Maize ZmFNSI Homologs Interact with an NLR Protein to Modulate Hypersensitive Response.

Nucleotide binding, leucine-rich-repeat (NLR) proteins are the major class of resistance (R) proteins used by plants to defend against pathogen infection. The recognition between NLRs and their cognate pathogen effectors usually triggers a rapid localized cell death, termed the hypersensitive response (HR). Flavone synthase I (FNSI) is one of the key enzymes in the flavone biosynthesis pathway. It also displays salicylic acid (SA) 5-hydroxylase (S5H) activity. A close homolog of FNSI/S5H displays SA 3-hydroxylase (S3H) activity. Both FNSI/S5H and S3H play important roles in plant innate immunity. However, the underlying molecular mechanisms and the relationship between S5H and S3H with the NLR-mediated HR are not known in any plant species. In this study, we identified three genes encoding ZmFNSI-1, ZmFNSI-2 and ZmS3H that are significantly upregulated in a maize line carrying an autoactive NLR Rp1-D21 mutant. Functional analysis showed that ZmFNSI-1 and ZmFNSI-2, but not ZmS3H, suppressed HR conferred by Rp1-D21 and its signaling domain CCD21 when transiently expressed in N. benthamiana. ZmFNSI-1 and ZmFNSI-2 physically interacted with CCD21. Furthermore, ZmFNSI-1 and ZmFNSI-2 interacted with HCT, a key enzyme in lignin biosynthesis pathway, which can also suppress Rp1-D21-mediated HR. These results lay the foundation for the further functional analysis of the roles of FNSI in plant innate immunity.


Introduction
To protect against the invasion of different pathogens, plants have evolved multiple sophisticated strategies including pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and effector-triggered immunity (ETI) [1,2]. ETI is activated when plant disease resistance (R) genes recognize the presence of specific pathogen-secreted molecules (effectors), triggering a series of defense responses which often include a localized programmed cell death at the point of pathogen challenge known as the hypersensitive response (HR) [3][4][5]. The HR can restrict the further expansion of pathogenic microorganisms and therefore is usually regarded as a hallmark for plant disease resistance [4,6,7].
Most R-genes encode so-called NLR proteins, which contain a central nucleotide-binding (NB) domain and a C-terminal leucine-rich repeat (LRR) domain [8,9]. The NLR proteins can be further divided into two major classes according to their N-terminal domains; CC-NLRs (CNLs) contain a predicted coiled-coil (CC) domain and TIR-NLRs (TNLs) carry an N-terminal homology to the intracellular Toll-IL1 receptors (TIRs) domain [9,10]. TNLs are restricted to dicots while CNLs have been found in both monocots and dicots. and ZmFNSI-2 physically interacted with the Rp1 protein as determined by coimmunoprecipitation (Co-IP) and yeast two hybrid (Y2H) assays. Furthermore, we showed that ZmFNSI-1 and ZmFNSI-2 interacted with HCT, which can also suppress Rp1-D21-mediated HR. We discuss the importance of FNSI in modulating plant defense response and lay the foundation for the further functional analysis of the roles in plant innate immunity.

The Genes Encoding ZmFNSI and ZmS3H
Are Induced in Rp1-D21 Compared to the Corresponding Wild Type We have previously performed RNA-seq analysis to compare a near-isogenic pair of lines with and without Rp1-D21 in two hybrid backgrounds, B73 × H95 and Mo17 × H95 [21]. We demonstrated that the genes encoding HCT1806, HCT4918 and CCoAOMT2 in the lignin biosynthesis pathway were highly induced in Rp1-D21 compared to the corresponding wild type, and all of them suppressed Rp1-D21-mediated HR when transiently expressed in N. benthamiana [20,21].
To further identify the genes modulating Rp1-D21-mediated HR or the plant defense response, we selected additional 19 genes that were upregulated in plants carrying Rp1-D21 compared to their wild type counterparts and which had been otherwise implicated in plant disease resistance based on the literature for an initial screen. These genes included ZmFNSIs, ZmS3H, peptidoglycan-related genes, wall associated kinase, auxin-responsive genes, calmodulin binding proteins, etc. (Table 1). When transiently expressed in N. benthamiana, eleven of them had no obvious effect on Rp1-D21-mediated HR, two of them enhanced Rp1-D21-mediated HR and three of them showed autoactive HR, and we therefore concentrated on three genes related to ZmFNSI/ZmS3H that gave the most promising results. FNSI displays both FNSI activity by catalyzing the conversion of flavanones into flavones and S5H activity by hydroxylating SA into 2,5-DHBA, while S3H is able to hydroxylate SA into 2,3-DHBA ( Figure 1). ZmFNSI-1 and ZmFNSI-2 physically interacted with the Rp1 protein as determined by coimmunoprecipitation (Co-IP) and yeast two hybrid (Y2H) assays. Furthermore, we showed that ZmFNSI-1 and ZmFNSI-2 interacted with HCT, which can also suppress Rp1-D21-mediated HR. We discuss the importance of FNSI in modulating plant defense response and lay the foundation for the further functional analysis of the roles in plant innate immunity.

The Genes Encoding ZmFNSI and ZmS3H Are Induced in Rp1-D21 Compared to the Corresponding Wild Type
We have previously performed RNA-seq analysis to compare a near-isogenic pair of lines with and without Rp1-D21 in two hybrid backgrounds, B73 × H95 and Mo17 × H95 [21]. We demonstrated that the genes encoding HCT1806, HCT4918 and CCoAOMT2 in the lignin biosynthesis pathway were highly induced in Rp1-D21 compared to the corresponding wild type, and all of them suppressed Rp1-D21-mediated HR when transiently expressed in N. benthamiana [20,21].
To further identify the genes modulating Rp1-D21-mediated HR or the plant defense response, we selected additional 19 genes that were upregulated in plants carrying Rp1-D21 compared to their wild type counterparts and which had been otherwise implicated in plant disease resistance based on the literature for an initial screen. These genes included ZmFNSIs, ZmS3H, peptidoglycan-related genes, wall associated kinase, auxin-responsive genes, calmodulin binding proteins, etc. (Table 1). When transiently expressed in N. benthamiana, eleven of them had no obvious effect on Rp1-D21mediated HR, two of them enhanced Rp1-D21-mediated HR and three of them showed autoactive HR, and we therefore concentrated on three genes related to ZmFNSI/ZmS3H that gave the most promising results. FNSI displays both FNSI activity by catalyzing the conversion of flavanones into flavones and S5H activity by hydroxylating SA into 2,5-DHBA, while S3H is able to hydroxylate SA into 2,3-DHBA ( Figure 1). Figure 1. The phenylpropanoid pathway, modified according to previous studies [21,28] The Arabidopsis genome contains one AtDMR6/AtS5H/AtFNSI and two DMR6-like oxygenases, AtDLO1 (AtS3H) and AtDLO2. Using AtDMR6 and AtS3H as queries to blast the maize genome database (MaizeGDB), we identified two ZmFNSI homologs, ZmFNSI-1 (Zm00001d029744) and ZmFNSI-2 (Zm00001d027423), and one ZmS3H homolog (Zm00001d002564) in maize (Table 1). Phylogenetic analysis showed that FNSIs and S3Hs from different plant species are divided into two groups, and each group was further separated into two clades containing FNSIs or S3Hs from dicots and monocots ( Figure 2

ZmFNSI-1 and ZmFNSI-2 have no Obvious Suppression Roles on Other Elicitor-Induced HR
To determine whether ZmFNSI-1 and ZmFNSI-2 could suppress HR mediated by other NLR proteins, we co-expressed them with barley MLA10(D502V) and Arabidopsis RPM1(D505V), two CNL proteins conferring an autoactive HR when transiently expressed in N. benthamiana [39,40]. The results showed that ZmFNSI-1 and ZmFNSI-2 have no obvious suppression roles on either Ion leakage conductivity (average ± standard error (SE), n > 5) was measured at 61 h after co-expression of GUS, HCT, ZmFNSIs or ZmS3H with Rp1-D21. Significant differences (p < 0.05) between samples are indicated by different letters (a-b). The protocol was measured according to our previous study [38]. (C) Total protein was extracted from agro-infiltrated leaves at 30 hpi. Anti-HA was used to detect the expression of Rp1-D21, and anti-GFP was used to detect the expression of GUS, HCT, ZmFNSIs and ZmS3H. Equal loading of protein samples was shown by Coomassie brilliant blue (CBB) staining of Rubisco.
INF1 is a cell death elicitor from Phytophthora infestans and Bax is a death-promoting protein of the Bcl-2 family from mouse, both of which can induce cell death when transiently expressed in N. benthamiana [41,42]. When co-expressed with Bax or INF1, ZmFNSI-1 and ZmFNSI-2 did not suppress Bax-or INF1-induced cell death, which is similar to the co-expression of ZmS3H or HCT1806 with Bax or INF1, respectively ( Figure S1B).

The Enzymatic Activity of ZmFNSI-1 and ZmFNSI-2 Might Not Be Required for Suppressing Rp1-D21-Mediated HR
In Arabidopsis AtS5H/AtDMR6, the histidine (H) at positions 212 and 269 is the catalytic residue for binding the ferrous iron atom and is critical for its FNSI enzyme activity [36]. The substitution proteins AtDMR6(H212Q) and AtDMR6(H269D) cannot restore the susceptibility of Atdmr6 to H. arabidopsisdis, indicating that the enzyme activity is required for the function of AtDMR6 [36]. Based on the sequence alignment, we found that the corresponding histidines (H211 and H268 in ZmFNSI-1 and ZmFNSI-2, respectively) are conserved among different plant FNSIs ( Figure 4A). To investigate whether the enzymatic activity of ZmFNSI-1 and ZmFNSI-2 is important for inhibiting Rp1-D21-mediated HR, we performed the corresponding mutations in ZmFNSI-1 and ZmFNSI-2 to generate ZmFNSI-1(H211Q), ZmFNSI-2(H211Q), ZmFNSI-1(H268D), and ZmFNSI-2(H268D). When co-expressed with Rp1-D21, all the mutants still suppressed Rp1-D21-mediated HR ( Figure 4B). All the HA-and EGFP-tagged proteins were expressed at substantial and broadly comparable levels ( Figure 4B). These data suggest that the enzymatic activity of ZmFNSI-1 and ZmFNSI-2 is not required for its function in suppressing Rp1-D21-mediated HR.

ZmFNSI-1 and ZmFNSI-2 Suppress CC D21 -Mediated HR and Interact with CC D21
The Rp1-D21 protein can be sub-divided into three major domains: an N-terminal coiled-coil (CC) domain termed CC D21 , a middle NB-ARC (APAF1, certain R gene products and CED-4) domain and a C-terminal LRR domain [38]. We have shown previously that CC D21 , but not NB or LRR, conferred autoactive HR when fused with EGFP and expressed transiently in N. benthamiana [38]. To determine whether ZmFNSI-1 and ZmFNSI-2 can suppress CC D21 -mediated HR, we co-expressed them and CC D21 in N. benthamiana, and found that ZmFNSI-1 and ZmFNSI-2, but not ZmS3H, suppressed CC D21 -mediated HR ( Figure 5A). Ion leakage analysis further verified the visual observations: co-expression of ZmFNSI-1 and ZmFNSI-2, but not of ZmS3H, with CC D21 reduced ion leakage due to CC D21 -mediated HR ( Figure 5B).  conferred autoactive HR when fused with EGFP and expressed transiently in N. benthamiana [38]. To determine whether ZmFNSI-1 and ZmFNSI-2 can suppress CCD21-mediated HR, we co-expressed them and CCD21 in N. benthamiana, and found that ZmFNSI-1 and ZmFNSI-2, but not ZmS3H, suppressed CCD21-mediated HR ( Figure 5A). Ion leakage analysis further verified the visual observations: co-expression of ZmFNSI-1 and ZmFNSI-2, but not of ZmS3H, with CCD21 reduced ion leakage due to CCD21-mediated HR ( Figure 5B). We previously showed that HCT1806 interacts with CCD21 and suppresses CCD21-mediated HR [20]. To investigate whether ZmFNSIs can interact with CCD21, we used yeast two hybridization (Y2H) to show that ZmFNSI-1 and ZmFNSI-2 physically associated with CCD21 at 72 h after co-expression, while ZmS3H showed weak or no interaction with CCD21 ( Figure 6A). We further tested the interaction via co-immunoprecipitation (Co-IP) and found that, like HCT1806, ZmFNSI-1 and ZmFNSI-2 interacted strongly with CCD21, while ZmS3H interacted weakly with CCD21 ( Figure 6B), We previously showed that HCT1806 interacts with CC D21 and suppresses CC D21 -mediated HR [20]. To investigate whether ZmFNSIs can interact with CC D21 , we used yeast two hybridization (Y2H) to show that ZmFNSI-1 and ZmFNSI-2 physically associated with CC D21 at 72 h after co-expression, while ZmS3H showed weak or no interaction with CC D21 (Figure 6A). We further tested the interaction via co-immunoprecipitation (Co-IP) and found that, like HCT1806, ZmFNSI-1 and ZmFNSI-2 interacted strongly with CC D21 , while ZmS3H interacted weakly with CC D21 (Figure 6B), indicating that the interactions between ZmFNSI-1/ ZmFNSI-2 and CC D21 are stronger than the interaction between ZmS3H and CC D21 .
indicating that the interactions between ZmFNSI-1/ ZmFNSI-2 and CCD21 are stronger than the interaction between ZmS3H and CCD21.
Previously we showed that Rp1-D21 and CCD21 localized to both the cytoplasm and nucleus, and that nucleo-cytoplasm trafficking was important for HR induction [44]. To determine whether the different suppression effects between ZmFNSI-1, ZmFNSI-2 and ZmS3H were due to their different
Previously we showed that Rp1-D21 and CC D21 localized to both the cytoplasm and nucleus, and that nucleo-cytoplasm trafficking was important for HR induction [44]. To determine whether the different suppression effects between ZmFNSI-1, ZmFNSI-2 and ZmS3H were due to their different effects on changing the subcellular localization of CC D21 , we co-expressed ZmFNSI-1, ZmFNSI-2 and ZmS3H with CC D21 :RFP in N. benthamiana. When CC D21 :RFP was co-expressed with GUS:EGFP, it mainly localized in cytoplasm and nucleus, which is consistent with previous results [44]. Similarly, when co-expressed with ZmFNSI-1, ZmFNSI-2 and ZmS3H, CC D21 :RFP also mainly localized in the cytoplasm and nucleus ( Figure S5). The results suggest that the suppression effect of ZmFNSI-1 and ZmFNSI-2 on CC D21 -mediated HR was not due to change the subcellular localization of CC D21 .

ZmFNSI Functions in Plant Defense Response
FNSI is involved in the biosynthesis of secondary metabolites [29,30]. ZmFNSI-1 was reported to have FNSI activity which can catalyze the conversion of flavanones into flavones [28]. AtDMR6, the Arabidopsis homologous enzyme to ZmFNSI, also has FNSI activity [28]. The loss-of-function mutant Atdmr6 confers resistance to multiple pathogens [29,36], and overexpression of ZmFNSI-1 in drm6-1 mutant plants results in reduced resistance to P. syringae [28], indicating that the function of FNSI/DMR6 in Arabidopsis and maize is conserved in plant defense response. CRISPR-Cas9 mediated knockout of the tomato ortholog SlDMR6-1 confers resistance against different pathogens, including P. syringae, Xanthomonas spp. and P. capsici [45]. These results indicate that FNSI/DMR6 from different plant species acts as a negative regulator in the plant defense response.
We had previously found that many genes predicted to be involved in secondary metabolism were induced by Rp1-D21, including the phenylpropanoid pathway [20,21]. The CCoAOMT2 and HCT1806 proteins were predicted to catalyze consecutive steps in the lignin biosynthesis pathway and have been implicated to play important roles in Rp1-D21-mediated HR [21]. CCoAOMT2 has also been shown to confer resistance to southern leaf blight and gray leaf spot in maize [46]. From previous transcriptional data, ZmFNSI-1 and HCT1806 were highly induced in maize lines inoculated by several fungal pathogens which cause ear rot, including Fusarium graminearum, Fusarium verticillioides and Aspergillus flavus [27,[47][48][49][50]. ZmFNSI-1, ZmS3H, CCoAOMT2 and HCT1806 are also upregulated in maize lines expressing WtsE from Pnss [51], suggesting these genes are also involved in maize disease resistance. Several examples have shown that negative regulators of plant immunity are induced during the defense response to control and prevent the over-activation of the defense response, e.g., Nudix hydrolase-encoding NUDT7 [52], HCT1806 and CCoAOMT2 [20,21]. Here, we found that the transcriptional levels of the genes encoding ZmFNSI-1, ZmFNSI-2, HCT1806 and CCoAOMT2 were significantly induced in the Rp1-D21 mutant. ZmFNSI-1 and ZmFNSI-2 acted as negative regulators to suppress Rp1-D21-mediated HR and they interacted with the CC signaling domain of Rp1-D21, suggesting that ZmFNSI/ZmS5H regulates Rp1-D21-mediated HR, likely through physical interaction. We speculate that the increased levels of ZmFNSI-1 and ZmFNSI-2 can serve as a reservoir to inhibit the further activation of Rp1 protein and the spread of HR.
Tricin, a flavone constituted by modified apigenins, is a structural monomer of lignin in maize [53]. Falcone Ferreyra et al. (2015) suggested that ZmFNSI-1 might have a role in lignin biosynthesis. In this study, we performed Y2H assays and found that ZmFNSI-1 and ZmFNSI-2 interacted with HCT, the key enzyme in the lignin biosynthesis pathway. This interaction, together with the interaction between HCT and CCoAOMT2 [21], may reflect the formation of a metabolon, a complex formed between sequential enzymes of a metabolic pathway believed to facilitate metabolite processing [54]. Components of the phenylpropanoid pathway have been shown to associate with each other in poplar and other species [55].

FNSI Enzyme Activity Is not Required for HR and ZmFNSIs Are not General Cell Death Suppressors
FNS1 belongs to the 2OGD superfamily members. In Arabidopsis, AtDMR6 acts as a negative regulator in plant immunity. Expression of AtDMR6(H212Q) and AtDMR6(H269D), the substitution proteins in the histidine (H) residues required for the putative catalytic activity of AtDMR6, fail to restore the susceptibility of Atdmr6 to H. arabidopsidis, suggesting that the oxygenase enzymatic activity of AtDMR6 is required for its function in disease resistance [36]. For the maize ZmFNS1-1 and ZmFNS1-2, however, the corresponding mutation in ZmS5H-1(H211Q), ZmS5H-2(H211Q), ZmS5H-1(H268D), and ZmS5H-2(H268D) still suppress Rp1-D21-mediated HR (Figure 4), suggesting that the enzyme activity is not required for the suppression effect. It is also possible that the substitution mutants retained low enzymatic activity which is sufficient to suppress HR. To exclude the possibility, the enzymatic activity for these mutants will be further determined. As mentioned above, ZmFNSI is induced by multiple pathogens. ZmFNSI-1 is located in a quantitative trait loci (QTL) conferring resistance to southern leaf blight [56], and it is also located at approximately 380 kb upstream of a SNP (s-ingle nucleotide polymorphism) which is associated with multiple disease resistance [57]. These results suggest that ZmFNSI might be also involved in maize disease resistance. It is possible that the differential enzyme activity requirement of FNS1/DMR6 in disease resistance and HR might be uncoupled. AtFNS1/AtDMR6 in Arabidopsis also has S5H activity [36]; it will be interesting to investigate whether ZmFNSI-1 and ZmFNSI-2 have S5H activity and whether the S5H activity is involved in Rp1-D21-mediated HR.
ZmFNSI-1 and ZmFNSI-2 suppressed Rp1-D21-and CC D21 -mediated HR, however, they did not substantially suppress the HR phenotype caused by the other two CNL autoactive mutants RPM1(D505V) and MLA10(D502V), and they have no obvious effect on Bax-and INF1-mediated cell death. These data suggested that ZmFNSI-1 and ZmFNSI-2 might not be general HR suppressors.

FNSI/S5H and S3H have Overlapping and Different Roles in Plant Defense Response
Plant resistance to biotrophic/hemibiotrophic pathogens is largely controlled by SA-mediated signaling pathways [58]. The compounds 2,5-DHBA and 2,3-DHBA are two major catabolic products of SA and they are catalyzed by hydroxylating SA via S5H and S3H, respectively [35,37]. In Arabidopsis, AtDMR6/AtS5H was reported to have high S5H activity and low FNSI activity [35]. Therefore, the enhanced disease resistance of Atdmr6 is most likely caused by increased total SA levels [35]. After Pst DC3000 treatment, the transcript levels of AtDMR6/AtS5H and AtS3H are induced, and the accumulation of 2,5-DHBA and 2,3-DHBA is enhanced, implying that SA hydroxylation plays a role in the detoxification of excessive SA [35]. AtDMR6/AtS5H is more sensitive to SA treatment and pathogen induction than AtS3H and the Ats5h mutant exhibited much stronger pathogen resistance than the Ats3h mutant [29, 35,36]. The SA levels are higher in the Atdmr6 mutant compared to wild type, and there is no significant different between the dlo1 mutant and wild type. However, in the dmr6dlo1 double mutant, the SA levels are approximately 20 times higher than that in the Atdmr6 mutant [36]. Therefore, AtDMR6/AtS5H and AtS3H have overlapping and distinct functions as negative regulators in innate immunity. Here, we found that the fold change of the transcript levels of ZmFNSI-1 and ZmFNSI-2 is higher than that of ZmS3H in the Rp1-D21 mutant compared to the corresponding wild type backgrounds (Table 1), and ZmFNSI-1 and ZmFNSI-2, but not ZmS3H, suppress Rp1-D21and CC D21 -mediated HR, indicating the different roles of ZmFNSI and ZmS3H in the maize defense response. We also found ZmFNSI-1 and ZmFNSI-2 formed stronger homomers than ZmS3H ( Figure S3), however, it needs to be further investigated whether the formation of homomers is related with the different roles of ZmFNSIs and ZmS3H in Rp1-D21-mediated HR.

ZmFNSI Regulates Rp1-D21-mediated HR possibly through Forming a Protein Complex
Plant NLR genes are usually expressed at low levels in living cells. The over-accumulation of NLRs can lead to autoimmune responses, causing dwarfism and an autoactive HR phenotype [59]. To reduce the pleiotropic effects to plant growth and development, NLR proteins are usually kept in an inactive state by a delicate combination of intra-molecular and inter-molecular interactions. Several different categories of host proteins (including transcription factors, kinases, E3 ligases, etc.) physically interact with NLRs to modulate plant innate immunity and some NLRs can interact with different host proteins targeted by different pathogen effectors [60]. For instance, in Arabidopsis, the NLR protein ZAR1 (HOPZ-ACTIVATED RESISTANCE1) interacts with several kinases or pseudokinases to recognize effectors secreted from different pathogens [61][62][63]. Recent cryo-electron microscopy studies have characterized the structures of ZAR1 complexes in different states; the inactive ZAR1-RKS1, the intermediate ZAR1-RKS1-PBL2 UMP and the activated ZAR1-RKS1-PBL2 UMP complexes [64,65]. A wheel-like pentameric ZAR1 "resistosome" is formed in the activated state that is believed to directly induce the defense response, including HR [65].
Rp1-D confers resistance to common rust caused by P. sorghi [66]. However, the cognate effector of Rp1-D from P. sorghi is still unknown. In our previous studies, we derived models to elucidate the transition between the resting and activation states of Rp1 proteins [20,21,38]. We proposed that in the resting state, Rp1 proteins are maintained by intramolecular interactions, some of which we defined, and also by interaction with other host proteins which acted as negative regulators, including HCT and CCoAOMT2. We now propose that ZmFNSI-1 and ZmFNSI-2 are also involved in the negative regulation of Rp1-D. In the activated state, we speculate that unknown effectors from a pathogen can target one of the components from HCT, CCoAOMT2 or ZmFNSIs and disrupt their association with Rp1 proteins, leading to the activation of Rp1. We have also previously shown that Rp1 proteins can form homomers [38]. By analogy with the ZAR1 resistosome, we speculate that Rp1-D, ZmFNSIs, HCT or CCoAOMT2 might also form a 'Rp1-D resistosome' to function in the maize defense response.
In summary, we have demonstrated that ZmFNSI-1 and ZmFNSI-2, two homologs of enzymes which catalyze key steps in the flavone biosynthesis and SA catabolism pathway, interact with the CC domain of Rp1-D21 to regulate Rp1-D21 -mediated HR. ZmFNSI-1 and ZmFNSI-2 also interact with HCT. To our knowledge, this is the first evidence that the homologs of enzymes in the flavonoid and lignin biosynthesis pathways may function together to regulate the NLR-mediated defense response. Though substantial data were derived from agrobacteria-mediated transient expression in N. benthamiana, we believe that it can largely reflect the situation in maize for the following reasons. Firstly, this system has been widely employed for exploring the HR phenotype mediated by NLRs from diverse plant species [39,[67][68][69][70]. Secondly, we have used this system to verify the suppression roles of HCT1806, HCT4918 and CCoAOMT2 in Rp1-D21-mediated HR [20,21]. Thirdly, we have also used this system to verify the HR conferred by 12 EMS maize mutants derived from Rp1-D21 and it can recapitulate the HR phenotype in maize [38]. With the maize genetic materials available in the future, we hope to further verify the results in this study.

Plant Materials and Growth Condition
Maize (Zea mays) line B73 was grown at 24 • C with 12 h light/12 h dark conditions and used for isolating ZmFNSI-1, ZmFNSI-2 and ZmS3H. Wild type Nicotiana benthamiana and the transgenic N. benthamiana harboring histone 2B (H2B)-TaqRFP were grown at 23 • C with a cycle of 16 h light and 8 h dark.

Sequence Alignment and Phylogenetic Analysis
For phylogenetic analysis, the protein sequences from the FNSI and S3H families were aligned using Clustal X v2.1 [71]. Based on this alignment, a neighbor-joining tree was constructed using MEGA 6.0 software with 1000 bootstrap replicates [72].

Agrobacterium tumefaciens-Mediated Transient Expression
A. tumefaciens strain GV3101 (pMP90) transformed with binary vector constructs was grown at 28 • C overnight in 8 mL of L-broth medium supplemented with appropriate antibiotics. The detailed procedures were performed according to our previous study [38]. Unless otherwise indicated, all the experiments were repeated three times with similar results.

Y2H
ZmFNSI-1, ZmFNSI-2 and ZmS3H were constructed into the pGADT7 (AD) cloning vector (Clontech, Mountain View, CA, USA) by LR reactions, and CC D21 were constructed into the pGBKT7 (BD) vector (Clontech, Mountain View, CA, USA). AD-and BD-derived constructs were co-transformed into competent cells of the yeast strain Y2HGold. The Y2H interaction assay was performed according to the protocol provided by the manufacturer (Clontech, Mountain View, CA, USA).

Confocal Microscopy
Confocal microscopy observation was performed according to our previous study [44] with some modifications. After agrobacteria infiltration, the abaxial sides of N. benthaminana leaves were used for observation at 48 hpi by a confocal microscope (LSM 700, Carl Zeiss, Jena, Germany) with a PMT detector. The representative areas were photographed under a 20× objective. EGFP fluorescence was excited at 488 nm and observed between 495 and 550 nm. RFP was excited at 561 nm and observed between 580 and 675 nm.

Accession Number
Sequence data from this article can be found in the EMBL/GenBank data libraries under the following accession numbers. Nucleotide sequence and protein sequence of Rp1-D21 is KF951062 and AIW65617, respectively. RNA-seq data: SRP060286.