The BTB/POZ‐MATH Protein ZmBPM1 Interacts With Autophagy‐related Protein ZmATG6 and Modulates the NLR Protein Rp1‐D21‐Mediated Defense Response in Maize Authors: Chang‐Xiao Tang, Tong‐Tong Liu, Qi‐Dong Ge, Xiao‐Ying Liu, Wanxiao Wang, Yang Sun, Yanfei Liu, Faqiang Li, Guan‐Feng Wang Published: 2026 ABSTRACT ABSTRACTBTB/POZ‐MATH (BPM) proteins are key regulators to modulate the stability of their substrates via the 26S proteasome. However, their function in plant immunity, particularly in modulating the activity of nucleotide‐binding, leucine‐rich repeat (NLR) receptor, remains unknown. Here, we demonstrate that maize ZmBPM1, but not its close homolog ZmBPM2, negatively regulates the autoactive NLR protein Rp1‐D21 by promoting its degradation through both the 26S proteasome and the autophagy pathways. ZmBPM1 overexpression inhibits Rp1‐D21‐mediated hypersensitive response (HR), whereas its mutation enhances the HR strength in maize. Notably, ZmBPM1 is primarily localized in the autophagosome‐like punctate structures and facilitates relocation of Rp1‐D21 from the nucleo‐cytoplasmic compartment to these punctate structures. Furthermore, ZmBPM1 promotes the degradation of Rp1‐D21 through autophagy pathway by interacting with the autophagy‐related protein ZmATG6a/ZmATG6b, therefore inhibiting Rp1‐D21‐mediated HR. Strikingly, ZmBPM1 also acts as a negative regulator of resistance against southern corn rust caused by Puccinia polysora, a fungus that appears to enhance ZmBPM1‐mediated autophagic activity. Our study thus identifies an autophagy‐dependent mechanism by which a BPM protein finely controls NLR homeostasis and plant immunity, revealing a previously unrecognized immune regulation mechanism in plants. Introduction Plant immunity is a complicated defense system that enables plants to defend against different pathogens. Traditionally, plant immune system is classified into two primary layers: PAMP‐triggered immunity (PTI) and effector‐triggered immunity (ETI) [1]. In PTI, pathogen‐associated molecular patterns (PAMPs) can be detected by cell‐surface pattern recognition receptors (PRRs) to trigger immunity. In ETI, pathogens secreted effectors delivered into plant cells are recognized by intracellular immune receptors, which mostly encode nucleotide‐binding, leucine‐rich repeat (NLR or NB‐LRR) proteins [2]. Compared to PTI, ETI is a stronger immune response, which is often accompanied by a rapid localized cell death known as the hypersensitive response (HR) [1, 3]. PTI and ETI mutually coordinate to act in plant immune responses [4, 5, 6]. Classical NLRs are composed of three primary domains: a variable N‐terminal domain for signaling, the central NB‐ARC (APAF1, certain R gene products and CED‐4) domain for ATP/ADP binding or hydrolysis, and the C‐terminal LRR domain for determining the specific recognition of pathogen effectors [2, 7]. Typically, NLRs are grouped into two major classes based on the N‐terminal structures; Toll‐interleukin 1 receptor (TIR)‐NLRs (TNLs) and coiled‐coil (CC)‐NLRs (CNLs). NLRs detect their specific effectors either directly or indirectly through interactions with their associated proteins, referred to as NLR regulators [2]. NLRs are distributed across different subcellular compartments, and some NLR regulators can modulate NLR activities by changing the subcellular localization of NLRs [2, 8, 9, 10, 11, 12, 13]. Additionally, the structural integrity of NLRs is meticulously controlled. Inappropriate activation of NLRs often triggers autoimmunity even in the absence of their cognate effectors. One well‐documented case is the CNL protein Rp1‐D21 in maize. Rp1‐D21 originates from intragenic recombination between Rp1‐D and Rp1‐dp2, two highly similar CNL‐encoding genes located within the same Rp1 locus [14, 15]. Maize lines that harbor Rp1‐D21 exhibit nonspecific resistance to common rust caused by Puccinia sorghi and southern rust caused by Puccinia polysora [16, 17]. Moreover, in maize lines that harbor Rp1‐D21, autoimmune phenotypes manifest in these lines even in the absence of pathogens; these include stunted growth and spontaneous HR lesions on the leaves [8, 18, 19]. The autoimmune phenotype depends on genetic background, therefore Rp1‐D21 is considered as a valuable resource for identifying novel defense regulators [8, 19, 20]. For example, caffeoyl CoA O‐methyltransferase (CCoAOMT) and hydroxycinnamoyl transferase (HCT), both encoding potential homologs of enzymes involved in the lignin biosynthesis pathway, negatively influence Rp1‐D21‐mediated HR through interacting with its CC domain [21, 22]. Additionally, recent studies have identified two metacaspases ZmMC1 and ZmMC2, the key components in the endosomal sorting complex required for transport (ESCRT)‐I machinery ZmVPS23/ZmVPS23L and the F‐box protein ZmFBK1, regulate Rp1‐D21‐conferred HR likely via alterations in its subcellular localization [8, 11, 13]. Autophagy represents a conserved degradation process that maintains cellular homeostasis by clearing and recycling intracellular components through double membrane vesicles known as autophagosomes [23, 24, 25]. This process allows cells to remove misfolded proteins, defective organelles, and miscellaneous intracellular refuse, thereby promoting cell survival under stress conditions. During the formation of autophagosome, autophagy‐related (ATG) proteins play critical roles and they can be classified into several complexes [26]. One of the critical components is ATG6/Beclin1/VPS30, which plays a critical role in the nucleation of vesicles during the initial stages of autophagosome formation via the phosphatidylinositol 3‐kinase (PI3K) pathway [27]. In plant immunity, autophagy acts as a double‐edged sword―either providing protection for plants to degrade damaged cellular constituents and maintain immune homeostasis, or being manipulated by pathogen effectors to facilitate their infection [28, 29, 30]. Consequently, the mechanisms through which autophagy regulates plant immunity are complex and multifaceted. The BPM (BTB/POZ‐MATH) proteins are substrate adaptors of Cullin3 (CUL3)‐based E3 ubiquitin ligases and mostly act to degrade their substrates by the 26S proteasome system [31, 32, 33]. They contain two major domains: an N‐terminal MATH (Meprin and TRAF homology) domain and a C‐terminal BTB/POZ (broad complex, tramtrack, bric‐a‐brac/POX virus, and zinc finger) domain [34]. The BTB/POZ domain determines the protein interaction with CUL3 E3 ligase and the MATH domain is responsible for the interaction with the BPM substrates [35, 36, 37]. Plant BPM proteins are divided into two main clades according to the phylogenetic analysis; the core group and the expanded group [38]. The core group contains all the six Arabidopsis BPM proteins (AtBPM1‐AtBPM6), and the expanded group might be evolved as a monocot‐specific clade [38, 39]. Up to now, most advances in plant BPM proteins are from the core group in Arabidopsis. They usually interact with and modulate the stability of various transcription factors and other signaling molecules, which allows BPM proteins to act as adaptors or scaffolds in different signaling pathways. They play critical roles in plant growth and development, fertility, flowering, stomatal dynamics, fatty acid metabolism, the abscisic acid and jasmonic acid signaling, and abiotic stress responses, etc [32, 33, 35, 36, 40, 41, 42, 43, 44, 45, 46, 47]. However, little is known about the function of BPM proteins in plant immunity, especially in NLR‐mediated defense response. Here we report that the maize expanded group BPM protein ZmBPM1, but not its homolog ZmBPM2 plays negative roles in the NLR protein Rp1‐D21‐triggered autoactive HR (auto‐HR). ZmBPM1 interacts with the CC domain of Rp1‐D21 (CCD21) and leads to the trafficking from nucleo‐cytoplasm to the autophagosome‐like structures. Moreover, ZmBPM1 modulates the stability of Rp1‐D21 most likely through interacting with the autophagy‐related protein ZmATG6a/ZmATG6b and promoting autophagy. Strikingly, ZmBPM1 negatively regulates the resistance to southern corn rust (SCR) in maize. This study uncovers the molecular mechanism for BPM protein in modulating the activity of NLR and in plant immunity. Results ZmBPM1, But Not Its Close Homolog ZmBPM2, Inhibits Rp1‐D21‐Triggered Auto‐HR upon Expressed in N. benthamiana Based on the amino acid sequences of Arabidopsis AtBPMs, we searched the maize genome database (MaizeGDB) and identified 28 ZmBPM homologs which contain the MATH and BTB/POZ domains (Figure S1 and Table S1). According to our previous RNA‐seq results [21, 48], we identified that the genes encoding ZmBPM1 and ZmBPM2 were significantly up‐regulated in the Rp1‐D21 mutant relative to wild type, especially for the ZmBPM1‐encoding gene, which was 178.0 and 275.5 times up‐regulated in Rp1‐D21 in B73 and Mo17 background, respectively (Table S1). However, the expression levels of other ZmBPM encoding genes were either too low to be detected or were not differentially expressed (Table S1). Phylogenetic analysis showed that maize ZmBPMs and Arabidopsis AtBPMs were divided into two groups, and ZmBPM1 and ZmBPM2 were in the same clade belonging to the expanded group, while the six Arabidopsis AtBPMs were in the core group (Figure S1). Rp1‐D21:HA (with a C‐terminal 3× HA tag) confers an auto‐HR upon transiently expressed in N. benthamiana [13, 14, 22]. To investigate whether ZmBPM1 and ZmBPM2 affect Rp1‐D21‐mediated auto‐HR, we co‐expressed ZmBPMs:EGFP (with the C‐terminal EGFP) and Rp1‐D21:HA by agro‐infiltration into N. benthamiana. As previously reported [11, 14, 22], HCT1806:EGFP and GUS:EGFP were used as positive and negative controls here. We found that ZmBPM1 inhibited Rp1‐D21‐triggered auto‐HR, and the suppression effect was similar to that caused by HCT1806, while ZmBPM2 showed similar effect to GUS (Figure 1A). Consistently, ZmBPM1, but not ZmBPM2 significantly decreased ion leakage conductivity levels compared to GUS (Figure 1B). Interestingly, ZmBPM1, but not ZmBPM2, appeared to decrease the protein amounts of Rp1‐D21 relative to GUS (Figure 1C). ZmBPM1 negatively regulates Rp1‐D21‐triggered auto‐HR. (A) ZmBPM1, but not ZmBPM2, inhibits Rp1‐D21‐triggered auto‐HR in N. benthamiana. Representative leaves showing co‐expression of ZmBPMs:EGFP and Rp1‐D21:HA were taken at 72 h post‐infiltration (hpi) (left: before clearing with alcohol; right: after clearing with alcohol). The numbers in parentheses indicate the count of leaves that developed HR out of the total infiltrated. (B) Quantification of cell death via electrolyte leakage at 62 h after co‐expression of different proteins and Rp1‐D21 in N. benthamiana. Data are displayed as means ± standard errors (n = 4 independent biological replicates); one‐way ANOVA with Tukey's multiple‐comparison test identified statistically significant differences (p < 0.05), indicated by distinct letters (a, b). (C) Immunoblotting detection of different proteins extracted at 40 hpi by anti‐HA and anti‐GFP antibodies. Coomassie Brilliant Blue (CBB)‐stained Rubisco serves as the loading reference. Band intensities were normalized to the band signal in the first lane, which was set to 1.0. The experiments were conducted three times with consistent results. (D) ZmBPM1 partially inhibits Rp1‐D21‐mediated HR when overexpressed in maize. The overexpression lines containing homozygous ZmBPM1 were crossed as female with B104:Rp1‐D21, and the phenotype of the relevant F1 plants were assessed at V6 stage in the field. The right panel of each genotype displays the images of the entire seedlings, with the fifth leaves highlighted in white rectangles and enlarged in the left panel. (E) Overexpression of ZmBPM1 partially inhibits Rp1‐D21‐associated plant height. Data are displayed as means ± SE (n > 5). Student's t‐test identified statistically significant differences (*p < 0.05). (F) Chlorophyll contents are measured at V6 stage. Data are displayed as means ± SE (n > 5). Student's t‐test identified statistically significant differences (*p < 0.05). (G‐H) The transcript levels of ZmBPM1 (G), PR1 and PR5 (H) are detected by real‐time RT‐PCR. Data are presented as means ± SE (n = 3). Student's t‐test identified statistically significant differences (***p < 0.001). (I, J) ZmBPM1 mutation enhances Rp1‐D21‐triggered auto‐HR in maize. The BC1F1 progeny was produced from Zmbpm1‐1 and B104:Rp1‐D21. For the genotypes lacking Rp1‐D21, no obvious HR was observed regardless of Zmbpm1‐1 genotypes (I). For the genotypes carrying Rp1‐D21, HR was more pronounced in Zmbpm1‐1 homozygous seedlings than heterozygous seedlings (J). Enlarged images in the white boxes were displayed on the left or right panels of their respective plants. (K) Plant height are measured in Zmbpm1‐1 as Figure 1E. (L) Chlorophyll contents are measured in Zmbpm1‐1 as Figure 1F. (M, N) The transcript levels of ZmBPM1 (M), PR1 and PR5 (N) in Zmbpm1‐1 are assessed by real‐time RT‐qPCR as Figure 1G,H (**p < 0.01, ***p < 0.001). ZmBPM1 Overexpression Partially Inhibits Rp1‐D21‐Conferred Auto‐HR in Maize To validate whether ZmBPM1 suppresses Rp1‐D21‐triggered auto‐HR in maize, we generated ZmBPM1:GFP overexpression (OE) lines in B104. We selected two ZmBPM1 OE lines (OE‐Y12 and OE‐Y49) and their corresponding negative (Neg) siblings (Neg‐Y12 and Neg‐Y49) for subsequent experiments. Immunoblotting confirmed that ZmBPM1 was only substantially expressed in two OE lines (Figure S2A). These lines were crossed as females to B104:Rp1‐D21 (heterozygous for Rp1‐D21) [8]. The resulting F1 plants exhibited 1:1 segregation for Rp1‐D21, as plants carrying Rp1‐D21 were dwarf and exhibited auto‐HR phenotype [8, 13]. We compared plants carrying Rp1‐D21 from OE and Neg crosses grown in the field, including OE‐Y12 × D21 and Neg‐Y12 × D21, OE‐Y49 × D21 and Neg‐Y49 × D21. At V6 stage, Neg‐Y12 × D21 and Neg‐Y49 × D21 seedlings exhibited strong HR lesions on the leaves, whereas the corresponding OE crosses displayed much weaker HR (Figure 1D). Consistent with the HR phenotype, OE‐Y12 × D21 and OE‐Y49 × D21 were significantly taller and contained more chlorophyll contents than Neg‐Y12 × D21 and Neg‐Y49 × D21, respectively (Figure 1E,F). Moreover, the transcript levels of PR1 and PR5, the well‐known defense marker genes, were significantly decreased in the two OE crosses (OE‐Y12 × D21 and OE‐Y49 × D21) compared to their relevant Neg siblings (Neg‐Y12 × D21 and Neg‐Y49 × D21), which is in contrast to the expression pattern of ZmBPM1 (Figure 1G,H). In the absence of Rp1‐D21, no height difference was observed between OE and Neg plants (Figure S2B), indicating that overexpression of ZmBPM1 per se in maize had no obvious effect on plant height. Collectively, these results demonstrate that ZmBPM1 partially suppressed Rp1‐D21‐mediated autoimmunity when it was overexpressed in maize. ZmBPM1 Mutation Augments Rp1‐D21‐Conferred Auto‐HR in Maize To further confirm the function of ZmBPM1 in Rp1‐D21‐conferred auto‐HR in maize, we ordered the Zmbpm1‐1 homozygous mutant in which the UniformMu transposon was inserted in the 3’‐UTR of ZmBPM1 (Figure S3A,B). ZmBPM1 transcript abundance was markedly lower in Zmbpm1‐1 than in its wild type (WT) siblings (Figure S3C), indicating Zmbpm1‐1 is a knock‐down mutant. We crossed Zmbpm1‐1 as female into B104:Rp1‐D21, and backcrossed the resulting F1 progeny containing Rp1‐D21 to Zmbpm1‐1 and generated the BC1F1 population. Field‐grown plants were genotyped and grouped into four classes based on the presence of Zmbpm1‐1 (heterozygous or homozygous) and Rp1‐D21 (Figure 1I,J; Figure S4). Among plants carrying Rp1‐D21, those homozygous for bpm1‐1 (bpm1‐1/bpm1‐1;D21+) displayed more severe autoimmune phenotypes (stronger HR lesions, lower height, less chlorophyll contents) than their heterozygous siblings (BPM1/bpm1‐1;D21+). In the absence of Rp1‐D21, no autoimmune phenotypes were observed regardless of Zmbpm1‐1 genotypes (BPM1/bpm1‐1;D21− and bpm1‐1/bpm1‐1;D21−) (Figure 1I–L). RT‐qPCR analysis revealed that t ZmBPM1 transcript abundance was markedly lower in bpm1‐1/bpm1‐1;D21+ than in BPM1‐1/bpm1‐1;D21+, which is in contrast to the expression patterns of PR1 and PR5 (Figure 1M,N). These results indicated that ZmBPM1 mutation augments Rp1‐D21‐conferred auto‐HR in maize. ZmBPM1 interacts with CCD21 and inhibits CCD21‐triggered auto‐HR. (A–C) ZmBPM1 inhibits CCD21‐triggered auto‐HR in N. benthamiana. Representative leaves showing co‐expression of ZmBPMs:EGFP and CC D21:EGFP at 72 hpi (A). Quantification of cell death via electrolyte leakage at 62 hpi. Statistical analysis was conducted as in Figure 1B (n = 8). (B). Immunoblot analysis of protein accumulation using anti‐GFP. CBB‐stained Rubisco is shown as the loading control. Band intensities were normalized to the signal in the first lane, which was set to 1.0 (C). (D) ZmBPM1 interacts with CCD21 as determined by Co‐IP assays. Immunoprecipitation was conducted using anti‐GFP microbeads, while the expression of CCD21 and ZmBPMs was detected using anti‐myc and anti‐GFP antibodies, respectively. (E) ZmBPM1 interacts with CCD21 as determined by SLCA. A representative image was shown at 40 hpi. (F) ZmBPM1 interacts with CCD21 as determined by BiFC assays in maize protoplasts. ZmBPMs:YN (the N‐terminus of YFP) and CCD21:YC (the C‐terminus of YFP) were driven by their native promotersImages were taken at 18 hpi. The punctate signal is indicated by yellow arrows. Bars, 20 µm. The Effect of ZmBPM1 on Other NLR or Elicitor‐Triggered Cell Death We further investigated the effects of ZmBPM1 on HR or cell death triggered by other NLRs or elicitors. Two CNLs, RPM1(D505V) from Arabidopsis and MLA10(D502V) from barley, confer auto‐HR in N. benthamiana [49, 50]. INF1 and Bax are two cell death elicitors from Phytophthora infestans and mouse, respectively. We co‐expressed ZmBPM1 with them in N. benthamiana and found that ZmBPM1 partially suppressed RPM1(D505V)‐triggered HR, but did not affect cell death triggered by MLA10(D502V), INF1 and Bax (Figure S5). ZmBPM1 Inhibits CCD21‐Mediated Auto‐HR and Interacts with CCD21 Among the three major domains in Rp1‐D21, the N‐terminal CCD21 domain, but not the central NB‐ARC or C‐terminal LRR domains is required for triggering auto‐HR in N. benthamiana [11, 14, 22]. We co‐expressed ZmBPMs:EGFP and CCD21:EGFP in N. benthamiana, and found that ZmBPM1, but not ZmBPM2, inhibited CCD21:EGFP‐mediated auto‐HR (Figure 2A,B). Similar to the full length of Rp1‐D21, the protein abundance of CCD21 was reduced when coexpressed with ZmBPM1 but not with ZmBPM2 (Figure 2C). To determine whether ZmBPMs interacted with CCD21, we firstly conducted co‐immunoprecipitation (Co‐IP) assays by transiently co‐expression of ZmBPMs:EGFP and CCD21:Myc (fused with a C‐terminal 4× c‐Myc tag) in N. benthamiana. Both ZmBPM1 and ZmBPM2 interacted with CCD21, although ZmBPM2 binding was weaker (Figure 2D). These interactions were confirmed by split‐luciferase complementation assays (SLCA) with similar results (Figure 2E). Moreover, we also confirmed the interactions in maize protoplasts via bimolecular fluorescence complementation (BiFC) assays with ZmBPMs and CCD21 driven by their native promoters. Strikingly, ZmBPM1 interacted with CCD21 predominantly in the punctate structures, whereas ZmBPM2 interacted with CCD21 in plant cells without obvious puncta (Figure 2F). ZmBPM1 Exhibits Different Subcellular Localization with ZmBPM2 To explore whether ZmBPM1 and ZmBPM2 have different locations, we conducted the subcellular localization experiments by introducing ZmBPMs:GFP into maize protoplasts. As expected, ZmBPM1 exhibited punctate dot localization, while ZmBPM2 did not show clear punctate dots (Figure 3A). To further validate the results, we expressed ZmBPMs:EGFP in N. benthamiana lines which contain the nuclear marker H2B:TagRFP [51]. Consistent with the results from maize protoplasts, ZmBPM1 was predominantly localized in the puncta in both the cytoplasm and nucleus (Figure 3B,C). In contrast, ZmBPM2 was mainly located in the cytoplasm and nucleus, with much less puncta (Figure 3B,C). Subsequently, we conducted microsomal fractionation assays, revealing that both ZmBPM1 and ZmBPM2 accumulated in soluble and microsomal fractions, with ZmBPM1 accumulating higher amounts in the microsomal fraction (Figure 3D). These data indicated that ZmBPM1 is present in the nucleus, cytoplasm, and microsomal fractions, characterized by punctate structures. ZmBPM1, but not ZmBPM2, translocates CCD21 from nucleo‐cytoplasm to the puncta. (A‐D) The subcellular localization of ZmBPM1 differs from that of ZmBPM2. (A) The localization of ZmBPM1:EGFP and ZmBPM2:EGFP were observed in maize protoplasts. The punctate signal is indicated by yellow arrows. Scale bars, 20 µm. (B) ZmBPMs:EGFP were agro‐infiltrated into N. benthamiana (stably expressing the nuclear marker H2B:TagRFP). Enlarged images in the white boxes were displayed in the right panel. The punctate signal is indicated by yellow arrows. Scale bars, 20 µm. (C) Quantification of puncta numbers expressing ZmBPM1:EGFP or ZmBPM2:EGFP. The data are presented as means ± SE (n = 10). Statistical analysis was conducted by Student's t‐test (***p < 0.001). (D) Protein microsomal fractionation analyses for ZmBPMs:EGFP. Total (T), microsomal (M) and soluble (S) fractions were extracted at 40 hpi. Anti‐H+‐ATPase—and anti‐APX2 antibodies were used as markers for detecting the microsomal and soluble fractions, respectively. (E‐G) ZmBPM1 leads to the subcellular relocation of CCD21. ZmBPMs and CCD21 fused with distinct fluorescence proteins were co‐expressed in maize protoplasts (E) or N. benthamiana (F). Protein microsomal fractionation assays were conducted for co‐expression of ZmBPMs and CCD21 in N. benthamiana (G). ZmBPM1, But Not ZmBPM2 Translocates Rp1‐D21 from Nucleo‐Cytoplasm to Puncta Rp1‐D21 and its CCD21 domain were observed to be predominantly locate in the cytoplasm and nucleus, and both localizations are crucial for them to trigger auto‐HR in N. benthamiana [13, 52]. Since ZmBPM1 and ZmBPM2 show distinct localizations in plant cells, we speculated that they potentially impact the subcellular localization of Rp1‐D21 or CCD21 differently. Consistent with our hypothesis, co‐expressing ZmBPMs:EGFP and CCD21:mRFP in maize protoplasts revealed that CCD21:mRFP was moved to the punctate dots by ZmBPM1, but not by ZmBPM2 (Figure 3E). We also validated these results in N. benthamiana with similar results (Figure 3F). We further verified the results by biochemical fractionation analyses. CCD21:EGFP accumulated more in the microsomal fraction when co‐expressed with ZmBPM1 than with ZmBPM2 (Figure 3G). These data demonstrated that ZmBPM1 relocates CCD21 from the nucleo‐cytoplasmic compartment to the microsomal fraction, particularly in the puncta. The N‐Terminal MATH Domain Is Required for ZmBPM1 to Suppress Auto‐HR ZmBPM1 contains an N‐terminal MATH domain and a C‐terminal BTB/POZ domain (Figure 4A). To determine which domain is critical for ZmBPM1 to inhibit CCD21‐triggered HR, we transiently co‐expressed different domains with CCD21:EGFP in N. benthamiana. It revealed that the MATH domain, but not the BTB/POZ domain inhibited CCD21:EGFP‐triggered HR, consistent with the effect caused by the full‐length ZmBPM1 (Figure 4B). Intriguingly, the MATH domain also reduced the accumulation of CCD21 (Figure 4B). We further showed that the MATH domain, but not the BTB/POZ domain interacted with CCD21 as revealed by Co‐IP and SLCA (Figure 4C,D). Consistent with the full‐length ZmBPM1, the MATH domain, but not the BTB/POZ domain displayed punctate localization and altered CCD21 to puncta (Figure 4E,H). These results indicated that the MATH domain is required for ZmBPM1 to suppress CCD21:EGFP‐mediated auto‐HR. The MATH domain of ZmBPM1 inhibited the HR mediated by CCD21 in N. benthamiana. (A) The schematic diagram illustrates the domain architecture of ZmBPM1. The top section indicates the amino acid positions of different domains. (B) The N‐terminal MATH domain suppresses CCD21‐triggered auto‐HR in N. benthamiana. Representative leaves showing co‐expression of different domains of ZmBPM1 and CCD21:EGFP at 72 hpi. Quantification of cell death and immunoblot analysis were performed as Figure 2B,C. (C) Investigating the interactions between various fragments of ZmBPM1 and CCD21 using Co‐IP assays as Figure 2D. (D) The interactions between CCD21 and various fragments of ZmBPM1 were assessed using SLCA, with a representative image captured at 60 hpi. (E) The subcellular localization of different fragments of ZmBPM1 in N. benthamiana (carrying H2B:RFP). The images within the white boxes were magnified and displayed in the right panel. The punctate signal is indicated by yellow arrows. Scale bars, 20 µm. (F, H) The quantification of puncta numbers from different fragments of ZmBPM:EGFP was co‐expressed with nuclear markers H2B:RFP or CCD21:RFP. Statistical analysis (means ± SE, n = 10) was conducted by Student's t‐test (***p < 0.001). (G) The N‐terminal MATH domain changes the subcellular localization of CCD21 to the punctate dots. Scale bars, 20 µm. To investigate the molecular mechanisms of the functional difference between ZmBPM1 and ZmBPM2 on suppressing CCD21‐mediated HR, we compared their amino acid sequences and found that ZmBPM1 and ZmBPM2 have largely similar MATH domain; however, ZmBPM1 lacks partial of the C‐terminal BTB‐POZ domain and the entire BACK domain, which are present in ZmBPM2 and most of other BPM proteins (Figures S1 and S6). Domain truncation assays showed that the MATH domain of ZmBPM2 (2‐MATH), but not its BTB/POZ‐BACK domain (2‐BTB/POZ‐BACK), also suppressed CCD21:EGFP‐mediated HR (Figure S7A,B). Interestingly, not only 2‐MATH, but 2‐BTB/POZ‐BACK interacted with CCD21 (Figure S7C). The 2‐MATH domain displayed punctate localization both in the cytoplasm and in the nucleus, while 2‐BTB/POZ‐BACK formed large condensates (Figure S7D,E). Furthermore, 2‐MATH, but not 2‐BTB/POZ‐BACK, changed the subcellular localization of CCD21 to the cytoplasmic puncta (Figure S7F,G). These results indicated that the functional difference between ZmBPM1 and ZmBPM2 on suppressing CCD21‐mediated HR might come from the difference in the C‐terminal part of BTB/POZ and the BACK domains, but not the MATH domains. We further used Alphafold3 to predict the protein structure and it showed that the MATH domains of ZmBPM1 and ZmBPM2 displayed similar structure (Figure S8A), which is consistent to their conserved function in suppressing CCD21‐mediated HR. Interestingly, for the full length proteins, ZmBPM1 adopted an open, loose conformation, while ZmBPM2 formed a highly compact conformation driven by the C‐terminal part of the BTB‐POZ and the BACK domains (Figure S8B), which might contribute to self‐association. Indeed, ZmBPM2, but not ZmBPM1 can self‐associate as detected by Y2H and split luciferase assays (Figure S8C,D). These results suggested that self‐association mediated by the BTB/POZ‐BACK domain compacts full‐length ZmBPM2 and sterically blocks the substrate‐binding pocket of the MATH domain, while ZmBPM1 without homotypic multimerization retains an open conformation to enable its normal MATH function in affecting CCD21‐mediated HR. ZmBPM1 Displays Overlapping Subcellular Localization with the Autophagic Markers To investigate where ZmBPM1 locates in plant cells, we co‐expressed ZmBPM1 with different cellular markers displaying punctate structures. When co‐expressed with autophagosome marker proteins in maize protoplasts, ZmBPM1 but not ZmBPM2 showed overlapping co‐localization with RFP:AtATG6 in punctate structures (Figure 5A). Similarly, ZmBPM1 partially co‐located with maize TagRFP:ZmATG8d in the puncta in N. benthamiana, while ZmBPM2 showed much less co‐localization (Figure 5B). Notably, co‐expression of ZmBPM1:EGFP with TagRFP:ZmATG8d led to a significant increase in the number of TagRFP:ZmATG8d punctate dots compared to ZmBPM2:EGFP (Figure 5C), suggesting that ZmBPM1 promotes the accumulation of autophagic bodies. ZmBPM1 also partially co‐localized with the endosomal markers SYP61 and VSR2 both in maize protoplasts and N. benthamiana (Figure S9A,B). Additionally, ZmBPM1 partially co‐localized with the endoplasmic reticulum marker, but not with the peroxisome and golgi markers (Figure S9C–E). ZmBPM1 exhibits co‐localization with autophagy markers and causes the degradation of CCD21 through both the autophagy and proteasome pathways. (A) Co‐localization of ZmBPMs:GFP with the autophagy marker RFP:AtATG6 in maize protoplasts. Scale bar, 20 µm. (B) Co‐localization of ZmBPMs:EGFP with the autophagy marker TagRFP:ZmATG8d in N. benthamiana. Scale bar, 20 µm. (C) The quantification of the number of TagRFP:ATG8d punctate dots in (B). Statistical analysis (means ± SE, n = 10) was conducted by Student's t‐test (***p < 0.001). (D) The degradation of CCD21 by ZmBPM1 depends on both 26S proteasome and autophagy pathways. ZmBPM1:EGFP or ZmBPM2:EGFP was co‐infiltrated with CCD21:TaqRFP into N. benthamiana. After 28 h, leaves were infiltrated with DMSO, 10 mm 3‐MA or 100 µm MG132 for an additional 6 h. Total proteins were immunoblotted with anti‐GFP and anti‐RFP. CCD21 levels were normalized to CBB staining. Band intensities were normalized to the value in the lane co‐expressed by CCD21 and ZmBPM2 under DMSO treatment, which was set to 1.0. (E) The cell‐free ubiquitination of CCD21 in Zmbpm1‐1. Protoplast lysates from CCD21:mRFP‐transfected cells were incubated with lysates from WT or Zmbpm1‐1 plants. CCD21:mRFP was enriched by anti‑RFP immunoprecipitation, and the ubiquitinated fraction was visualized with anti‑ubiquitin under both ATP‑replete and ATP‑depleted conditions. (F) In vivo ubiquitination of CCD21 analysis was examined by transfecting CCD21:mRFP into protoplasts extracted from the ZmBPM1:GFP overexpression line. Immunoblotting was performed as in (E). (G) ZmBPM1 might promote autophagic flux. ZmBPM1:EGFP or ZmBPM2:EGFP was co‐expressed with TagRFP:ZmATG8d in N. benthamiana. Immunoblotting was conducted with anti‐GFP (left) and anti‐RFP (right) to detect free EGFP and TagRFP, respectively. (H) Autophagy inhibitor 3‐MA increases CCD21:mRFP accumulation in protoplasts from the ZmBPM1:GFP overexpression line. ZmBPM1 Degrades CCD21 through Both the 26S Proteasome and the Autophagy Pathways Co‐expression of ZmBPM1 appeared to reduce the protein levels of Rp1‐D21 or CCD21 (Figures 1C and 2C). To further validate the results, we co‐expressed ZmBPMs with CCD21, Rp1‐D21 or Rp1‐dp2 (the parental NLR of Rp1‐D21 without conferring auto‐HR) and detected the protein accumulation at different time points. It showed that ZmBPM1, but not ZmBPM2 reduced the protein levels of CCD21, Rp1‐D21 or Rp1‐dp2 at both 36 and 48 hpi (Figure S10A–C). To explore which pathways ZmBPM1 used to modulate the protein stability of CCD21, we treated the samples with DMSO, the autophagy inhibitor 3‐MA or 26S proteasome inhibitor MG132. Relative to DMSO controls, CCD21 abundance was substantially elevated upon co‑expression with ZmBPM1 in the presence of either 3‑MA or MG132 (Figure 5D). These results indicated that ZmBPM1 modulates the homoeostasis of CCD21 through both the autophagy and 26S proteasome pathways. Notably, the stability of ZmBPM1 and ZmBPM2 was also modulated by the autophagy and 26S proteasome pathways (Figure 5D). To assess whether ZmBPM1 facilitates ubiquitination and subsequent degradation of CCD21, we performed cell‐free ubiquitination assays with extracts from Zmbpm1‐1. Proteins from protoplasts expressing CCD21:mRFP were incubated with extracts from either wild‐type or Zmbpm1‐1 plants. Following immunoprecipitation with anti‐RFP, ubiquitination of CCD21:mRFP was remarkably diminished in Zmbpm1‐1 compared to wild‐type in the presence of ATP, coinciding with increased accumulation of CCD21:mRFP in Zmbpm1‐1 (Figure 5E). Complementary in vivo experiments were further conducted by transfection of CCD21:mRFP into protoplasts derived from OE‐Y49 (Ubi::ZmBPM1:GFP). In this context, CCD21:mRFP ubiquitination was substantially elevated in OE‐Y49 relative to Neg‐Y49, accompanied by enhanced CCD21:mRFP degradation mediated by ZmBPM1 (Figure 5F). Since ZmBPM1 also degraded CCD21 via autophagy pathway, we next investigated whether ZmBPM1 promoted the autophagic flux by detecting the free fluorescent protein (xFP) cleaved from xFP‐tagged proteins [53, 54]. Co‐expression of ZmBPM1:EGFP and TagRFP:ZmATG8d resulted in higher levels of both free TagRFP and free EGFP compared with co‐expression of ZmBPM2:EGFP and TagRFP:ZmATG8d (Figure 5G). These results suggested that ZmBPM1 promotes the autophagic flux. Furthermore, when CCD21:mRFP was transfected into protoplasts extracted from the ZmBPM1 overexpression line OE‐Y49 (Ubi::ZmBPM1:GFP), 3‐MA treatment substantially increased CCD21 and ZmBPM1 accumulation relative to DMSO control (Figure 5H), consistent with the results from N. benthamiana (Figure 5D). Consistently, ZmBPM1 partially restored CCD21‐mediated auto‐HR after 3‐MA treatment (Figure S11). ZmBPM1 Interacts with ZmATG6a/ZmATG6b, Which Also Suppress CCD21‐Mediated Auto‐HR We speculated that ZmBPM1 might associate with maize ATGs in modulating autophagy‐dependent degradation of Rp1‐D21. To validate the hypothesis, we conducted Co‐IP assays to investigate the interactions between ZmBPM1 and a few maize ATG homologs, including ZmATG5, ZmATG6a, ZmATG6b and ZmATG8d [13, 55]. Interestingly, ZmBPM1 interacted with all these ZmATGs investigated (Figure 6A). ZmBPM1 interacts with ZmATG6 to inhibit Rp1‐D21‐mediated auto‐HR. (A) The interactions between ZmBPM1 and ZmATG proteins were investigated by Co‐IP assays. Immunoprecipitation (IP) was conducted using anti‐GFP microbeads, and immunoblotting was conducted by anti‐RFP and anti‐GFP antibodies. (B) The effects of different ZmATGs on suppressing CCD21‐mediated auto‐HR in N. benthamiana. For detailed procedures, see Figure 2A–C. (C‐D) The stability of CCD21 is synergistically regulated by ZmBPM1 and ZmATG6a or ZmATG6b. (E) ZmATG6a/ZmATG6b, ZmBPM1 and CCD21 form ternary complexes. To further explore whether these ZmATGs play a role in Rp1‐D21‐mediated auto‐HR, we co‐expressed them with CCD21:TagRFP in N. benthamiana. It showed that ZmATG6a and ZmATG6b, but not ZmATG5 or ZmATG8d, inhibited CCD21:EGFP‐triggered auto‐HR and decreased the protein accumulation of CCD21: TagRFP relative to GUS (Figure 6B). Moreover, both the MATH domain and the BTB/POZ domain of ZmBPM1 also interacted with ZmATG6a/ZmATG6b as detected by Y2H (Figure S12A). ZmATG6a/ZmATG6b displayed punctate localization when they were co‐expressed with GUS control (Figure S12B,C). Interestingly, the MATH domain promoted while the BTB/POZ domain suppressed the punctate localization of ZmATG6a/ZmATG6b (Figure S12B,C). ZmATG6 Stabilizes ZmBPM1 to Degrade CCD21 in Ternary Complexes To investigate the functional relevance of the ZmBPM1–ZmATG6 interaction in attenuating CCD21‐mediated HR, ZmBPM1:Myc, EGFP:ZmATG6a/ZmATG6b and CCD21:TagRFP were co‐infiltrated in N. benthamiana. In prior experiments, CCD21:TagRFP at an OD600 of 0.4 was fully suppressed by either ZmBPM1 or ZmATG6 alone. Here, a higher infiltration density (OD600 = 0.7) of CCD21:TagRFP was employed, under which individual expression of ZmBPM1 or ZmATG6a/ZmATG6b only partially suppressed the auto‐HR (Figure 6C,D). In contrast, co‐expression of ZmBPM1 and ZmATG6 led to a marked reduction in HR severity (Figure 6C,D). Interestingly, quantitative analysis revealed that ZmBPM1 and ZmATG6a/ZmATG6b exhibited a synergistic effect on CCD21 degradation (Figure 6C,D; Figure S13). Notably, the stability of ZmBPM1 was enhanced upon co‐expression with ZmATG6a/ZmATG6b compared with GUS, while ZmATG6a/ZmATG6b protein abundance remained unchanged regardless of co‐expression with ZmBPM1 or GUS (Figure 6C,D). To determine whether ZmBPM1, ZmATG6a/ZmATG6b, and CCD21 assemble into ternary complexes, co‐IP assays were performed. ZmATG6a/ZmATG6b, but not GUS, formed complexes with both ZmBPM1 and CCD21 (Figure 6E). Collectively, these findings demonstrate that ZmBPM1 and ZmATG6a/ZmATG6b function synergistically to suppress Rp1‐D21‑triggered HR, with ZmATG6a/ZmATG6b stabilizing ZmBPM1 and promoting CCD21 degradation within ternary complexes. ZmBPM1 Negatively Regulates the Resistance to Southern Corn Rust in Maize To determine the function of ZmBPM1 in maize disease resistance, we firstly investigated the reactive oxygen species (ROS) burst in the Zmbpm1‐1 mutant triggered by chitin, and found that Zmbpm1‐1 significantly increased the ROS burst compared to WT (Figure 7A). To determine the response of ZmBPM1 to Puccinia polysora, we detected the transcript levels of ZmBPM1 via RT‐qPCR from the resistant line Qi319 and the susceptible line Ye478 infected by P. polysora at different time points [56]. It showed that ZmBPM1 was significantly up‐regulated at 24 hpi and 48 hpi in both lines (Figure S14). We further scored southern corn rust (SCR) caused by P. polysora for plants grown in the field and found that Zmbpm1‐1 conferred more resistant to SCR compared to WT siblings (Figure 7B,C). In contrast, the ZmBPM1 overexpression lines decreased the ROS burst triggered by chitin, and they were more susceptible to SCR than their negative siblings after artificial inoculation by P. polysora (Figure 7D–G). These results indicated that ZmBPM1 plays a negative role on the resistance to SCR in maize. The function of ZmBPM1 in conferring resistance to southern corn rust.(A) Reactive oxygen species (ROS) production was significantly increased in Zmbpm1‐1 compared to WT following chitin treatment. The data are presented as means ± SE (n = 8). (B‐C) Mutation of ZmBPM1 conferred increased resistance to southern corn rust (SCR) caused by P. polysora. (B) The SCR phenotype of Zmbpm1‐1 was evaluated for natural infection in the field conditions. (C) Disease symptoms on the ear leaves were assessed on a scale from 0 (resistant) to 9 (susceptible) at the VT stage in the field. Statistical analysis (means ± SE, n = 15) was conducted by Student's t‐test (***p < 0.001). (D) Overexpression of ZmBPM1 reduced the ROS production compared to their negative siblings following chitin treatment. The data are presented as means ± SE (n = 8). (E, F) The ZmBPM1 overexpression line exhibited increased susceptibility to SCR induced by P. polysora. (E) The SCR phenotype was evaluated for artificial infection in the phytotron. (F) The statistical analysis (means ± SE, n = 6) was conducted to determine the number of P. polysora spores per leaf by Student's t‐test (***p < 0.001). (G) Fungal biomass measurement of P. polysora in maize overexpression line OE‐Y49 infected at 48 hpi. Fungal biomass was normalized to maize genomic DNA via qPCR. Statistical analysis (means ± SE, n = 3) was conducted by Student's t‐test (***p < 0.001). (H–J) P. polysora is likely to promote autophagy in the ZmBPM1 overexpression line. (H) In the ZmBPM1 overexpression line, P. polysora induced a higher number of punctate structures (Scale bars: 20 µm). (I) The puncta numbers were quantified. The data are presented as means ± SE (n = 7). Student's t‐test was used to determine the significant difference between samples (***p < 0.001). (J) P. polysora induces more free EGFP on ZmBPM1 overexpression line. (K) Western blot detection of autophagy activation in OE‐Y49 plants inoculated with P. polysora. Anti‐ATG8 antibody recognizes two forms of ATG8: free ATG8 and ATG8‐phosphatidylethanolamine (PE) adducts. The ratio of normalized ATG8‐PE to normalized ATG8a was calculated to reflect the autophagic lipidation activity. Anti‐NBR1 was used to detect the protein accumulation of the autophagic cargo receptor NBR1. Ponceau S (PS) staining served as the total protein loading reference for normalization of all band intensities. The relative abundance of ATG8 and NBR1 proteins (normalized to PS) under H2O mock treatment was set as 1.00. Notably, P. polysora infection significantly induced puncta formation in the ZmBPM1:EGFP overexpression line and increased free GFP release from ZmBPM1:EGFP (Figure 7H–J). NBR1 is a well‑established autophagy receptor and substrate, and the conjugation of ATG8 to phosphatidylethanolamine (PE) to produce ATG8‑PE adducts is a key event that reflects autophagic flux [24, 57, 58]. We therefore quantified the ATG8‑PE/ATG8 ratio and found that the ratio was increased, whereas the protein accumulation of NBR1 was decreased in the ZmBPM1:EGFP overexpression line after P. polysora treatment (Figure 7K). These results indicated that P. polysora promotes ZmBPM1‐mediated autophagy in maize. Discussion BPM proteins, particularly within the core group, function as adaptors in the 26S proteasome pathway to modulate the stability of their substrates, most are transcription factors, thereby regulating diverse plant biological processes. However, the roles and mechanisms of BPMs—from both core and expanded groups—in plant immunity remain largely unexplored. While tomato SlBPM4 was recently shown to positively regulate resistance to Botrytis cinerea by degrading negative regulators of defense [59], here we demonstrate that maize ZmBPM1 (expanded group) negatively regulates both autoimmunity mediated by the NLR protein Rp1‐D21 and resistance to southern corn rust. Mechanistically, ZmBPM1 interacts with ZmATG6 to facilitate autophagy‐mediated degradation of Rp1‐D21, thereby suppressing the HR. Unlike in mammals, few studies have shown the direct interactions between autophagy components and NLRs in plants [60]. Together with our previous study [13], we showed that ZmATG6a/ZmATG6b interacted with NLR protein Rp1‐D21 and inhibited the HR. Our findings reveal a previously unrecognized, autophagy‐dependent function for a BPM protein in plant immunity, highlighting the need for further investigation into this regulatory family. BPM proteins contain an N‐terminal MATH domain and a C‐terminal BTB/POZ domain. The MATH domain, is also called the tumor necrosis factor receptor (TNFR)‐associated factor (TRAF) domain [34]. More than 70 TRAF/MATH domain‐containing proteins were identified in Arabidopsis and they were divided into five classes, including TRAF/MATH‐only proteins and BPM proteins, etc [34]. Two TRAF/MATH‐only proteins MUSE13/14 (also named as TRAF1a/TRAF1b) regulate the turnover of NLR proteins SNC1 and RPS2 likely by interacting with the E3 ligase CPR1 to form a TRAFasome, and they play negative roles in plant immunity [61]. In human and mouse cells, the BTB/POZ‐MATH protein SPOP modulates the stability of the immune transcription factor IRF1 in proteasome‐dependent pathway through its MATH domain [62]. In this study, we demonstrated that the MATH‐BTB/POZ protein ZmBPM1 modulates the protein stability of NLR protein Rp1‐D21 and Rp1‐D21‐triggered auto‐HR. ZmBPM1 interacts with itsCCD21 domain and changes its subcellular localization mainly through the MATH domain, which is structurally similar to the TRAF/MATH‐only proteins MUSE13/14 and the BTB/POZ‐MATH protein SPOP. As ZmBPM1 not only modulates the stability of the autoactive Rp1‐D21, but its parental non‐autoactive Rp1‐dp2 (Figure S10C), indicating that ZmBPM1 contributes to NLR turnover regardless of their activation state. Interestingly, ZmBPM2 also interacts with CCD21, but it neither modulates the stability of CCD21 nor inhibits CCD21‐triggered auto‐HR. These results indicated that the interactions between CCD21 and ZmBPM proteins are required but not sufficient for inhibiting HR. Notably, the MATH domain of ZmBPM2 also interacted with CCD21 and suppressed the HR (Figure S7), indicating that the functional difference between ZmBPM1 and ZmBPM2 in modulating Rp1‐D21 is not due to the difference in their MATH domains. Protein structure modelling predicted that ZmBPM1 adopts an open, loose conformation, whereas ZmBPM2 forms a highly compact conformation (Figure S8B). This compact structure, driven by the C‐terminal parts of the BTB‐POZ and BACK domains, may contribute to self‐association. The exact molecular mechanisms need to be further explored. Interestingly, ZmBPM1 not only suppresses Rp1‐D21‐mediated HR, but also partially suppresses Arabidopsis RPM1(D505V)‐mediated HR (Figure S5A). ZmBPM1 is so far the only plant BPM protein implicated in NLR regulation. The role of BPM proteins in other NLR‐mediated defense response needs to be further investigated. Most BPM proteins regulate the turnover of their substrates, which are mostly transcription factors, through 26S proteasome pathway. Strikingly, we revealed that ZmBPM1, modulates the stability of the NLR protein Rp1‐D21 and CCD21 not only through the 26S proteasome pathway but also through the autophagy pathway (Figure 5). ZmBPM1 also changed the localization of CCD21 from nucleo‐cytoplasm to punctate structures. These results are consistent with previous observation that aggregate‐prone proteins are prone to be simultaneously degraded by both pathways, especially by the autophagy pathway [63, 64, 65]. Notably, we further showed that ZmBPM1 interacts with ZmATG6, an autophagy‐related protein which also suppresses HR (Figure 6). Moreover, ZmATG6a/ZmATG6b stabilized ZmBPM1 in ternary complexes by interacting with CCD21 to regulate the homeostatic of Rp1‐D21. Consistently, the MATH‐only proteins TRAF1a/1b regulates autophagy likely by interacting with ATG6 in Arabidopsis [66]. Previously, ATG6 from Arabidopsis and N. benthamiana has been reported to be required for suppressing HR mediated by several R proteins [67, 68]. ATG6 might act as a brake to prevent the inappropriate activation of Rp1‐D21. Collectively, we demonstrated a novel mechanism that the homeostasis or activity of an NLR protein is finely modulated by the BPM1 and ATG6. NLRs are distributed across different subcellular compartments, and their proper localization is essential for optimal activity [69]. The mechanisms governing subcellular transport are crucial for dynamic spatial communication and the formation of complexes between NLRs and their interacting proteins [13, 70]. The CNL receptor PigmR confers broad‐spectrum resistance to the rice blast caused by M. oryzae [71]. It localizes to the nucleus and cell periphery, where it triggers the HR. This resistance mechanism involves a key interaction with PIBP4 (PigmR‐INTERACTING and BLAST RESISTANCE PRO‐TEIN 4), a prenylated Rab acceptor protein. PIBP4 binds both PigmR and the active Rab GTPase OsRab5a, mediating the loading of PigmR onto trafficking vesicles. These vesicles subsequently deliver PigmR to plasma membrane (PM) microdomains, where it contributes to pathogen defense [12]. Additionally, the NLR protein NRG1 localizes at several subcellular compartments, and it oligomerizes and forms punctate structures at PM once activated [72, 73]. We previously demonstrated that both the nuclear and cytoplasmic localizations are important for Rp1‐D21‐ and CCD21‐conferred HR, and alter either the distribution affects the HR phenotype [52]. We have recently elucidated that the endosome localized ZmVPS23L and ZmVPS23, the homologs of the key components in ESCRT‐I complex, interact with CCD21 and relocalize Rp1‐D21 to the endosomes to suppress HR both in N. benthamiana and maize [8]. Interestingly, ZmBPM1 also partially co‐localized with ZmVPS23L and the endosomal markers (Figures S9 and S15). We further determined that ZmBPM1 interacted with ZmVPS23L as analyzed by Co‐IP and BiFC assays (Figure S15), suggesting that they are in the same complex in modulating Rp1‐D21‐mediated auto‐HR. ZmBPM1 is primarily found in punctate structures that coincide with ATG8, a well‐established marker for autophagosomes [24]. Notably, ZmBPM1 interacts with CCD21, facilitating the translocation of CCD21 from the nucleo‐cytoplasm to structures resembling autophagosomes (Figures 2 and 3). Additionally, ZmBPM1 also enhances autophagic flux and modulates the degradation of both Rp1‐D21 and CCD21 through both the proteasome and the autophagy pathways (Figure 5D). The ubiquitin‑26S proteasome pathway mediates rapid degradation and preferentially targets soluble proteins, while autophagy acts as a selective mechanism for the degradation of large insoluble protein aggregates, organelles, or invaded pathogens [24, 65]. Our cell‐free ubiquitination assays showed that ZmBPM1 facilitates ubiquitination and proteasomal degradation of CCD21 (Figure 5E,F), which might allow rapid and flexible control of soluble CCD21 pools. Ubiquitination frequently serves as a critical targeting signal for autophagic degradation, especially for misfolded protein aggregates, damaged organelles, and large protein complexes that cannot be processed by the 26S proteasome [74]. Mechanistically, ZmBPM1 interacts with ZmATG6 to form a ternary complex with CCD21, which might direct ubiquitinated CCD21 toward autophagic degradation. Based on the aforementioned findings, we therefore propose that the ubiquitin‑26S proteasome serves as a rapid response system for basal and soluble CCD21 degradation, while autophagy acts as a complementary, sequential and inducible pathway to ensure robust CCD21 turnover. The ZmBPM1‐ ZmATG6 module acts as a molecular hub that coordinates the flux and substrate partitioning between these two degradation systems to maintain immune homeostasis. Many pathogens secret effectors and target host proteins to infect plants. Though BTB/POZ‐MATH proteins have not been identified to be targeted by pathogen effectors, however, some BTB/POZ domain‐containing proteins without MATH domain can be targeted by pathogen effectors. For example, Phytophthora infestans RXLR effector Pi02860 can target the BTB‐domain protein NRL1, which can degrade the positive immune regulator SWASP70, a guanine nucleotide exchange factor associated with endosomes [75]. Additionally, many pathogen effectors target components in autophagy pathway to facilitate their infection [76, 77]. Conversely, some effectors are also targeted for autophagy‐mediated degradation [77, 78, 79]. Though we are not sure whether ZmBPM1 or ZmATG6 can be targeted by effectors, however, we have demonstrated that ZmBPM1 acts negatively on the resistance to SCR caused by P. polysora, which also promotes autophagy flux in ZmBPM1 overexpression lines (Figure 7). These results suggest that ZmBPM1 might be targeted by effectors from P. polysora. This hypothesis is waiting for the further investigation in the future. We proposed a model to elucidate the function of ZmBPM1 and ZmATG6 in NLR‐mediated plant immunity (Figure 8). In the absence of pathogen infection, ZmBPM1 either ubiquitinates Rp1‐D21 to degrade it by 26S proteasome pathway or is engage with ZmATG6a/ZmATG6b to form preactivation complexes with Rp1‐D21 and potentially other Rp1 proteins. These proteins are primarily sequestered within autophagosomes for degradation to prevent inappropriate NLR accumulation and autoimmunity. Upon infected by pathogens, some secreted effectors may target either ZmBPM1 or ZmATG6, leading to reduce their interaction with Rp1 proteins. This could either liberate Rp1 proteins or alter the conformation of the preactivation complexes, facilitating NLR activation to trigger the HR. However, this model does not fully explain our observations that ZmBPM1 negatively regulate resistance to southern corn rust. We propose that certain pathogens may actively induce high‐level expression of ZmBPM1 or ZmATG6 as a virulence strategy. Elevated levels of these negative regulators could substantially degrade either the effectors or the NLR proteins themselves, thereby reducing or abolishing the HR. This would render plants with increased ZmBPM1 or ZmATG6 expression more susceptible to biotrophic pathogens such as those causing southern corn rust. In this context, the negative regulatory role of ZmBPM1 and ZmATG6 may represent an adaptive mechanism that restricts HR to infection sites, preventing complete plant mortality. The capacity of BPM proteins to act as negative regulators of NLR proteins like Rp1‐D21 is of particular interest, as ZmBPM1 is so far the only plant BTB/POZ‐MATH protein implicated in NLR regulation and plant immunity and it extends the growing set of BPM substrates. This study highlights the basis to further elucidate the molecular mechanisms of BPM proteins in NLR‐mediated defense response and plant immunity. The working models for elucidating the mechanisms of ZmBPM1/ZmATG6 in maize defense responses. (A) In the absence of pathogen challenge, ZmBPM1 promotes the degradation of Rp1‐D21 through two distinct pathways. It either ubiquitinates Rp1‐D21 for 26S proteasome‑mediated proteolysis, or associates with ZmATG6a/ZmATG6b to form preactivation complexes with Rp1‐D21. These complexes are predominantly routed to autophagosomes for clearance, thereby preventing untimely immune activation. (B) Upon infected by pathogens, the putative effectors may potentially target ZmBPM1 or ZmATG6, thereby releasing Rp1 proteins or altering the conformation of the preactivation complex to initiate HR. To prevent uncontrolled spread of the HR, a limited amount of Rp1 molecules are still bound by ZmBPM1/ZmATG6 for degradation. (C) Certain pathogens actively induce high expression of ZmBPM1 or ZmATG6, which can markedly degrade both effectors and NLR proteins, thereby diminishing or abolishing HR. As a result, plants with increased ZmBPM1 or ZmATG6 expression become more susceptible to pathogens, e.g. P. polysora. The model was generated using BioRender (https://BioRender.com). Experimental Section Plant Materials and Growth Conditions Nicotiana benthamiana were cultivated in a phytotron maintained at a constant temperature of 23°C, with 16 h of light and 8 h of darkness. The ZmBPM and ZmATG genes were cloned from the maize inbred line B73. Maize seedlings were grown in a phytotron at 25°C, with 12 h light and 12 h darkness. The inbred line B104 served as the donor for the generation of ZmBPM1 transgenic lines. The UniformMu mutant Zmbpm1‐1, which is in the W22 genetic background, was obtained from the maize stock center (MaizeGDB). B104:Rp1‐D21 contained the heterozygous Rp1‐D21 maintained in B104 background [8, 13, 19]. Pollen from B104:Rp1‐D21 was crossed with two ZmBPM1 overexpression T2 lines (OE‐Y12 and OE‐Y49). The resulting F1 progeny were cultivated in the field for subsequent phenotypic evaluation. The Zmbpm1‐1 mutant was backcrossed into W22 and subsequently selfed to get both the homozygous Zmbpm1‐1 and the cognate wild‐type controls. The homozygous Zmbpm1‐1 was then crossed as female to B104:Rp1‐D21. The resulting F1 plants harboring Rp1‐D21 were backcrossed to Zmbpm1‐1 to produce BC1F1 progeny for field phenotyping. Genotyping was performed using primers as indicated in Figure S4. Plasmid Construction For the agrobacterium‐mediated transient assays in N. benthamiana, constructs such as HCT1806:EGFP, GUS:EGFP, Rp1‐D21:HA, CCD21:EGFP, CCD21:TagRFP, CCD21:Myc, ZmATG6a:EGFP and ZmATG6b:EGFP were previously developed [13, 14, 21]. The ZmBPM and ZmATG genes were isolated from B73 using primers listed in Table S2 and were subsequently inserted into the pENTR directional TOPO cloning vector (Invitrogen). The constructs were then transferred through LR reactions into the destination vectors TagRFP‐pSITEII‐N1, EGFP‐pSITEII‐N1, pSITEII‐N1‐EGFP, or pSITEII‐N1‐TagRFP [13, 51]. For transient assays in maize protoplasts, constructs GUS:GFP and CCD21:mRFP were produced in our prior study [11]. The ZmBPMs and ZmATG8d were integrated into PUC19‐derived vectors that harbor either GFP or mRFP. Ubi::ZmBPM1:GFP was generated by ligation of ZmBPM1:GFP into the Spe I‐digested pZZ01523 vector containing the ubiquitin promoter [80], and used for creating the ZmBPM1:GFP overexpression lines. Phylogenetic Analysis The sequences of BPM proteins from arabidopsis and maize were aligned using Muscle of MEGA 11 software [81]. The Maximum likelihood phylogenetic analysis was performed with 1000 bootstrap replicates using the JTT+G4 model by IQ‐TREE [82, 83]. Agrobacterium‐Mediated Ectopic Transient Expression in N. benthamiana The details for Agrobacterium‐mediated transient expression in N. benthamiana were conducted in accordance with our previous research [13, 14]. Protein Analysis Protein extraction was carried out as previously described [8, 14]. The details for Co‐IP assays were conducted as previously described [13]. For immunoblotting detection of proteins fused with Myc, GFP, RFP and HA tags, the details were described as we described previously [8, 11, 13]. NBR1 was detected by anti‐NBR1 (Agrisera, Cat# AS14 2805) with a dilution of 1:5000. ATG8 was detected as described previously [84]. Protein Degradation Analyses For protein stability analyses, various constructs were agro‐infiltrated into N. benthamiana leaves. At 28 hpi, phosphate‐buffered saline with dimethyl sulfoxide (DMSO), 10 mm 3‐MA or 100 µm MG132 was further infiltrated and samples were collected for a further 6 h for immunoblotting analysis. Microsomal Fractionation Assays Microsomal fractionation assays were conducted based on our previous studies [8, 52]. Ubiquitination Assays The procedures for ubiquitination assays were carried out as previously described [13]. Cell Free Ubiquitination Assays The protocol was described previously [13, 85] with the following modifications. Briefly, proteins were extracted from wild‐type and Zmbpm1‐1 maize leaves with NB1 buffer (50 mm TRIS‐MES pH 8.0, 0.5 m sucrose, 10 mm EDTA, 1 mm MgCl2, 5 mm DTT, and protease inhibitor cocktail [CompleteMini tablets]). Cell lysates harboring CCD21:mRFP were isolated from transfected maize protoplasts with the same buffer. The two protein samples were combined at a 1:1 ratio. To preserve proteasome activity, the mixture was supplemented with 10 µm ATP and 50 µm MG132; a parallel reaction without ATP served as a negative control. After gentle rotation at 4°C for 4 h, a 30 µL aliquot was taken as the input fraction, mixed with SDS loading buffer, and denatured at 95°C for 5 min. The remaining mixture was then incubated with 15 µL of RFP‐Trap beads at 4°C for another 4 h. All subsequent steps were carried out according to the ubiquitination assay protocol described above. Split Luciferase Complementation Assays The protocol was described previously [13]. Bimolecular Fluorescence Complementation Analyses in Maize Protoplasts and in N. benthamiana The comprehensive procedures for maize mesophyll protoplasts isolating and the transfection method were carried out as previously reported [8, 11, 13, 86]. For BiFC assays in maize protoplasts, different constructs were produced by fusing different proteins into the pVYNE(R) and pVYCE(R) vectors, respectively [8, 13, 87]. Different vector combinations were co‐transfected into maize protoplasts. Confocal microscopy (LSM 900, Carl Zeiss) observation was taken at 18 hpi. For BiFC assays in N. benthamiana, different constructs were fused with split YFP‐derived YN and YC vectors, respectively [8]. Different vectors were transiently co‐infiltrated in N. benthamiana, and images were taken at 48 hpi using confocal microscopy. Confocal Microscopy The abaxial surfaces of N. benthamiana leaves were utilized for Agrobacterium‐mediated transient expression and confocal microscopy observation was conducted at 36 hpi. In the case of maize protoplasts, confocal images were observed at 18 hpi. Fluorescence detection was performed using specific excitation and observation wavelengths as previously described [13, 52]. Artificial Inoculation of Puccinia Polysora in the Growth Chamber The procedures for artificial inoculation of P. polysora were conducted as previously described [13]. Confocal microscopy observation and free GFP detection were conducted at 48 hpi by P. polysora. The disease symptoms were evaluated at 10 dpi. Quantification and Statistical Analysis Statistical analysis was conducted by GraphPad Prism and analyzed by two‐tailed Student's t‐test. Quantification of protein levels was determined by Image J software. The details for each experiment can be found in figure legends. Primers All primers used in this study are listed in Table S2. Author Contributions G.‐F. Wang conceived and designed the study. C.‐X. Tang, T.‐T Liu, Q.‐D. Ge, X.‐Y. Liu, and W. Wang performed the experiments and analyzed the data with help from Y. Sun, Y. Liu, and F. Li. G.‐F. Wang wrote the manuscript. C.‐X. Tang edited the manuscript. All authors read and approved the final manuscript. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File 1: advs77193‐sup‐0001‐SuppMat.pdf. Supporting File 2: advs77193‐sup‐0002‐SuppTables.docx.