar to the positive control (NPR3-nLuc + cLuc-NPR4), whereas control combinations yielded no signal (Fig. 3B and fig. S10, A and B). R8 interacts with Avr8 in the cytoplasm. (A) Co-IP assay showing that HA-R8 interacts with RFP-Avr8 in planta. The presence (+) or absence (−) of each construct in nrc2/3/4 knockout N. benthamiana leaves is indicated. IB: HA and IB: RFP indicate the antibodies used. Protein size markers are shown in kDa. PS staining shows equal protein loading. Red asterisks indicate target proteins. (B) LUC complementation assay confirming the interaction between R8 and Avr8 in nrc2/3/4 knockout N. benthamiana leaves. The LUC signal was detected only when Avr8-nLuc and cLuc-R8 were coexpressed. NPR3-nLuc + cLuc-NPR4 served as a positive control. Red circles indicate agroinfiltration sites. (C) Projection of a confocal Z-series image showing that GFP-R8, coexpressed with a nucleocytoplasmic RFP, localized to the cytoplasm and nucleus. Images were collected 48 hours after agroinfiltration. (D) Confocal projection showing that GFP-Avr8 localizes to the cytoplasm and nucleus. (E) BiFC assay showing the interaction between YN-R8 and YC-Avr8 in the cytoplasm. The reconstituted YFP signal in the confocal projection is shown in green. Nucleocytoplasmic RFP derived from the EV in (C) to (E) is shown in magenta. Inset images are single optical sections through the nuclei to show the degree of nucleocytoplasmic labeling. Scale bars indicate 15 μm. The plots on the right of confocal images show the fluorescence intensity profiles along the white arrows in (C) to (E). We subsequently determined the subcellular localization of the R8/Avr8 interaction. GFP (green fluorescent protein)-R8 was located in the cytoplasm and nucleus with the signal overlapping with EV-RFP (Fig. 3C), not with the plasma membrane (PM) marker RFP-AtREM1.3 (fig. S10C). Similarly, RFP-Avr8 was also localized to both the nucleus and cytoplasm (Fig. 3D). Bimolecular fluorescence complementation (BiFC) revealed that their interaction occurs specifically within the cytoplasm, as evidenced by the overlap of the yellow fluorescent protein (YFP) signal with the cytoplasmic marker EV-RFP, rather than the PM marker RFP-AtREM1.3 (Fig. 3E and fig. S10D). YN-R8 and YC-Avr8 were normally expressed (fig. S10E). Collectively, these findings illustrate that R8 and Avr8 directly interact in the cytoplasm. R8/Avr8–mediated HR and immunity require the helper NLRs NRC2, NRC3, and NRC4 (fig. S11) (28). Given that NLR recognition often triggers helper oligomerization, we used NRC2 as a representative indicator, due to its functional redundancy with the other helper NLRs NRC3 and NRC4 (29–31). To prevent HR-associated cell death, the inactive MADA-motif mutant NRC2EEE was used (30). Native-polyacrylamide gel electrophoresis (PAGE) analysis in nrc2/3/4 knockout N. benthamiana revealed that R8/Avr8 recognition induced a pronounced electrophoretic mobility shift, with the ∼200-kDa NRC2EEE monomer assembling into an ∼900-kDa complex (Fig. 4A and fig. S12, A and B). This transition occurred without altering the total protein stability or the primary molecular weight of NRC2EEE (Fig. 4B and fig. S12, A and B), demonstrating that R8/Avr8 recognition triggers the assembly of the downstream helper NRC2 into a high-molecular-weight complex. R8/Avr8 recognition induces NRC2 oligomerization and PM translocation. (A) BN-PAGE analysis showing NRC2EEE-Myc oligomerization upon R8/Avr8 recognition. The presence (+) or absence (−) of each construct in nrc2/3/4 knockout N. benthamiana leaves is indicated. IB: Myc indicates the antibody used. kDa indicates the protein molecular weight. Three replicates show similar results [two replicates are shown in fig. S12 (A and B)]. (B) Samples from (A) were denatured and analyzed by SDS-PAGE. IB: HA, IB: RFP, and IB: Myc indicate antibodies used. PS staining shows equal protein loading. Red asterisks indicate target proteins. (C) Confocal images showing the localization of inactive and activated NRC2. Unfused RFP derived from the EV and RFP-AtREM1.3 (magenta) serve as markers for the nucleocytoplasm and PM, respectively. Upper panel: Projection of a Z-series showing that, in the absence of Avr8, NRC2EEE-GFP localized to the cytoplasm. The plot on the right shows the fluorescence intensity along the path of the white arrow. Lower panel image: In the presence of Avr8, a single optical section shows that NRC2EEE forms puncta associated with PM. The area within the yellow rectangle is enlarged on the right to highlight the puncta at the PM. Scale bars, 15 μm. (D) Subcellular fraction enrichment assay showing that activated NRC2EEE-Myc accumulated in the M fraction. Inactivated NRC2EEE-Myc distributed evenly in S and M fractions, whereas activated NRC2EEE-Myc is predominantly present in the M fraction, and R8 was mainly present in the S fraction. T, total fraction; S, soluble fraction; M, membrane fraction. H+-ATPase serves as a PM marker. Red triangles indicate NRC2EEE-Myc protein bands. Rubisco, visualized by PS staining, serves as a loading control for total and soluble fractions. Two replicates show similar results (one replicate is shown in fig. S12C). Furthermore, confocal microscopy showed that, while inactive NRC2EEE-GFP was diffusely localized in the cytoplasm, its activation by R8/Avr8 induced redistribution into punctate structures associated with the PM (Fig. 4C). This translocation was confirmed by membrane protein fractionation assays, which showed that activated NRC2EEE accumulated predominantly in the membrane (M) fraction (Fig. 4D and fig. S12C). Notably, the sensor R8 remained exclusively in the soluble (S) fraction (Fig. 4D and fig. S12C), suggesting that it does not physically incorporate into the activated NRC2 resistosome. RGA 10-65-1 associates with Avr8 but fails to activate NRC2 oligomerization The RGA 10-65-1, which lacks an independent resistance function (Fig. 1, A and C), encodes a 1054–amino acid protein. It features a Solanaceae-specific domain, a CC domain, an NB-ARC domain, and a truncated C-terminal LRR region, sharing 97.58% nucleotide sequence identity and 94.59% amino acid sequence identity with R8 (fig. S13 and tables S5 and S6). To test whether this RGA interacts with Avr8, we performed co-IP assays in wild-type N. benthamiana. These demonstrated that GFP-10-65-1 coimmunoprecipitated with RFP-Avr8 but not with the control RFP-Avr2 or EV-RFP (Fig. 5A and fig. S14A). This interaction was further supported by a LUC complementation assay, in which luminescence was detected only when cLuc-10-65-1 and Avr8-nLuc were coexpressed (Fig. 5B). cLuc-10-65-1 and Avr8-nLuc were normally expressed (fig. S14B). Furthermore, BiFC assays revealed that, similar to the localization of R8/Avr8, RGA 10-65-1 interacts with Avr8 in the cytoplasm (Fig. 5C and fig. S14C). RGA 10-65-1 associates with Avr8 but fails to trigger NRC2 oligomerization. (A) Co-IP assay showing that GFP-10-65-1 interacts with RFP-Avr8 in planta. The presence (+) or absence (−) of each construct is indicated. IB: GFP and IB: RFP indicate the antibodies used. kDa indicates protein size. PS staining shows equal protein loading. Red asterisks indicate target proteins. Three replicates show similar results (two replicates are shown in fig. S14A). (B) LUC complementation assay confirming the interaction between 10-65-1 and Avr8 in nrc2/3/4 knockout N. benthamiana plants. The LUC signal was detected in leaf areas where Avr8-nLuc and cLuc-10-65-1 were coexpressed. NPR3-nLuc + cLuc-NPR4 serves as a positive control. Red circles represent the agroinfiltration sites. (C) BiFC assay revealing the cytoplasmic interaction between YN-10-65-1 and YC-Avr8. EV-RFP was used as a marker to confirm the cytoplasmic localization of the reconstituted YFP signal. Images were merged confocal slices. YFP and RFP signals are shown in green and magenta, respectively. Scale bars, 15 μm. (D) NRC2EEE-Myc failed to form the oligomer upon the coexpression of Avr8 and RGA 10-65-1. SDS-PAGE (right panel) shows the stable expression of GFP-10-65-1, NRC2EEE-Myc, and RFP-Avr8. The presence (+) or absence (−) of each construct is indicated. IB: GFP, IB: RFP, and IB: Myc indicate the antibodies used. kDa indicates protein size. PS staining showing equal protein loading. Two replicates show similar results (one replicate is shown in fig. S14D). However, unlike R8/Avr8 recognition, coexpressing GFP-10-65-1 and RFP-Avr8 failed to trigger NRC2 oligomerization, with NRC2EEE-Myc remaining in its monomeric form (Fig. 5D and fig. S14D). Similarly, the nonsuppressive analogs RGA 10-58-11 and RGA JX1-6 also associated with Avr8 but failed to induce NRC2 assembly (fig. S15). These findings indicate that, while RGA 10-65-1 can bind the effector Avr8, this interaction is nonproductive and fails to initiate downstream immune signaling. RGA 10-65-1 attenuates the R8/Avr8 association The RGA 10-65-1 interacts with Avr8 (Fig. 5) and compromises R8-mediated resistance (Figs. 1C and 2). We therefore sought to determine the molecular basis of this suppression. Given that RGA 10-65-1 interacts with Avr8 in the cytoplasm, we hypothesized that it might compete with R8 for Avr8 binding, thereby blocking immune signaling. In BiFC assays, RGA 10-65-1 substantially attenuated the YFP signal generated by YN-R8/YC-Avr8 complementation in the cytoplasm, whereas the control HA-GUS and the nonsuppressive RGA 10-58-11 had no such effect (Fig. 6A and fig. S16). Similarly, a LUC complementation assay revealed that RGA 10-65-1 strongly diminished the luminescence intensity produced by cLuc-R8 and Avr8-nLuc (Fig. 6B and fig. S17, A and B). The expression of GFP-10-65-1, EV-GFP, and GFP-10-58-11 did not alter the protein levels of the sensor or effector (fig. S17C). Furthermore, co-IP assays confirmed that RGA 10-65-1 perturbs R8/Avr8 association, as evidenced by the weakened GFP-Avr8 band coprecipitated by HA-R8 in the presence of RGA 10-65-1 compared to the RGA 10-58-11 control (Fig. 6C and fig. S18). Together, these results demonstrate that RGA 10-65-1 interferes with the formation of the R8/Avr8 complex through competitive binding. RGA 10-65-1 disturbs the R8/Avr8 association. (A) Representative confocal images showing that HA-10-65-1 reduces the YFP signal intensity generated by YN-R8 and YC-Avr8 complementation, whereas HA-10-58-11 and the HA-GUS control do not. Constructs were coexpressed in nrc2/3/4 knockout N. benthamiana. EV-RFP serves as a cytoplasmic marker. HA-GUS and HA-10-58-11 served as negative controls. The YFP signal is shown in green. The RFP signal is shown in magenta. Scale bars, 15 μm. Fluorescence intensity plots across the indicated paths are shown on the right. Five biological replicate experiments showed similar results (shown in fig. S16). (B) Representative leaf image showing that GFP-10-65-1 reduces the LUC signal from cLuc-R8 and Avr8-nLuc, whereas controls EV-GFP and GFP-10-58-11 do not. Bar graphs represent relative LUC activity. Significant differences were determined by Tukey’s multiple-range test and are indicated with different letters (one-way ANOVA; P < 0.001; n = 13). Relevant constructs were expressed in leaves of nrc2/3/4 knockout N. benthamiana plants. Red circles represent the agroinfiltration sites. Western blotting shows that GFP-10-65-1, EV-GFP, and GFP-10-58-11 did not affect protein expression levels of cLuc-R8 and Avr8-nLuc (shown in fig. S17C). (C) Co-IP showing that less GFP-Avr8 was immunoprecipitated by HA-R8 in the presence of RFP-10-65-1. This interference is dose-dependent, as shown by increasing OD600 ratios (1× and 2×) of the RGA construct. The presence (+) or absence (−) of each construct is indicated. IB: HA and IB: GFP indicate antibodies used. kDa indicates protein molecular weight. Actin was used as an internal control. Red asterisks indicate target protein bands. Three replicates showed similar results (two replicates are shown in fig. S18). RGA 10-65-1 attenuates R8/Avr8–activated NRC2 oligomerization The observation that RGA 10-65-1 disrupts the R8/Avr8 interaction suggested that it might also impair downstream NRC2 oligomerization. To test this, NRC2EEE-3×Myc was coexpressed with HA-R8/RFP-Avr8 alongside either R9a-GFP, GFP-10-58-11, or GFP-10-65-1 in nrc2/3/4 knockout N. benthamiana. Native-PAGE and immunoblot analysis revealed a significant attenuation of NRC2 oligomerization in the presence of GFP-10-65-1, whereas GFP-10-58-11 had no effect. This effect was not due to reduced protein levels, as SDS-PAGE confirmed stable accumulation of all proteins, including NRC2EEE-Myc (Fig. 7, A and B, and fig. S19). RGA 10-65-1 suppresses R8/Avr8–activated NRC2 oligomerization. (A) GFP-10-65-1 attenuates R8/Avr8–activated NRC2EEE-Myc oligomerization. GFP-10-58-11 served as a negative RGA control, and R9a-GFP served as an R protein control. Protein lysates from nrc2/3/4 knockout N. benthamiana leaves were analyzed by BN-PAGE. An aliquot of the native protein samples was denatured and analyzed by SDS-PAGE to show the protein expression of HA-R8, NRC2EEE-Myc, and RFP-Avr8. GFP-10-65-1 did not affect NRC2EEE-Myc accumulation in planta. The presence (+) or absence (−) of each construct is indicated. IB: GFP, IB: RFP, IB: Myc, and IB: HA indicate the antibodies used. kDa indicates the protein molecular weight. Actin was used as an internal control. Three replicates showed similar results (two replicates are shown in fig. S19). (B) Relative quantification of NRC2EEE-Myc oligomer levels across three biological replicates. To quantify oligomer levels, BN-PAGE signal intensities were normalized to actin bands using ImageJ. Different letters indicate statistically significant differences determined by Tukey’s multiple-range test (one-way ANOVA; P < 0.05; three biological replicates). (C) Subcellular protein fraction enrichment assay demonstrating that RGA 10-65-1 interferes with the membrane translocation of activated NRC2EEE-Myc. In the presence of R9a-GFP or RGA 10-58-11, activated NRC2EEE-Myc predominantly partitioned into the M fraction. In contrast, the membrane-associated NRC2EEE was markedly reduced in the presence of GFP-10-65-1. All constructs were transiently expressed in N. benthamiana leaves by agroinfiltration. T, total fraction; S, soluble fraction; M, membrane fraction. H+-ATPase and Rubisco are used as markers for the membrane and total/soluble fractions, respectively. IB: Myc and IB: HA indicate the antibodies used. kDa indicates the protein molecular weight. Three replicates showed similar results (two replicates are shown in fig. S20). Subcellular fractionation assays further supported these findings. Activated NRC2EEE was predominantly enriched in the M fraction in the presence of the R9a-GFP control. However, its membrane accumulation was markedly reduced in the M fraction upon coexpression with GFP-10-65-1. In contrast, GFP-10-58-11 did not interfere with the PM translocation of NRC2 (Fig. 7C and fig. S20). Collectively, these findings reveal that, as RGA 10-65-1 interferes with R8/Avr8 association, less R8 becomes activated, which consequently results in reduced NRC2 activation. This attenuates both the oligomerization and PM translocation of NRC2 activated by R8/Avr8 recognition, thereby compromising R8-mediated immune signal transduction (Fig. 8). Proposed model for the modulation of R8-mediated immunity by suppressive RGAs. During the immune response of S. tuberosum against P. infestans, R8-mediated defense is dynamically regulated by specific suppressive RGAs. In the absence of this suppressor, the pathogen secretes the effector Avr8 into the host cytoplasm, where it is directly perceived by the NLR sensor R8. This recognition triggers the downstream helper NLR NRC2 to oligomerize, translocate to the PM, and assemble into active resistosome complexes. This process orchestrates a robust HR and systemic immunity. In the presence of a suppressor such as RGA 10-65-1, the RGA intercepts Avr8. This competitive interception attenuates the signaling flux from R8 to NRC2, substantially weakening NRC2 activation and its subsequent recruitment to the membrane, thereby suppressing HR-mediated cell death and leading to reduced pathogen resistance. DISCUSSION Late blight has remained a persistent threat to global potato production for more than 180 years, dating back to the Irish potato famine of the 1840s. Because of the rapid evolution of P. infestans populations, pathogens frequently overcome Rpi genes by evading R protein recognition. R8, a race-specific Rpi gene derived from the wild species S. demissum, is an intriguing exception. Despite recognizing only the effector Avr8, it has maintained its efficacy for nearly a century (19, 20, 22). Intriguingly, we successfully cloned the major late blight resistance QTL from the B3C1HP population—a subset of the International Potato Center (CIP) B3 population (22, 32). This gene is identical to the R8 cloned from the differential plant MaR8 (19). The B3 population originated from “Population A,” which integrated diverse resistance resources through four-way hybrids (33). In an effort to prioritize durable resistance, breeders attempted to exclude major Rpi genes by selecting only those progenies that were not fully resistant, resulting in Population B (34). Despite this selection against “strong” resistance, clones containing R8 in the B3C1HP population typically exhibit expanded lesions, and the resistance observed during severe epidemics aligns with what is traditionally considered a quantitative type (35, 36). The fact that R8 was retained through several rounds of recurrent selection suggests that its quantitative-like characteristics allowed it to “slip through” the screen that eliminated more typical, high-impact Rpi genes (37). Clear evidence demonstrates that, although R8 encodes a typical CC-NB-LRR NLR, it confers a quantitative resistance phenotype characterized by slow lesion expansion in the field (19, 22). In various breeding lines, the observed variation in resistance levels cannot be explained by a simple “presence-or-absence” R8 model (19–21). This distinct resistance pattern likely exerts moderate selection pressure on P. infestans, contributing to its substantial durability. In this study, we have uncovered “the tip of the iceberg”: RGAs may modulate R8-mediated immunity through molecular competition for Avr8 recognition (Fig. 8). This provides a theoretical foundation for understanding how durable resistance is fine-tuned within a plant’s genetic background. The results showed that both R8 CDS and multiple RGAs (sharing 90.87 to 94.76% identity with R8) are coexpressed in R8 marker–positive potato clones (fig. S3 and table S5). While most RGAs appear nonfunctional (Fig. 1, A and B, and table S4), specific variants—such as RGA 10-65-1, RGA 10-65-2, and RGA 10-58-2—definitively suppress R8-mediated resistance (Fig. 1C and fig. S4, A and B). We confirmed this suppression through stable transformation: Overexpressing RGA 10-65-1 effectively neutralized endogenous R8 resistance in potato and ectopically expressed R8 resistance in N. benthamiana (Fig. 2 and fig. S8, H to J). Mechanism dissection of the downstream signaling cascade indicates that RGA 10-65-1 interferes with R8/Avr8 association, leading to less R8 activation, which reduces the NRC2 oligomerization triggered by R8/Avr8 recognition, as well as its subsequent enrichment in the M fractions (Figs. 6 and 7, A to C). Collectively, these findings demonstrate that the expression of certain R8-RGAs acts as a molecular “brake,” modulating or attenuating R8-mediated late blight resistance. This explains the phenotypic mystery of why some R8 marker–positive potato clones exhibit variable resistance levels or, in some cases, total susceptibility. Plant NLRs are frequently organized in genomic clusters, arising from unequal crossing over, tandem duplication, and gene conversion. This clustering provides a flexible genetic reservoir for evolving new resistance specificities through domain swaps (38, 39). Clusters of the NLRs could result in massive expansions of pathogen-specific recognition resulting from homo- and heterodimerization/oligomerization possibilities among NLRs encoded by that cluster (7, 40). Most cloned potato late-blight resistance (Rpi) genes belong to NLRs that are typically organized in genomic clusters. A classic example includes the R1 cluster in S. demissum. The R1 locus exhibits marked variation in overall length and R1 homolog number. Frequent sequence exchanges occurred among R1 homologs within each clade but not among those in different clades (39). The R8 locus on chromosome IX is a prime example, containing at least 10 RGAs. Previous studies identified RGA 0.20 as the functional R8 (fig. S21A) (19). RGA 0.20 shares high nucleotide identity (94.01%) and amino acid identity (88.8%) with RGA 0.21 but shares lower nucleotide identities with other RGAs from 54.9 to 86.5% and amino acid identities from 33.6 to 76%. In this study, we cloned several RGAs with high protein identity (90.87 to 94.76%) to R8 (tables S5 and S6 and fig. S21, B and C), likely representing homologs from different haplotypes. We propose that random recombination and mutation within these high-identity homologs generate suppressive RGAs. By attenuating the R8-mediated response, these RGAs exert lower selection pressure on P. infestans populations, potentially explaining why R8 has remained effective for nearly a century while other race-specific genes have failed. From an evolutionary perspective, the regulatory function of these RGAs is likely not accidental. NLR genes are among the most polymorphic families in plant genomes, providing the raw material for natural selection to fine-tune immunity. Hyperactive immune systems often come with significant growth penalties. Specific RGAs, such as 10-65-1, might have been evolutionarily selected as “braking” elements to prevent these growth costs. Our findings suggest that durable resistance relies not merely on a single functional NLR but on a coordinated network of “sensors” (functional NLRs) and “regulators” (suppressive RGAs). Within genomic clusters, these regulatory elements fine-tune R gene activity to moderately attenuate resistance, thereby alleviating selection pressure on pathogen populations and promoting the long-term durability of the resistance locus. The competitive suppression mechanism we observed mirrors other characterized systems: The PigmS variant within the rice Pigm locus balances between disease resistance and vegetative growth through the competitive suppression of PigmR activity (41). In recent years, an increasing number of resistance systems conferred by pairs of distinct NB-LRR proteins have been described (8, 42). These paired NLR genes are arranged in a head to-head orientation in plant genomes, forming “dimers of heterodimers,” upon activation to initiate cell death (42–45). A notable parallel is found in wheat, where the neighboring NLR2-V negatively regulates the broad-spectrum resistance protein Pm21 (46). Our results further expand this paradigm by demonstrating that RGAs within a cluster serve as critical modulators of functional R protein activity, providing a flexible mechanism to ensure immune homeostasis. The rapid evolution of P. infestans enables the pathogen to readily overcome individual Rpi genes (13, 47). Pyramiding multiple Rpi genes, especially those broad-spectrum and durable R genes, would provide long-lasting resistance (48–50). Nevertheless, stacking multiple Rpi genes theoretically imposes higher evolutionary pressure on the pathogen (18). A more sustainable approach involves balancing the function of individually stacked R genes to provide broad-spectrum resistance while effectively extending their effective life span. Although transcriptional and translational control strategies have been explored to balance R gene function (51), using natural R gene-RGA combinations offers a promising alternative. In breeding programs, for instance, potato lines containing R8 with moderate (satisfactory) resistance, rather than those with hypersensitive high-level resistance, could be prioritized as parents to achieve more durable outcomes. Despite the insights gained regarding how RGAs regulate the R gene function, several questions remain to be explored. First, are there additional regulatory factors within the potato plant that modulate R8-mediated resistance? The quantitative resistance characteristic observed in F1 progeny carrying the broad-spectrum resistance gene RB (52) indicates that R gene efficacy depends not only on its presence but also on the specific genetic background of the recipient genotype. Second, the structural basis by which suppressive RGAs bind Avr8 without triggering immune signaling remains elusive. Given our observation that some RGAs exhibit weak interactions with Avr8 without interfering with the R8 function, further investigation into RGA structural variations and their functional implications is required. It remains unclear how nonsuppressive RGA variants, which also recognize Avr8, differ from suppressive ones. Is suppression caused solely by truncation, or are additional polymorphisms involved? Testing whether restoring the truncated LRR domain abolishes suppression, or generating artificial truncated variants, would help determine whether truncation alone is sufficient to confer suppression. Such analyses would provide mechanistic and evolutionary insights and facilitate the design of artificial RGAs to fine-tune R gene activity. Third, it remains to be determined whether these expressed RGAs have other functions and how their expression is regulated in potato plants. In summary, this study uncovers a mechanism whereby RGAs regulate NLR receptor activity through competitive interference, deepening our understanding of plant immune homeostasis and the establishment of durable resistance. It provides an alternative strategy and a design framework for crop disease resistance breeding. Shifting the focus from individual “star resistance genes” toward the complex genetic regulatory networks in which they reside will be pivotal for achieving sustainable crop disease management. MATERIALS AND METHODS Plant growth conditions The diploid potato (Solanum tuberosum L.) line “98100,” which carries the R8 locus and served as a parent for the R8 mapping population (22), was used for genetic transformation and pathogen assays. Potato plants were maintained in a greenhouse, and 6-week-old individuals were selected for P. infestans infection. Wild-type and NRC2/NRC3/NRC4 knockout N. benthamiana lines (27) were cultivated in a controlled-environment growth chamber at 22° to 25°C under a 16-hour light/8-hour dark photoperiod with 60% humidity. Plasmid construction All expression vectors were assembled via homologous recombination. The following backbone vectors were used: N-pH7LIC for HA- and Myc-tagged constructs; pK7WGR2.0 for N-terminal RFP fusions; pK7WFG2.0 for N-terminal GFP fusions. For BiFC assays, the N-terminal fragment of YFP (YN) was fused with R8 and 10–65-1, respectively, while the C-terminal fragment (YC) was fused with Avr8. For LUC complementation assays, the vector backbone pCAMBIA1300 was used. All constructs were transformed into Agrobacterium tumefaciens strain GV3101. Detailed primer sequences and vector specifications are provided in table S7. Plant transformation The R8-containing diploid potato line “98100,” characterized by high late blight resistance, was used for genetic transformation. RFP-tagged constructs (RFP-10-65-1 and RFP-10-58-11) were introduced into the A. tumefaciens strain GV3101. The transformation was performed using a modified stem-segment protocol: Stem segments (2 to 5 mm in length, excluding axillary and apical buds) were excised from 4- to 6-week-old potato plants. Explants were precultured for 1 day at 24°C under a 16-hour photoperiod (2000 lx). This was conducted on R3B medium [MS medium, sucrose (30 g liter−1), naphthaleneacetic acid (2 mg liter−1), and 6-benzylaminopurine (1 mg liter−1), pH 5.8] using a single layer of filter paper and 2 ml of PACM [MS medium, sucrose (30 g liter−1), casein hydrolysate (2 g liter−1), 2,4-dichlorophenoxyacetic acid (1 mg liter−1), and KT (0.5 mg liter−1), pH 6.5]. Following Agrobacterium infection, explants were incubated in the dark on R3B medium for 2 days to facilitate bacterial attachment and tissue dedifferentiation. Subsequently, explants were transferred to shoot induction medium [MS medium, sucrose (30 g liter−1), indole-3-acetic acid (0.2 mg liter−1), GA (0.2 mg liter−1), 6-benzylaminopurine (0.5 mg liter−1), ZT (2 mg liter−1), cefoxitin (200 mg liter−1), timentin (200 mg liter−1), and kanamycin (50 mg liter−1), pH 5.8]. After 2 weeks, they were transferred to ZCVK medium [MS medium, sucrose (20 g liter−1), ZT (1 mg liter−1), cefoxitin (200 mg liter−1), timentin (200 mg liter−1), and kanamycin (50 mg liter−1), pH 5.8] and subcultured every 2 weeks until shoot emergence. Regenerated shoots were excised and transferred to rooting medium [MS medium, sucrose (30 g liter−1), cefoxitin (200 mg liter−1), timentin (200 mg liter−1), and kanamycin (50 mg liter−1), pH 5.8]. Well-rooted transgenic lines were verified by PCR, and target gene expression levels were quantified via qRT-PCR using the primers listed in table S7. N. benthamiana transformation followed the procedure described by Qi et al. (53). First, wild-type N. benthamiana was transformed with GFP-R8 (hygromycin selection) to generate the OE line OE-R8-23. Homozygous OE-R8-23 plants were subsequently retransformed with RFP-10-65-1 (kanamycin selection). Double-transgenic plants coexpressing the constructs were identified and maintained using dual antibiotic selection. P. infestans inoculation assay P. infestans isolates were maintained on rye agar medium and incubated at 20°C in darkness for 15 to 20 days. Before harvest, sporangia formation was verified via microscopic examination. To prepare the inoculum, culture plates were flooded with 2 to 3 ml of sterile water, and mycelial mats were gently scraped using a spreader. The resulting suspension was filtered through a 200-mesh nylon cloth to remove hyphal debris. The sporangial concentration was then adjusted to the target density. To induce zoospore release, the suspension was incubated at 4°C for 2 to 3 hours and used immediately for plant inoculation. For potato infection assays, P. infestans isolates HB0914-2 and UK3928A were applied at a standardized concentration of 0.9 × 105 sporangia/ml. Seven- to 11-week-old potato plants with uniform growth were selected, and the third to fifth fully expanded compound leaves (counting from the first fully expanded leaf) were used for inoculation. Disease progression was monitored daily, and lesion diameters were measured using digital calipers at 4 to 5 days postinoculation. For N. benthamiana infections, the isolate 88069 was adjusted to a concentration of 1.5 × 105 sporangia/ml and inoculated onto fully expanded leaves of 4-week-old plants, with disease assessment conducted similarly at 5 to 7 days postinoculation. Agrobacterium-mediated transient expression assays The binary constructs were introduced into the A. tumefaciens strain GV3101 via electroporation. For long-term storage, cultures were mixed with 30% glycerol and kept at −80°C. Strains were streaked onto solid YEB (yeast extract and beef) medium plates supplemented with kanamycin and rifampicin and incubated at 28°C for 2 days. Single colonies harboring the constructs were selected, inoculated into liquid YEB medium, and grown overnight with shaking. The Agrobacterium culture was harvested by centrifugation and resuspended in MMA buffer (10 mM MES, 10 mM MgCl2, and 200 μM acetosyringone). Following 1-hour incubation in the dark, the bacterial suspension was infiltrated into plant tissues using a 1-ml needleless syringe. HR assay The target constructs were transiently expressed in 4- to 6-week-old N. benthamiana leaves via A. tumefaciens infiltration. The bacterial suspensions were adjusted to final OD600 values of 0.01 for the Avr8 construct and 0.5 for all other constructs. HR was evaluated 4 to 5 days post-agroinfiltration under ultraviolet (UV) illumination to visualize cell death. HR indexes were quantified according to the established scoring system (54). Confocal imaging Transient expression assays were performed in 3- to 4-week-old N. benthamiana plants, with Agrobacterium suspensions adjusted to OD600 values of 0.01 for Avr8, 0.1 for R8 and NRC2, and 0.5 for BiFC assays. Confocal microscopy was conducted 48 hours postinfiltration using a Leica TCS-SPE system with the following settings: GFP: fluorescence was captured with specific excitation/emission settings for GFP (488 nm/498 to 534 nm), RFP (561 nm/593 to 628 nm), and YFP (514 nm/530 to 575 nm). Split LUC complementation assay For transient expression assays, 4-week-old N. benthamiana leaves were infiltrated with A. tumefaciens strains harboring the target constructs at a final OD600 of 0.5. At 48 hours postinfiltration, detached leaves were treated with 15 mM luciferin applied to the infiltrated areas. Following a 10-min dark adaptation period to minimize the background signal, luminescence was captured using a NightSHADE LB 985 imaging system, and the resulting images were processed using indiGO software. Total protein extraction A. tumefaciens strains harboring the respective constructs were co-infiltrated into 4- to 5-week-old N. benthamiana leaves at a final OD600 of 0.5. Samples were harvested at 72 hours postinfiltration for NRC2 analysis, while all other protein samples were collected at 48 hours postinfiltration. Leaf discs (8 to 10 per sample, 1.0 cm in diameter) were excised with a cork borer, flash-frozen in liquid nitrogen with tungsten beads, and stored at −80°C until processing. For protein extraction, frozen samples were ground to a fine powder. NRC2 extraction was performed using GTMN buffer [10% glycerol, 50 mM tris-HCl (pH 7.5), 5 mM MgCl2, and 50 mM NaCl] supplemented with 10 mM dithiothreitol (DTT), 1× protease inhibitor cocktail (Sigma-Aldrich), and 0.2% NP-40 substitute, as described previously (29). For other proteins, GTEN buffer [10% glycerol, 25 mM tris-HCl (pH 7.5), 1 mM EDTA, and 150 mM NaCl] was used with 10 mM DTT, 1 × protease inhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride, and 0.2% NP-40. Lysates were incubated on ice for 30 min with periodic vortexing and subsequently clarified by centrifugation at 5000g for 10 min at 4°C. The resulting supernatants were used for subsequent Blue Native-PAGE (BN-PAGE) and SDS-PAGE analyses. Co-IP assays Protein extraction was conducted as detailed above using 1.2 ml of GTEN buffer per sample. Before the procedure, an 80-μl aliquot of the supernatant was reserved and mixed with 5× SDS sample buffer to serve as the input control. The remaining supernatant was incubated with 15 μl of agarose beads at 4°C for 2 hours under gentle rotation. The beads were subsequently collected and subjected to five washes with GTEN-based washing buffer (containing 1× protease inhibitor cocktail and 1 mM phenylmethylsulfonyl fluoride). Bound proteins were eluted by boiling the beads in 2× SDS sample buffer at 95°C for 10 min. Both eluates and input controls were resolved via SDS-PAGE and analyzed by immunoblotting. SDS-PAGE and immunoblotting For SDS-PAGE, protein supernatants were mixed with SDS sample buffer and denatured at 95°C for 10 min. Protein were separated on 8 to 10% SDS-PAGE gels and transferred onto polyvinylidene fluoride membranes following the manufacturer’s instructions. For immunoblotting, membranes were blocked in 5% milk in PBST (1× PBS containing 0.1% Tween 20) for 1 hour at room temperature, followed by a 2-hour incubation with primary antibodies (anti-GFP, anti-HA, anti-Myc, or anti-RFP) diluted at 1:5000. After washing three times with PBST (10 min each), the membranes were incubated with a horseradish peroxidase–conjugated anti-mouse secondary antibody (1:5000 dilution) for 1 hour, followed by another 30-min wash with PBST. Protein bands on the immunoblot were visualized using chemiluminescent ECL reagents (Servicebio, Wuhan, China). BN-PAGE assays BN-PAGE was performed as previously described (30). Briefly, the supernatant was mixed with 4× NativePAGE sample buffer, 5% Coomassie G-250 additive, and double-distilled H2O. Protein samples were loaded onto NativePAGE 3 to 12% bis-tris protein gels (Thermo Fisher Scientific, cat. no. BN2001BOX) and subjected to electrophoresis at 150 V under 4°C. Proteins were transferred to polyvinylidene fluoride membranes using a methanol-free transfer buffer. For protein fixation, membranes were incubated in 8% acetic acid for 15 min with gentle agitation and rinsed twice with distilled water. After air-drying, the membranes were reactivated with ethanol and processed for immunoblotting, as described above. Membrane enrichment assays Membrane enrichment was conducted following the protocol described by Duggan et al. (55) with modifications. Briefly, A. tumefaciens strains harboring the target constructs were transiently co-infiltrated in leaves of 4-week-old N. benthamiana, with each construct adjusted to a final OD600 of 0.5. At 3 days post-agroinfiltration, samples (0.1 g) were harvested and pulverized into a fine powder before being homogenized in 2 ml of lysis buffer [20 mM tris-HCl (pH 7.5), 20 mM KCl, 2 mM EDTA, 2.5 mM MgCl2, 25% glycerol, 250 mM sucrose, 5 mM DTT, and 1× protease inhibitor cocktail]. The homogenate was filtered through Miracloth and clarified by centrifugation at 1500g at 4°C. A 160-μl aliquot of the resulting supernatant was reserved and mixed with 5× SDS sample buffer to serve as the total protein fraction (T). The remaining supernatant was subjected to centrifugation at 21,000g for 1 hour at 4°C to partition the soluble and membrane components. The cytoplasmic fraction (S) was obtained by mixing 160 μl of this secondary supernatant with 5× SDS sample buffer. After the supernatant was carefully discarded, the microsomal pellet was resuspended in Minute Non-Denatured Protein Solubilization Reagent (Invent) and mixed with 2× SDS sample buffer to represent the M fraction. All protein fractions (T, S, and M) were denatured at 95°C for 10 min and analyzed via immunoblotting. An anti–H+-dependent adenosine triphosphatase (H+-ATPase) antibody (AS07260, Agrisera) was used as an internal reference to detect endogenous PM ATPase. qRT-PCR Total RNA was extracted from N. benthamiana and potato leaf tissues using the Plant Total RNA Kit (ZOMANBIO) in accordance with the manufacturer’s instructions. First-strand cDNA was synthesized using All-In-One 5 × RT MasterMix (ABM), and qRT-PCR was performed using BlasTaq 2 × qPCR MasterMix (ABM) on an ABI 7300 Real-Time PCR System (Applied Biosystems). Actin was used as the internal reference gene for both N. benthamiana and potato. Relative transcript abundance was determined via the comparative Ct (cycle threshold) method (56). Detailed primer information is provided in table S7. Statistical analysis All data were processed and analyzed using GraphPad Prism 8.0 software. Statistical significance was evaluated by a one-way analysis of variance (ANOVA) followed by pairwise or multiple comparisons. All values and error bars presented are the means ± SEM of three or more experimental replicates.