Maize nicotinate N-methyltransferase interacts with the NLR protein Rp1-D21 and modulates the hypersensitive response.

Most plant intracellular immune receptors belong to nucleotide-binding, leucine-rich repeat (NLR) proteins. The recognition between NLRs and their corresponding pathogen effectors often triggers a hypersensitive response (HR) at the pathogen infection sites. The nicotinate N-methyltransferase (NANMT) is responsible for the conversion of nicotinate to trigonelline in plants. However, the role of NANMT in plant defence response is unknown. In this study, we demonstrated that the maize ZmNANMT, but not its close homolog ZmCOMT, an enzyme in the lignin biosynthesis pathway, suppresses the HR mediated by the autoactive NLR protein Rp1-D21 and its N-terminal coiled-coil signalling domain (CCD21 ). ZmNANMT, but not ZmCOMT, interacts with CCD21 , and they form a complex with HCT1806 and CCoAOMT2, two key enzymes in lignin biosynthesis, which can also suppress the autoactive HR mediated by Rp1-D21. ZmNANMT is mainly localized in the cytoplasm and nucleus, and either localization is important for suppressing the HR phenotype. These results lay the foundation for further elucidating the molecular mechanism of NANMTs in plant disease resistance.

INTRODUCTION

To defend against pathogen infection, plants have evolved a sophisticated immune system (Jones & Dangl, 2006). Plants employ intracellular immune receptors to directly or indirectly recognize pathogen‐secreted virulence factors (effectors) during infection (Jones et al., 2016; Wang et al., 2020). Most intracellular receptors belong to the superfamily of nucleotide‐binding, leucine‐rich repeat (NB‐LRR or NLR) receptors, which contain an N‐terminal coiled‐coil (CC) or toll/interleukin‐1 receptor (TIR) domain, a conserved central nucleotide‐binding (NB) domain, and a variable leucine‐rich repeat (LRR) domain at the C‐terminus (van der Burgh & Joosten, 2019; Wu et al., 2017). NLRs that contain the CC domain or the TIR domain are referred to as CC‐NLRs (CNLs) or TIR‐NLRs (TNLs), respectively (Bonardi et al., 2012; Meyers et al., 2003). Expression of the TIR domain or the CC domain alone can induce an autoactive hypersensitive response (auto‐HR) in plants and they are therefore considered to be the signalling domains for the induction of cell death and initiation of downstream immune responses in many NLRs (Collier et al., 2011; Maekawa et al., 2011; Swiderski et al., 2009; Wróblewski et al., 2018). NLRs are usually maintained in an autoinhibited state in the absence of pathogens, and they are activated upon effector recognition, which triggers a series of cellular defence responses (Collier & Moffett, 2009; Sun et al., 2020), including a rapid influx of calcium ions from external stores, a burst of reactive oxygen species (ROS), reprogramming of defence‐related genes expression, callose deposition in the cell wall, and a rapid localized cell death termed the hypersensitive response (HR) at the sites of attempted infection (Dangl et al., 2013; Dodds & Rathjen, 2010; Walters, 2015). Both TNLs and CNLs have been identified in dicotyledons, TNLs sometimes in more abundance than CNLs, while only CNLs have been reported in monocots (Shao et al., 2016; Wróblewski et al., 2018). Therefore, CNLs provide the core repertoire of receptors mediating intracellular immunity in monocots, and understanding the molecular mechanisms of the transition between the inhibited and activated states of CNLs is important for disease resistance breeding of monocotyledonous crops (Wang et al., 2020; Wróblewski et al., 2018).

The maize R protein Rp1‐D is a typical CNL and confers resistance to maize common rust caused by specific races of the fungus Puccinia sorghi (Collins et al., 1999; Hu et al., 1996). The encoding gene Rp1‐D is located at the Rp1 locus, in which multiple, tandemly repeated, CNL paralogs have been identified, including Rp1‐D and eight other gene homologs: Rp1‐dp1 to Rp1‐dp8 (Collins et al., 1999; Smith et al., 2004). These paralogs often mispair in meiosis and recombine unequally to create new haplotypes carrying novel gene combinations. One of them was named Rp1‐D21, which was derived from the combination of the 5′ end of Rp1‐dp2 and the 3′ end of Rp1‐D (Smith et al., 2010; Sun et al., 2001). The haplotype carrying Rp1‐D21 exhibits a spontaneous necrotic lesion mimic phenotype in the absence of pathogen infection. Transient expression of Rp1‐D21 or its CC domain (CCD21) alone in Nicotiana benthamiana triggers an auto‐HR (Smith et al., 2010; Wang, Ji, et al., 2015). The Rp1‐D21‐mediated auto‐HR phenotype is affected by temperature, light, developmental stage, and genetic background (Negeri et al., 2013; Olukolu et al., 2013). Rp1‐D21 has been used as an effective tool by crossing it into different mapping populations to identify loci and genes that were associated with modulation of auto‐HR and the defence response (Chaikam et al., 2011; Chintamanani et al., 2010; Olukolu et al., 2013; Wang, He, et al., 2015). From a nested association mapping and genome‐wide association study (NAM‐GWAS), the genes encoding homologs in the lignin biosynthesis pathway, hydroxycinnamoyltransferases (HCT1806 and HCT4918) and caffeoyl‐CoA O‐methyltransferase (CCoAOMT2), have been identified to be associated with variation in Rp1‐D21‐mediated auto‐HR (Olukolu et al., 2014). Functional analyses showed that CCoAOMT2, HCT1806, and HCT4918 associate with CCD21 and suppress Rp1‐D21‐mediated auto‐HR (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015).

In plants, lignin itself and some key enzymes involved in the lignin biosynthesis pathway have been reported to play important roles in the defence response to pathogens. Apart from the HCTs and CCoAOMT2 mentioned above, caffeic acid O‐methyltransferase (COMT) is another key enzyme involved in the lignin biosynthesis pathway and has been proved to contribute to plant resistance owing to its role in promoting lignin accumulation (Funnell‐Harris et al., 2010; Quentin et al., 2009; Wang, Zhu, et al., 2018). Simultaneous down‐regulation of COMT and methylenetetrahydrofolate reductase (MTHFR) in switchgrass caused a lesion mimic phenotype and up‐regulation of defence‐related genes (Liu et al., 2017). These results indicate that COMT may also participate in regulating programmed cell death in plants.

Nicotinate (NA) is a key intermediate of the nicotinamide adenine dinucleotide (NAD) salvage pathway, which plays an important role in maintaining the net balance of cellular NAD levels in all organisms (Gossmann et al., 2012; Wang & Pichersky, 2007). In plants, excessive accumulation of NA is toxic to plant cells; hence, NA is maintained at an appropriate concentration by converting NA into different derivatives (Li et al., 2015; Wang & Pichersky, 2007). Various conjugates of NA have been detected in plants, including glycosylation or methylation at the N‐position or carboxyl group of NA, which are catalysed by glycosyltransferases or methyltransferases, respectively (Li et al., 2015, 2017). Trigonelline (Tg), also known as N‐methyl nicotinate, is an alkaloid derived from NA methylation at the N‐position catalysed by S‐adenosyl‐l‐methionine‐dependent N‐methyltransferases and has been widely detected in various plants (Ashihara et al., 2012; Li et al., 2017; Mizuno et al., 2014). Tg is likely to function as a plant growth regulator in various biological processes (Minorsky, 2002; Zhou et al., 2012). In Arabidopsis thaliana, the nicotinate N‐methyltransferase 1 (AtNANMT1, At3g53140) was identified to be responsible for the conversion from NA to Tg and some NANMT homologs from different plant species have also been proved to have the activity by catalysing NA to Tg (Li et al., 2017). The closest homolog of AtNANMT1 in Arabidopsis is AtCOMT1, which is a key enzyme in the lignin biosynthesis pathway (Goujon et al., 2003).

In maize, the brown midrib mutant bm3 is caused by a mutation in ZmCOMT (Zm00001d049541) which has been proved to be involved in lignin biosynthesis (Vignols et al., 1995). ZM00001D051870 was previously identified as a homolog of ZmCOMT (Vélez‐Bermúdez et al., 2015). Here we found that it shows higher sequence similarity to AtNANMT1 than AtCOMT1, and thus is named ZmNANMT. COMT has been reported to play important roles in disease resistance in different plants (Liu et al., 2017; Wang, Zhu, et al., 2018). However, whether NANMT is involved in plant disease resistance, especially in the NLR‐mediated defence response, is unknown. In this study, we investigated the roles of both ZmCOMT and ZmNANMT in Rp1‐D21‐mediated auto‐HR through transient coexpression in N. benthamiana, and found that ZmNANMT, but not ZmCOMT, suppressed Rp1‐D21‐mediated auto‐HR. We further showed that ZmNANMT, but not ZmCOMT, interacted with CCD21 and also formed a complex with HCT and CCoAOMT2. We revealed the important role of ZmNANMT as a novel NLR regulator in the modulation of Rp1‐D21‐mediated defence response.

RESULTS

The effect of the key enzyme homologs in lignin biosynthesis on Rp1‐D21‐mediated auto‐HR

From our previous RNA‐Seq analysis, we found that most genes encoding key enzymes involved in the lignin biosynthesis pathway were up‐regulated in maize lines carrying Rp1‐D21 compared to wild type in both B73 × H95 and Mo17 × H95 backgrounds (Table 1). We have shown that maize HCT1806, HCT4918, and CCoAOMT2 suppress Rp1‐D21‐mediated auto‐HR when they are transiently coexpressed with Rp1‐D21 in N. benthamiana (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015). To investigate whether other enzyme homologs in the lignin biosynthesis pathway act in Rp1‐D21‐mediated auto‐HR, we cloned most of their encoding genes and coexpressed them with Rp1‐D21 in N. benthamiana. Of 16 proteins tested in this study, 14 of them showed no obvious effect on Rp1‐D21‐mediated auto‐HR, including ZmCOMT, MTHFR2, and one CAD homolog (Zm00001d015618) (Table 1). The maize brown midrib mutants bm1, bm2, and bm3 cause mutations in these three proteins, respectively (Halpin et al., 1998; Tang et al., 2014; Vignols et al., 1995). One CCR homolog (Zm00001d032152) showed a weak suppression effect on Rp1‐D21‐mediated auto‐HR and only Zm00001D051870, a close homolog of ZmCOMT, suppressed Rp1‐D21‐mediated auto‐HR (Table 1).

TABLE 1

The genes encoding enzyme homologs in the lignin biosynthesis pathway and investigation of their effects on Rp1‐D21‐mediated hypersensitive response (HR)

Gene name	Accession number	Chromosomal position	FC in B73 × H95 isogenic lines a	FC in Mo17 × H95 isogenic lines a	Rp1‐D21‐mediated HR phenotype	Reference
ZmCOMT(bm3)	Zm00001d049541	Chr4: 33,817,769–33,820,095	No read	No read	No effect	In this study
ZmNANMT	Zm00001d051870	Chr4: 172,950,400–172,953,761	1.6	1.9	Suppressed	In this study
HCT1806	Zm00001d027946	Chr1: 18,086,164–18,087,981	296.1	223.7	Suppressed	Wang, He, et al.(2015)
HCT4918	Zm00001d027948	Chr1: 18,147,157–18,148,937	1,115.0	568.5	Suppressed
HCT0436	Zm00001d030542	Chr1: 142,569,190–142,571,156	23.1	46.0	Weakly suppressed
HCT7251	Zm00001d037073	Chr6: 111,482,487–111,483,803	178.0	158.8	No effect
HCT5584	Zm00001d017186	Chr5: 188,273,104–188,277,915	6.1	4.5	No effect
HCT8083	Zm00001d003129	Chr2: 33,327,793–33,333,386	4.4	2.8	No effect
CCoAOMT1	Zm00001d036293	Chr6: 82,193,275–82,194,916	1.0	0.7	No effect	Wang & Balint‐Kurti  (2016)
CCoAOMT2	Zm00001d045206	Chr9: 16,076,158–16,081,626	2.2	1.7	Suppressed
PAL homologs	Zm00001d003015	Chr2: 29,467,931–29,470,598	4.3	13.5	No effect	In this study
	Zm00001d017276	Chr5: 191,474,696–191,476,810	10.8	13.9	No effect	In this study
	Zm00001d017274	Chr5: 191,418,711–191,422,345	4.5	2.2	No effect	In this study
	Zm00001d051163	Chr4: 146,793,740–146,798,809	17.6	10.4	No effect	In this study
CCR homolog	Zm00001d009666	Chr8: 75,117,321–75,122,209	6.6	6.9	No effect	In this study
	Zm00001d032152	Chr1: 214,573,739–214,579,407	10.9	5.3	Weakly suppressed	In this study
	Zm00001d011438	Chr8: 150,482,983–150,484,456	35.0	17.6	No effect	In this study
CAD homologs	Zm00001d015618	Chr5: 101,493,553–101,498,009	7.6	3.4	No effect	In this study
	Zm00001d045043	Chr9: 10,913,773–10,917,850	3.6	5.3	No effect	In this study
C4H homolog	Zm00001d009858	Chr8: 85,446,321–85,449,522	6.1	6.2	No effect	In this study
4Cl homologs	Zm00001d015459	Chr5: 91,459,227–91,463,857	8.6	5.6	No effect	In this study
	Zm00001d032103	Chr1: 213,126,776–213,129,233	13.1	10.5	No effect	In this study
C3H homolog	Zm00001d038555	Chr6: 159,789,930–159,794,329	16.4	6.9	No effect	In this study
MTHFR2 (bm2)	Zm00001d034602	Chr1: 297,605,177–297,611,407	4.3	4.8	No effect	Zhu et al. (2020)

The fold change (FC) of the transcript levels of maize genes in Rp1‐D21 compared to the corresponding wild type. bm, brown midrib.

Phylogenetic analysis of COMT and NANMT homologs in plants

To determine the relationship between ZmCOMT and its homolog (ZM00001D051870), the amino acid sequences of ZmCOMT and ZM00001D051870 were used as queries in a BLAST search against the plant genomics resource database (Phytozome, https://phytozome.jgi.doe.gov/pz/portal.html). Homologous sequences from different plant species were used for phylogenetic analysis. The phylogenetic tree was separated into two groups, the COMT group and the NANMT group, and each group was further divided into two clades containing COMT homologs and NANMT homologs from dicots and monocots, respectively (Figure 1). ZmCOMT was in the same clade as AtCOMT1, which was confirmed to have COMT activity (Li et al., 2017). ZM00001D051870 was in the same clade as AtNANMT1, which was confirmed to have NANMT activity (Li et al., 2017), therefore ZM00001D051870 was named ZmNANMT hereafter.

FIGURE 1

Phylogenetic analysis of COMT and NANMT homologs from diverse plant species. ZmNANMT and ZmCOMT are indicated in bold font, and AtNANMT1 and AtCOMT1 are indicated by underlines. Black triangles indicate the proteins showing NANMT enzyme activity, which was measured previously (Li et al., 2017). The tree was constructed with 1,000 bootstrap values. Ac, Aquilegia coerulea; At, Arabidopsis thaliana; Bd, Brachypodium distachyon; Gm, Glycine max; Hv, Hordeum vulgare; Nt, Nicotiana tabacum; Os, Oryza sativa; Sb, Sorghum bicolor; Sl, Solanum lycopersicum; St, Solanum tuberosum; Pv, Panicum virgatum

ZmNANMT, but not ZmCOMT, suppresses Rp1‐D21‐mediated auto‐HR in N. benthamiana

To further examine the function of ZmCOMT and ZmNANMT in Rp1‐D21‐mediated auto‐HR, we used the Agrobacterium‐mediated transient expression system in N. benthamiana, a widely used method for investigating the function of genes involved in auto‐HR (Hamel et al., 2016; Qi et al., 2012; Slootweg et al., 2018; Wang, Ji, et al., 2015; Wang, Grubb, et al., 2018). Rp1‐D21 was fused with a C‐terminal 3 × hemagglutinin (HA) tag (Rp1‐D21:HA), while ZmNANMT and ZmCOMT were fused with a C‐terminal enhanced green fluorescent protein (EGFP) tag (ZmNANMT:EGFP and ZmCOMT:EGFP). β‐glucuronidase (GUS):EGFP, and HCT1806:EGFP were used as negative and positive controls, respectively. When GUS:EGFP and Rp1‐D21:HA were transiently coexpressed in N. benthamiana, an obvious auto‐HR was observed at 3 days postinfiltration (dpi), while no obvious auto‐HR was observed in leaves coinfiltrated with HCT1806:EGFP and Rp1‐D21:HA (Figure 2a), which is consistent with our previous reports (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015). When ZmNANMT:EGFP or ZmCOMT:EGFP was transiently coexpressed with Rp1‐D21:HA in N. benthamiana, ZmNANMT:EGFP, but not ZmCOMT:EGFP, suppressed Rp1‐D21‐mediated auto‐HR (Figure 2a). To verify the visual observation results, we further measured the levels of ion leakage conductivity. Similar to HCT1806:EGFP, ZmNANMT:EGFP significantly reduced ion leakage levels compared to the GUS:EGFP control or ZmCOMT:EGFP when coexpressed with Rp1‐D21:HA (Figure 2b). Western blot results showed that coexpression of ZmNANMT:EGFP or ZmCOMT:EGFP did not affect the protein accumulation of Rp1‐D21 compared to coexpression of GUS:EGFP control, and all the EGFP‐tagged proteins were expressed at substantial levels (Figure 2c).

FIGURE 2

ZmNANMT suppressed the Rp1‐D21‐mediated auto‐hypersensitive response (HR). (a) The phenotypes resulting from transient coexpression of ZmCOMT, ZmNANMT, HCT1806, or β‐glucuronidase (GUS) with Rp1‐D21 in Nicotiana benthamiana. The representative pictures were taken at 3 days after inoculation. Two pictures show the same leaf before (left) and after (right) ethanol clearing. (b) Ion leakage conductivity was measured at 60 hr after coexpression of GUS, HCT1806, ZmCOMT, or ZmNANMT with Rp1‐D21. Significant differences (p < .05) between samples are indicated by different lowercase letters above the bars. (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinoculation (hpi). Anti‐HA was used to detect the expression of Rp1‐D21, and anti‐GFP was used to detect GUS, HCT1806, ZmCOMT, or ZmNANMT. Equal loading of protein samples was shown by Ponceau S (PS) staining of RuBisCO. The experiment was repeated three times with similar results

ZmNANMT is not a general cell death suppressor

To determine whether ZmNANMT:EGFP could inhibit auto‐HR mediated by other NLRs, we coexpressed it with two CNLs, barley MLA10(D502V) and Arabidopsis RPM1(D505V), that confer an auto‐HR when transiently expressed in N. benthamiana (Bai et al., 2012; Gao et al., 2011). Consistent with our previous results (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015), GUS:EGFP and HCT1806:EGFP had no obvious suppression roles on either MLA10(D502V)‐ or RPM1(D505V)‐mediated auto‐HR when transiently coexpressed in N. benthamiana leaves (Figure S1a,b). Similarly, neither ZmNANMT:EGFP nor ZmCOMT:EGFP suppressed the auto‐HR mediated by either CNL protein (Figure S1a,b).

INF1 is a cell death elicitor secreted from Phytophthora infestans and Bax belongs to a death‐promoting protein of the Bcl‐2 family from mouse. INF1 and Bax can trigger cell death when transiently expressed in N. benthamiana (Kamoun et al., 1998; Lacomme & Santa Cruz, 1999) and therefore were used as representatives to determine whether ZmNANMT:EGFP could suppress cell death mediated by other proteins. When coexpressed with INF1 or Bax, ZmNANMT:EGFP and ZmCOMT:EGFP showed no obvious suppression on either protein‐mediated cell death (Figure S1c,d). These results indicate that ZmNANMT might not be a general cell death suppressor.

The putative enzymatic activity of ZmNANMT might not be required for suppressing Rp1‐D21‐mediated auto‐HR

In Arabidopsis, AtNANMT1 has NA N‐methyltransferase activity, catalysing the conversion of NA to Tg. Mutation of the catalytic residue Thr at position 264 in AtNANMT1 to the corresponding His in AtCOMT to generate AtNANMT1(T264H) caused the loss of NANMT enzyme activity (Li et al., 2017). We found that the catalytic residue Thr is conserved among different plant NANMT proteins, including ZmNANMT(T278); the corresponding residue in ZmCOMT is His (Figure S2a). To investigate the effect of the predicted catalytic residue in ZmNANMT on suppressing Rp1‐D21‐mediated auto‐HR, we created the corresponding mutation to generate ZmNANMT(T278H) and fused it with a C‐terminal EGFP tag (ZmNANMT[T278H]:EGFP). When coexpressed with Rp1‐D21 in N. benthamiana, ZmNANMT(T278H):EGFP still suppressed Rp1‐D21‐mediated auto‐HR, which was similar to the suppression effect caused by ZmNANMT:EGFP (Figure 3a,b). Western blot results showed that Rp1‐D21:HA and all EGFP‐tagged proteins were expressed at substantial levels (Figure 3c).

FIGURE 3

Mutation in the conserved residue required for NANMT enzyme activity did not affect the suppressive role of ZmNANMT on the Rp1‐D21‐mediated auto‐hypersensitive response (HR). (a) The auto‐HR phenotypes resulting from transient coexpression of ZmNANMT and ZmNANMT(T278H) with Rp1‐D21. (b) Ion leakage conductivity was measured at 60 hr after coexpression of β‐glucuronidase (GUS), HCT1806, ZmNANMT, or ZmNANMT(T278H) with Rp1‐D21. Significant differences (p < .05) between samples are indicated by different lowercase letters above the bars. (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinfiltration. Anti‐HA was used to detect the expression of Rp1‐D21, and anti‐GFP was used to detect the expression of GUS, HCT1806, ZmNANMT, and its mutation. Equal loading of protein samples was shown by Ponceau S (PS) staining of RuBisCO. The experiment was repeated three times with similar results

ZmNANMT was predicted to contain three amino acid residues (N34, Y129, and H133) in the active site, which were also conserved among different plant NANMTs (Figure S2b). The protein structure of ZmNANMT was modelled using the SWISS‐MODEL platform (https://swissmodel.expasy.org/interactive) based on the structure of Sorghum bicolor COMT (SbCOMT, PDB no. 4PGH) (Green et al., 2014), which showed 35.5% sequence identity with ZmNANMT (Figure S2c). The results suggest that ZmNANMT might form a homo‐dimer with two activity centres (Figure S2d). Each activity centre contained the three predicted substrate‐binding sites (N34, Y129, and H133) and the predicted activity site (T278), which were close in space. Among these four sites, Y129, H133, and T278 belonged to chain A of the dimer while N34 belonged to chain B. To further investigate which region is required for inhibiting Rp1‐D21‐mediated auto‐HR, ZmNANMT was truncated into two segments: the 251 amino acids at the N‐terminus (ZmNANMT‐N251) and the 123 amino acids at the C‐terminus (ZmNANMT‐C123). ZmNANMT‐N251 contains three predicted substrate‐binding sites (N34, Y129, and H133) while ZmNANMT‐C123 contains the predicted enzyme catalytic site (T278). When ZmNANMT‐N251 and ZmNANMT‐C123 were coexpressed with Rp1‐D21, ZmNANMT‐C123 suppressed Rp1‐D21‐mediated auto‐HR to the same level as full‐length ZmNANMT, but ZmNANMT‐N251 showed a partial suppression effect (Figure S3a,b). Western blot results showed that Rp1‐D21:HA and all EGFP‐tagged proteins were expressed at substantial and broadly comparable levels (Figure S3c).

ZmNANMT suppresses CCD21‐mediated auto‐HR and interacts with CCD21

Rp1‐D21 contains three domains: the CC domain (CCD21), the NB domain, and the LRR domain. When they were fused with C‐terminal EGFP tags and transiently expressed in N. benthamiana, only CCD21:EGFP conferred an auto‐HR phenotype (Wang, Ji, et al., 2015). To determine the function of ZmNANMT and ZmCOMT in CCD21‐mediated auto‐HR, we transiently coexpressed them with CCD21:EGFP in N. benthamiana. The results showed that ZmNANMT, but not ZmCOMT, suppressed CCD21‐mediated auto‐HR (Figure 4a,b). All proteins were expressed at substantial levels (Figure 4c).

FIGURE 4

ZmNANMT suppressed CCD21‐mediated auto‐hypersensitive response (HR). (a) The phenotypes resulting from transient coexpression of ZmCOMT, ZmNANMT, HCT1806, or β‐glucuronidase (GUS) with CCD21. Pictures were taken at 3 days after inoculation. Two pictures show the same leaf before (left) and after (right) ethanol clearing. (b) Ion leakage conductivity was measured at 60 hr after coexpression of GUS, HCT1806, ZmCOMT, or ZmNANMT with CCD21. Significant differences (p < .05) between samples are indicated by different lowercase letters above the bars. (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinoculation (hpi). Anti‐GFP was used to detect GUS, HCT1806, ZmCOMT, ZmNANMT, and CCD21. Equal loading of protein samples was shown by Coomassie brilliant blue staining of RuBisCO. (d) ZmNANMT interacted with CCD21 as revealed by Co‐IP. EGFP‐ and 4 × c‐Myc‐tagged constructs were transiently coexpressed in Nicotiana benthamiana. Total protein was extracted from agroinfiltrated leaves at 40 hr postinfiltration for coimmuniprecipitation (Co‐IP) assay. Protein extracts were immunoprecipitated (IP) by anti‐GFP microbeads and detected (immunblotted [IB]) by anti‐GFP and anti‐c‐Myc antibodies. The experiment was repeated three times with similar results

To investigate whether ZmNANMT can interact with CCD21, we performed coimmunoprecipitation (Co‐IP) in N. benthamiana using EGFP‐tagged proteins and CCD21:Myc (with a 4 × c‐Myc tag fused at the C‐terminal of CCD21). The result showed that ZmNANMT, similar to the HCT1806 control, interacted with CCD21, while ZmCOMT and GUS did not interact with CCD21 (Figure 4d).

When ZmNANMT‐N251 and ZmNANMT‐C123 were coexpressed with CCD21:EGFP, ZmNANMT‐C123 suppressed CCD21‐mediated auto‐HR to the same level as full‐length ZmNANMT, but ZmNANMT‐N251 showed a partial suppression effect (Figure S4a,b). Western blot results showed that all EGFP‐tagged proteins were expressed at substantial levels (Figure S4c).

ZmNANMT interacts with HCT1806 and CCoAOMT2

In our previous studies (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015), HCT1806 and CCoAOMT2 interacted with CCD21 and suppressed CCD21‐ and Rp1‐D21‐mediated auto‐HR. To determine whether ZmNANMT or ZmCOMT can interact with HCT1806 or CCoAMOT2, we tested their interactions via yeast two‐hybrid (Y2H) assays. The results showed that ZmNANMT, but not ZmCOMT, interacted with both HCT1806 and CCoAMOT2 in yeast (Figure 5). Maize CCoAOMT1 has high sequence similarity to CCoAOMT2 but has no suppression role on Rp1‐D21‐mediated auto‐HR (Wang & Balint‐Kurti, 2016). We found that neither ZmNANMT nor ZmCOMT interacted with CCoAOMT1 in yeast (Figure S5). All proteins transformed into yeast were substantially expressed (Figure S6).

FIGURE 5

ZmNANMT, but not ZmCOMT, interacted with CCoAOMT2 and HCT1806 in yeast. CCoAOMT2 and HCT1806 were constructed in pGBKT7 vector fusion with GAL4 binding domain (BD). ZmCOMT and ZmNANMT were constructed in pGADT7 vector fusion with GAL4 activation domain (AD). pGBKT7‐53 and pGADT7‐T were used as positive controls, pGBKT7‐Lam and pGADT7‐T were used as negative controls. The colony was diluted and 5 μl dilution of different concentrations was dropped on the indicator plates and cultured for 3 days at 30 °C. The experiment was repeated three times with similar results

ZmNANMT, HCT1806, CCoAOMT2, and CCD21 form complexes

To test whether ZmNANMT, HCT1806, CCoAMOT2, and CCD21 could form complexes, a Co‐IP assay was performed by transiently coexpressed EGFP‐tagged ZmNANMT, ZmCOMT or GUS, HA‐tagged HCT1806, and Myc‐tagged CCoAMOT2 and CCD21 in N. benthamiana. The results showed that ZmNANMT, but not ZmCOMT or GUS, interacted strongly with all of the three proteins CCD21, CCoAMOT2, and HCT1806 (Figure 6), indicating that they can form a complex(es).

FIGURE 6

ZmNANMT, CCoAOMT2, HCT1806, and CCD21 formed a complex(es) as assayed by coimmunoprecipitation (Co‐IP). EGFP‐, 4 × c‐Myc‐ and 3 × HA‐tagged constructs were transiently coexpressed in Nicotiana benthamiana. Total protein was extracted from agroinfiltrated leaves at 40 hr postinfiltration for Co‐IP assay. Protein extracts were immunoprecipitated (IP) by anti‐GFP (α‐GFP) microbeads and detected (immunblotted [IB]) by anti‐GFP, anti‐HA (α‐HA), and anti‐c‐Myc (α‐c‐Myc) antibodies. The experiment was repeated three times with similar results

Either the cytoplasmic or the nuclear localization of ZmNANMT is important for suppressing auto‐HR

To determine the subcellular localization of ZmNANMT:EGFP or ZmCOMT:EGFP, they were transiently expressed in a transgenic N. benthamiana line carrying the stably expressed nuclear marker H2B:red fluoresecent protein (RFP) (Martin et al., 2009). It showed that ZmNANMT:EGFP, ZmCOMT:EGFP, and GUS:EGFP were all mainly located in the cytoplasm and nucleus (Figure S7).

Both Rp1‐D21 and CCD21 are located in the cytoplasm and nucleus, and nucleocytoplasm trafficking is required for auto‐HR induction (Wang & Balint‐Kurti, 2015). To investigate whether the suppression of ZmNANMT on CCD21‐mediated auto‐HR is due to a change in the subcellular localization of CCD21, we coexpressed ZmNANMT:EGFP and CCD21:RFP in N. benthamiana. Consistent with our previous results (Wang & Balint‐Kurti, 2015; Zhu et al., 2020), when transiently coexpressed with GUS:EGFP in N. benthamiana CCD21:RFP was mainly located in cytoplasm and nucleus (Figure S8). When transiently coexpressed with ZmNANMT:EGFP or ZmCOMT:EGFP in N. benthamiana CCD21:RFP was also mainly distributed in the cytoplasm and nucleus (Figure S8). The results indicated that the ZmNANMT suppression of CCD21‐mediated auto‐HR was not caused by affecting the CCD21 subcellular localization.

To determine whether the cytoplasmic or the nuclear localization of ZmNANMT is important for suppression of CCD21‐mediated auto‐HR, we added a nuclear localization signal (NLS), nuclear export signal (NES), or their corresponding mutants nls or nes (that differ from NLS and NES by one and three amino acids, respectively), to the C‐terminus of ZmNANMT and then fused them with EGFP tag to generate four proteins, ZmNANMT‐NLS:EGFP, ZmNANMT‐nls:EGFP, ZmNANMT‐NES:EGFP, and ZmNANMT‐nes:EGFP. When transiently expressed in N. benthamiana, ZmNANMT‐NLS:EGFP and ZmNANMT‐NES:EGFP were predominantly located in the nucleus and cytoplasm, respectively, while ZmNANMT‐nls:EGFP and ZmNANMT‐nes:EGFP were still localized in both the cytoplasm and nucleus, which was similar to the localization of ZmNANMT:EGFP (Figure 7a). When transiently coexpressed with CCD21:EGFP in N. benthamiana, both ZmNANMT‐NLS:EGFP and ZmNANMT‐NES:EGFP suppressed CCD21‐mediated auto‐HR (Figure 7b). Similarly, ZmNANMT‐nls:EGFP and ZmNANMT‐nes:EGFP also suppressed CCD21‐mediated auto‐HR (Figure 7b). Western blotting results showed that all EGFP‐tagged proteins were expressed at substantial levels (Figure 7c). These results indicate that both the nuclear and cytoplasmic localizations of ZmNANMT play important roles in suppression Rp1‐D21‐mediated auto‐HR.

FIGURE 7

Either the cytoplasmic or the nuclear localization of ZmNANMT is important for suppressing the CCD21–mediated auto‐hypersensitive response (HR). (a) Subcellular localization of ZmNANMT fused with nuclear localization signal (NLS) and nuclear export signal (NES). Additional signal peptide sequences (NLS, nls, NES, or nes) were added to the C‐terminus of ZmNANMT and then fused with enhanced green fluorescent protein (EGFP) tag. Different proteins were transiently expressed in transgenic Nicotiana benthamiana line carrying the stably expressed nuclear marker H2B:red fluorescent protein (RFP). Confocal images were taken at 48 hr postinoculation. The white arrow indicates the nucleus. The scale bar represents 50 µm. (b) The auto‐HR phenotypes resulting from transient coexpression of ZmNANMT, ZmNANMT‐NLS, ZmNANMT‐nls, ZmNANMT‐NES, ZmNANMT‐nes, or GUS with CCD21. The pictures were taken at 3 days after inoculation. Two pictures show the same leaf before (top) and after (bottom) ethanol clearing. (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinoculation (hpi). Anti‐GFP was used to detect the expression of different proteins. The arrow indicated the size of CCD21:EGFP. Equal loading of protein samples was shown by Coomassie brilliant blue staining of RuBisCO. The experiment was repeated three times with similar results

DISCUSSION

ZmNANMT might act as a novel NLR regulator in plant defence response

The lignin biosynthesis pathway plays important roles in plant disease resistance. There are at least 10 key enzymes involved in the lignin biosynthesis pathway in plants (Vanholme et al., 2008); however, little is known about their roles in NLR‐mediated defence response. We have shown that the maize HCT and CCoAOMT homologs, HCT1806, HCT4918, and CCoAOMT2, suppress the NLR protein Rp1‐D21‐mediated auto‐HR when they were transiently coexpressed with Rp1‐D21 in N. benthamiana (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015). However, here we found that none of the genes encoding homologs of other tested enzymes in the lignin biosynthesis pathway had an obvious effect on Rp1‐D21‐mediated auto‐HR, even though they were significantly up‐regulated in maize lines carrying Rp1‐D21 compared to the wild type (Table 1). Intriguingly, ZmNANMT, but not its close homolog ZmCOMT, suppressed Rp1‐D21‐mediated auto‐HR, and even the gene encoding ZmNANMT was only slightly up‐regulated in Rp1‐D21 (Table 1). As a key enzyme in the lignin biosynthesis pathway, COMT has been shown to play important roles in plant disease resistance (Liu et al., 2017; Wang, Zhu, et al., 2018). These data suggest that the role of COMT in disease resistance and NLR‐mediated auto‐HR might be uncoupled. Interestingly, AtCOMT1 and some of its homologs from different plants also showed weak NANMT activity, while NANMT homologs from different plants did show COMT activity. Phylogenetic and enzymatic analyses suggest that NANMT is probably derived from the duplication and subfunctionalization of an ancestral COMT gene during plant evolution (Li et al., 2017).

Tg is known to play important functions in plants and animals and it is believed to be derived from NA methylation catalysed by an N‐methyltransferase (Minorsky, 2002; Upmeier et al., 1988; Zhou et al., 2012). However, there was no direct genetic and molecular evidence for the genes responsible for this catalytic reaction until AtNANMT was identified in Arabidopsis (Li et al., 2017). NANMT1 has been reported to participate in Tg synthesis in Arabidopsis; however, there are no functional reports on other biological processes, including plant defence response. Many NLR regulators modulate the activity of NLRs through protein interaction in plants. Seven major classes of NLR regulators have been recently summarized, including transcription factors, E3 ligases, molecular chaperones, and kinases (Sun et al., 2020). In this study, we determined that ZmNANMT, a homolog of AtNANMT1 in maize, acted as a novel class of NLR regulator to negatively regulate the NLR protein Rp1‐D21‐mediated auto‐HR.

NANMT enzyme activity might not be required for suppression of Rp1‐D21‐mediated auto‐HR

In AtNANMT1, the Thr residue at position 264 is important for its catalytic activity and the mutation of AtNANMT1T264H failed to convert NA to Tg (Li et al., 2017). We performed the corresponding mutation in ZmNANMT to generate ZmNANMT(T278H) and found it had a similar suppressive effect with ZmNANMT on Rp1‐D21‐mediated auto‐HR (Figure 3a,b). Furthermore, when ZmNANMT was divided into the N‐terminal and C‐terminal parts, which contained the methyl donor‐binding sites and the catalytic site, respectively, both ZmNANMT‐N251 and ZmNANMT‐C123 suppressed Rp1‐D21‐mediated auto‐HR even though the suppression effect of ZmNANMT‐N251 was weaker than that of ZmNANMT‐C123 and the full‐length protein (Figure S3). These results suggest that the enzymatic activity of ZmNANMT is not required for its function in suppressing Rp1‐D21‐mediated auto‐HR. Interestingly, one of the breakthrough discoveries recently in the NLR field was that the TIR domains of TNLs act as NAD+‐cleaving enzymes to promote cell death (Horsefield et al., 2019; Wan et al., 2019). The TIR domains cleave NAD+ to produce nicotinamide, ADP‐ribose (ADPR), and a variant of cyclic ADPR (v‐cADPR), and the latter two can trigger calcium influx to the cytoplasm, which is important for NLR function (Wan et al., 2019; Wu et al., 1997). NAD+ can be converted to NA with nicotinamide as intermediate in plants. NANMT is a key enzyme by converting NA into Tg (Li et al., 2017). Here we showed that NANMT acts as a regulator in CNL Rp1‐D21‐mediated auto‐HR. However, the molecular mechanism of how NANMT functions is not clear. Unlike the TIR domain, no enzymatic activity has been demonstrated for the CC domain. According to cryoelectron microscopy examination of the structure of the CNL protein ZAR1 (HOPZ‐ACTIVATIED RESISTANCE 1) resistosome, the CC domain of ZAR1 can form a funnel‐like structure and insert into the plasma membrane to form a pore that might allow calcium influx to trigger auto‐HR (Wang, Hu, et al., 2019; Wang, Wang, et al., 2019). It seems that calcium influx/efflux across the plasma membrane might be a convergent event for the signalling pathways mediated by CNLs and TNLs (Saur et al., 2020; Sun et al., 2020). It will be interesting to investigate whether NANMT can affect calcium homeostasis and modulate NLR activity.

Both the cytoplasmic and nuclear localizations of ZmNANMT are important for auto‐HR suppression

The subcellular localization results showed that both ZmCOMT and ZmNANMT are located in the cytoplasm and nucleus (Figure S7), indicating that the different suppression effect on Rp1‐D21‐mediated auto‐HR is not due to their differential subcellular localizations. We previously reported that both Rp1‐D21 and CCD21 are located in the nucleus and cytoplasm, and that nucleocytoplasmic trafficking plays an important role in auto‐HR induction (Wang & Balint‐Kurti, 2015). We have recently shown that two type I maize metacaspases (ZmMCs), ZmMC1 and ZmMC2, interact with CCD21 and change its subcellular localization to suppress auto‐HR (Luan et al., 2020). When ZmNANMT was coexpressed with CCD21 in N. benthamiana, it did not cause the redistribution of CCD21 in plant cells (Figure S8). When ZmNANMT was moved to the cytoplasm or nucleus by adding NES or NLS, it still suppressed CCD21‐mediated auto‐HR (Figure 7). These results indicate that both the cytoplasmic and nuclear localization of ZmNANMT are important for suppression auto‐HR and blocking Rp1‐D21 anywhere in the nucleus or cytoplasm might destroy the nucleocytoplasmic trafficking and its ability to induce auto‐HR.

ZmNANMT may cooperate with other NLR suppressors to trap effectors in Rp1‐mediated immune response

NLR proteins can recognize pathogen effectors directly or indirectly. Several models have been proposed to explain how NLR receptors recognize different pathogen effectors, including the decoy model. According to this model, some plant proteins are employed as decoys by an NLR to selectively recognize different pathogen effectors (Cesari, 2018; Moffett, 2016). An NLR receptor can even interact with several host proteins to indirectly recognize multiple pathogen effectors from different pathogens (Sun et al., 2020), while direct recognition only allows NLRs to recognize one or a limited number of effectors. In addition, in the process of evolution, effectors can easily escape the direct recognition of receptors. Therefore, compared with direct recognition, indirect recognition through decoys by NLR receptors can provide more effective options for the plant immune system and increase the fitness costs of pathogens to avoid being recognized (Cesari, 2018). In Arabidopsis, the NLR protein ZAR1 is a typical example that can interact with and use several kinases or pseudokinases as decoys to perceive a few effectors from different pathogens (Baudin et al., 2017; Laflamme et al., 2020; Lewis et al., 2013; Martel et al., 2020; Seto et al., 2017; Wang, Roux, et al., 2015). Among them, ZAR1 and a pseudokinase RKS1 form a complex to recognize the uridylylated receptor‐like cytoplasmic kinase PBL2 (PBL2UMP) catalysed by Xanthomonas campestris effector AvrAC (Wang, Roux, et al., 2015). One of the most important recent discoveries is that the structure of the ZAR1‐RKS1‐PBL2UMP complex was characterized by cryoelectron microscopy and it showed that the activated complex forms a wheel‐like pentameric “resistosome,” which is required for the immune response (Wang, Hu, et al., 2019; Wang, Wang, et al., 2019).

In maize, HCT1806, CCoAOMT2, and ZmNANMT suppressed Rp1‐D21‐mediated auto‐HR and functioned as negative regulators in plant defence response. Rp1‐D confers disease resistance to common rust caused by the fungus P. sorghi (Collins et al., 1999). Due to the lack of the corresponding effectors of Rp1‐D from P. sorghi, we do not know whether these suppressors can act as decoys to trap effectors. From our Co‐IP and Y2H assays, we showed that ZmNANMT, HCT1806, and CCoAOMT2 interacted with CCD21 (Figure 4) (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015), and all three suppressors can interact with each other (Figure 5) and form a complex(es) with CCD21 (Figure 6). The structure of the N‐terminal CC domain of ZAR1 is similar to those of the CC domains of the potato CNL Rx and the wheat CNL Sr33, suggesting structural and functional conservation of the CC domains of CNLs across plant lineages (Saur et al., 2020). Similar to the CC domain of ZAR1, CCD21 can also trigger auto‐HR and interact with several different NLR regulators. By analogy with the ZAR1 resistosome mentioned above, we speculate that Rp1‐D21 might form a similar resistosome, which might include ZmNANMT and other regulator proteins (e.g., HCT1806 and CCoAOMT2) to recognize unknown pathogen effector(s) for activation.

In summary, we have demonstrated that ZmNANMT acts as a novel negative regulator to suppress the NLR protein Rp1‐D21‐mediated auto‐HR. To our knowledge, it has not been reported previously in any species that NANMTs can play a role in the plant defence response. Although most data were derived from the ectopic system using Agrobacterium‐mediated transient over‐expression in N. benthamiana, we believe it can largely reflect the situation in the endogenous maize system for the reasons mentioned previously (Luan et al., 2020; Wang, Ji, et al., 2015). We hope to verify these results when maize genetic materials are available in the future. Hopefully, our novel findings here will lay the foundation for elucidating the molecular mechanism of NANMTs in plant innate immunity.

EXPERIMENTAL PROCEDURES

Plant material and growth conditions

N. benthamiana wild‐type and H2B‐RFP transgenic plants were grown in the greenhouse under conditions of 24 °C with a cycle of 16 hr light and 8 hr dark. Maize (Zea mays) line B73 was grown in a plant growth chamber under conditions of 24 °C with a cycle of 12 hr light and 12 hr dark.

Sequence alignment and phylogenetic analysis

The homologs of COMT and NANMT from different plant species were downloaded in FASTA format, and then aligned by ClustalW (www.ebi.ac.uk). The phylogenetic tree was constructed using MEGA v. 6.0 software applying neighbour‐joining (NJ) algorithms with 1,000 bootstrap replicates (Tamura et al., 2013).

Genes cloning and plasmid construction

RNA was extracted from B73 seedlings at the four‐leaf stage using TRIzol reagent (Invitrogen) following the manufacturer's instructions. Reverse transcription (RT) first‐strand cDNA synthesis was performed using TransScript First‐Strand cDNA Synthesis SuperMix (Transgene). The coding sequences (CDS) of ZmNANMT and ZmCOMT were amplified from the cDNA above and cloned into the Gateway Entry vector pENTR/D‐TOPO (Invitrogen). The sequences of NLS, NES, nls, and nes were separately added to the C‐terminus of ZmANANMT CDS by PCR (Wang & Balint‐Kurti, 2015). Overlap extension PCR was performed to generate ZmNANMT CDS containing the site‐directed mutation of ZmNANMT(T278H). All of the PCR products were cloned into the pENTR/D‐TOPO vector. The primers used in PCR are listed in Table S1. After verification by sequencing, different target sequences carried in the pENTR/D‐TOPO vectors were transferred into pSITEII‐N1‐EGFP vector (Martin et al., 2009) by LR recombination reaction following the manufacturer's instructions (Invitrogen) and used for transient expression in N. benthamiana. Rp1‐D21:HA, GUS:EGFP, HCT1806:EGFP, CCD21:EGFP, CCD21:Myc, and HCT1806:HA were generated in our previous studies (Wang & Balint‐Kurti, 2016; Wang, He, et al., 2015). CCoAOMT2:Myc was generated by transferring CCoAOMT2 carried in the entry vector into the pGWB617 vector by LR reaction.

For construction of Y2H vectors, the target genes were amplified using primers with restriction enzyme sites (Table S1) by PCR. The PCR products and pGBKT7 and pGADT7 vectors (Clontech) were digested by the corresponding restriction enzymes and then ligated using DNA T4 ligase (TransGen).

Transient expression in N. benthamiana

Agrobacterium tumefaciens GV3101 (pMP90) was used for transformation of different vector constructs. The detailed procedures of Agrobacterium‐mediated transient expression in N. benthamiana were according to the previous study (Wang, Ji, et al., 2015). Briefly, the agrobacteria were separately cultured using L‐broth medium followed by resuspension and mixing according to the experiment requirements using resuspension buffer (10 mM MES pH5.6, 10 mM MgCl2, 200 μM acetosyringone). After being placed at room temperature for 3 hr, the solution was infiltrated into the abaxial side of 4‐week‐old N. benthamiana leaves using a syringe with needle removed. At least 20 individual leaves were infiltrated by each construct, and the HR phenotype is illustrated with one representative leaf for each experiment. After infiltration, the plants were placed under the same growing conditions for auto‐HR phenotype observation or protein extraction.

Ion leakage measurement

The infiltrated region of N. benthamiana leaves was sampled using a hole puncher (1.2 cm diameter) and used for ion leakage measurement. At least 12 leaf discs were sampled from different leaves for each protein and every three leaf discs were put in a 10‐ml polypropylene tube containing 4 ml of sterile water. After shaking for 3 hr at room temperature, the conductivity of different samples was measured individually using a conductivity meter (METTLER TOLEDO) and the values were recorded as C1. Subsequently, samples were boiled for 15 min and naturally cooled to room temperature. The conductivity was measured again as C2, which represented the total conductivity. The ion leakage was calculated as the C1/C2 ratio and the mean was used for final analysis (Wang & Balint‐Kurti, 2015).

Protein extraction and analysis

Protein extraction from N. benthamiana leaves, protein expression, and Co‐IP assays were performed according to previous studies (Wang, Ji, et al., 2015; Zhu et al., 2020) with minor modifications. Briefly, for protein expression detection, three leaf discs (each 1.2 cm diameter, total about 50 mg) from different leaves infiltrated by agrobacteria were collected at 30 hr postinfiltration (hpi). Total protein was extracted in 150 µl of western blotting buffer (20 mM Tris.HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 1% Triton X‐100, 0.1% SDS, 10 mM dithiothreitol [DTT], 40 µM MG132, and 1 × plant protein protease inhibitor mixture [Sigma‐Aldrich]). Subsequently, total protein was loaded to sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS‐PAGE) and analysed by western blotting. The target proteins were detected by immunoblot (IB) with α‐HA (cat. # H6908; Sigma) or α‐EGFP (cat. # A01704S; GenScript) antibodies at a 1:5,000 dilution. For the Co‐IP assay, three whole leaves infiltrated by agrobacteria were collected at 48 hpi and 0.6 g of leaf samples was used for total protein extraction in 1.8 ml of Co‐IP buffer (50 mM HEPES pH 7.5, 50 mM NaCl, 10 mM EDTA, 0.5% Triton 100, 4 mM DTT, 40 µM MG132, and 1 × plant protein protease inhibitor mixture). Subsequently, total protein was used for Co‐IP assay using the kit containing anti‐GFP microbeads (cat. # 130–091–125; Miltenyi Biotec) and the detailed procedures were performed following the manufacturer's instructions and our previous studies (Wang, Ji, et al., 2015; Zhu et al., 2020). The target proteins were detected by IB with α‐HA, α‐EGFP, or α‐Myc (cat. # A00704S; GenScript) antibodies at 1:5,000 dilution.

Yeast two‐hybrid assay

Saccharomyces cerevisiae Y2HGold (Clontech) was used for the Y2H assay. pGBKT7‐ and pGADT7‐derived constructs were transformed into Y2HGold and the Y2H experiment was performed according to the protocols provided by the manufacturer. DDO (minimal synthetic defined base with added double dropout supplement − Leu/−Trp) culture plate was used for positive transformation screening and QDO (SD base with added quadruple dropout supplement − Ade/−His/−Leu/−Trp)/X‐α‐Gal culture plate was used for protein interaction verification. Vectors pGBKT7‐53 and pGADT7‐T (Clontech) were used as positive control because the murine p53 protein (53) can interact with SV40 large T‐antigen (T) in yeast, while pGBKT7‐Lam (Clontech) and pGADT7‐T were used as negative control because human lamin C protein (Lam) cannot interact with T in yeast.

Confocal microscopy

The abaxial sides of N. benthaminana leaves infiltrated by agrobacteria were used for fluorescence observation by a confocal microscope (LSM 880; Carl Zeiss) at 48 hpi. 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.

CONFLICT OF INTEREST

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

FIGURE S1 ZmCOMT and ZmNANMT did not suppress other elicitor‐induced hypersensitive response (HR) or cell death in Nicotiana benthamiana. (a) ZmCOMT and ZmNANMT did not suppress the HR mediated by MLA10 (top) or MLA10(D502V) (bottom). (b) ZmCOMT and ZmNANMT did not suppress RPM1(D505V)‐mediated HR. (c) ZmCOMT and ZmNANMT did not suppress INF1‐mediated cell death. (d) ZmCOMT and ZmNANMT did not suppress Bax‐mediated cell death. The two pictures show the same leaf before (left) and after (right) ethanol clearing. Agrobacterium carrying MLA10, MLA10(D502V), RPM1(D505V), INF1, and Bax was diluted to a final concentration of OD600 = 0.8, 0.02, 0.1, 0.3, and 0.3, respectively, while Agrobacterium carrying EGFP‐tagged proteins construct was diluted to a final concentration of OD600 = 0.45 plus p19 with OD600 = 0.2. The pictures were taken at 2 days (MLA10(D502V), RPM1(D505V), INF1 and Bax) or 2.5 days (MLA10) after inoculation. Ion leakage conductivity was measured just after taking the pictures for MLA10, MLA10(D502V), and RPM1(D505V). The experiment was repeated three times with similar results

Click here for additional data file.

FIGURE S2 Multiple sequence alignment of plant COMT and NANMT homologs. (a) The region containing predicted enzyme activity sites in COMT and NANMT homologs. The conserved enzyme activity sites are in red boxes. (b) The regions containing precited substrate binding sites in NANMT homologs. The conserved substrate‐binding sites are in green boxes. Blue triangles indicate the proteins showing NANMT enzyme activity measured previously (Li et al., 2017). (c) The amino acid sequence alignment of ZmNANMT and SbCOMT (Sobic.007G047300). (d) Structure homology modelling of ZmNANMT was modelled based on the structure of SbCOMT (PDB no. 4PGH) (Green et al., 2014) by homology modelling (Waterhouse et al., 2018). The main chain is shown in cartoon, and the predicted substrate binding sites and activity site are labelled and shown in sticks. The C‐terminus of chain A is in magenta, the N‐termini of chain A and chain B are in green and cyan, respectively

Click here for additional data file.

FIGURE S3 The suppression effect of the N‐ and C‐termini of ZmNANMT on Rp1‐D21‐mediated auto‐HR in Nicotiana benthamiana. (a) The phenotypes resulting from transient coexpression of GUS, HCT1806, ZmNANMT, and the N‐ or C‐termini of ZmNANMT with Rp1‐D21. The pictures were taken at 3 days after inoculation. The two pictures show the same leaf before (left) and after (right) ethanol clearing. (b) Ion leakage conductivity was measured at 60 hr after coexpression of different constructs. Significant differences (p < .05) between samples are indicated by different letters (a, b, and c). (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinoculation (hpi). Anti‐HA and anti‐GFP were used to detect the expression of different proteins. Equal loading of protein samples is shown by Ponceau S (PS) staining of RuBisCO. The meaning of numbers 1–6 is shown in the annotation. The experiment was repeated three times with the similar results

Click here for additional data file.

FIGURE S4 The suppression effect of the N‐ and C‐termini of ZmNANMT on CCD21‐mediated hypersensitive response (HR) in Nicotiana benthamiana. (a) The phenotypes resulting from transient co‐expression of GUS, ZmNANMT and the N‐ or C‐termini of ZmNANMT with CCD21. The pictures were taken at 3 days after inoculation. The two pictures show the same leaf before (left) and after (right) ethanol clearing. (b) Ion leakage conductivity was measured at 60 hr after coexpression of different constructs. Significant differences (p < .05) between samples are indicated by different letters (a, b, and c). (c) Total protein was extracted from agroinfiltrated leaves at 30 hr postinoculation (hpi). Anti‐GFP was used to detect the expression of different proteins. The blue arrow indicates the size of ZmNANMT‐N251:EGFP, which is similar to the size of CCD21:EGFP. Equal loading of protein samples was shown by Coomassie brilliant blue staining of RuBisCO. The experiment was repeated three times with similar results

Click here for additional data file.

FIGURE S5 Neither ZmCOMT nor ZmNANMT interacted with CCoAOMT1 in yeast. CCoAOMT1 was constructed in pGBKT7 vector fusion with GAL4 binding domain (BD). ZmCOMT and ZmNANMT were constructed in pGADT7 vector fusion with GAL4 activation domain (AD). The colony was diluted and 5 μl dilution was dropped on the indicating plate and cultured for 3 days at 30 °C. The experiment was repeated three times with similar results

Click here for additional data file.

FIGURE S6 Detecting the protein expression in yeast. Anti‐c‐Myc and anti‐HA were used to detect the expression of different proteins constructed into pGBKT7 and pGADT7, respectively. The experiment was repeated three times with similar results

Click here for additional data file.

FIGURE S7 Subcellular localization of ZmCOMT and ZmNANMT. GUS, ZmCOMT, and ZmNANMT were fused with C‐terminal EGFP, and transiently expressed in transgenic Nicotiana benthamiana line carrying the stably expressed nuclear marker H2B:RFP. Confocal images were taken at 48 hr postinfiltration. The white arrow indicates the nucleus. The scale bar represents 50 µm. The experiment was repeated three times with the same results

Click here for additional data file.

FIGURE S8 ZmCOMT and ZmNANMT did not change the subcellular localization of CCD21. GUS, ZmCOMT, and ZmNANMT were fused with C‐terminal EGFP, and transiently coexpressed with CCD21:RFP in N. benthamiana. Confocal images were taken at 48 hr postinfiltration. The white arrow indicates the nucleus. The scale bar represents 50 µm. The experiment was repeated three times with the same results

Click here for additional data file.

TABLE S1 The primers used in this study

Click here for additional data file.