An RNAi-based binary vector for host-induced gene silencing (HIGS) of the ZmPRms gene (Figures 1A,B) was constructed using an In-fusion HD Cloning Kit (Clontech Laboratories, Inc.; Cat# 011614). Briefly, a 1115 bp Zein promoter (Shepherd and Scott, 2009), a 460 bp 5′ arm (the sense strand) and a 454 bp 3′ arm (the antisense strand) of the ZmPRms gene and a PR10 intron (Chen et al., 2010) were PCR amplified using sequence-specific primers (Supplementary Table S1) containing a 15 bp overlap at the 5′ ends. The three PCR fragments were joined together with ScaI-SpeI restriction enzyme digested pMCG1005 binary plant transformation vector (McGinnis et al., 2005) using the In-Fusion^® HD Cloning Kit to generate the vector, pMCG–ZmPRms–RNAi (Figure 1C). The PR10 intron inserted between the sense and antisense ZmPRms fragments facilitated efficient splicing and formation of hpRNA in the transgenic ZmPRms–RNAi plants.

Agrobacterium-mediated transformation of maize (inbred B104) with the pMCG–ZmPRms–RNAi vector was carried out at the Iowa State University Plant Transformation Center (Frame et al., 2011). Several independent transgenic events were regenerated for further analysis. Putative transgenic plants and their progenies were grown in moist soil mix containing three parts Scott’s 360 Metro-Mix (Scotts Company, Marysville, OH, United States) and one part perlite in 3^′′ (7.6 cm) pots. Seedlings were first grown in a growth chamber at 25°C under 16-h photoperiod (80 μmol m^-2 s^-1) for 4 weeks prior to transfer to 5 gal (0.02 cubic meter) pots in the greenhouse (27 ± 2°C).

Plants were screened by PCR using the ‘Phire Plant Direct PCR Kit’ (ThermoFisher Scientific; Cat# F160S) according to the manufacturer’s protocol. Primers used to screen putativeZmPRms–RNAi plants for the presence of the RNAi transgene cassette were 27zn945-F 5′-ccatgaagctgcctacagc-3′ and PR10_R 5′-cggaattccgtatggcaa-3′, while primers Bar_F: 5′-caccatcgtcaaccactacatcgagac-3′ and Bar_R: 5′-cagctgccagaaacccacgtcatgc-3′ (Rajasekaran et al., 2017) were used to screen empty vector transformed control plants. A 55°C annealing temperature and 30 s elongation time were used to amplify 1156 bp and 433 bp fragments that confirmed the presence of the RNAi expression cassette and the bar gene, respectively, in the transgenic plants.

An aflatoxin-producing A. flavus 70 strain expressing GFP (AF70-GFP; Rajasekaran et al., 2008) was obtained from the SRRC fungal collection (SRRC 1436; ARS-USDA, New Orleans, LA, United States). The fungal strain was grown on 2x concentrated V8 agar media (2x V8; Cary et al., 2015) for 7 days at 30°C with illumination. Conidia were harvested by flooding each plate with 20 ml of 0.02% (v/v) sterile Triton X-100 solution and gently dislodging conidia from the surface mycelia with a sterile scraper. Conidial suspensions were adjusted to 4 × 10⁶ spores/ml prior to the kernel inoculation.

Undamaged and roughly uniform size T₁ maize kernels (B104) collected from transgenic ZmPRms–RNAi and control plant lines were randomly assigned and processed according to a kernel-screening assay (KSA; Rajasekaran et al., 2013). All kernels were surface sterilized in 70% ethanol, air dried, and stored in sterile tubes. Kernels were inoculated by immersion into a 4 × 10⁶ spores/ml suspension of the AF70-GFP strain followed by stirring for 3 min. The inoculum was then drained off and the kernels were transferred to plastic caps that were placed in a tray for each transgenic line. The kernels were incubated under high RH ( > 90%) at 31°C for 7 days in the dark. The filter paper inside the tray was kept moist by adding extra water when needed during the incubation period.

AF70-GFP infected maize kernels were harvested 7 days post inoculation (dpi), longitudinally sectioned, and photographed using a stereomicroscope (Nikon SMZ25, Melville, NY, United States) equipped with a GFP filter and a camera to capture images of GFP fluorescence (excitation 485 nm, emission 528 nm). Individual kernels frozen in liquid nitrogen were homogenized using a SPEX SamplePrep (Geno/Grinder, Metuchen, NJ, United States) and the ground seed samples (∼25 mg fresh weight) were extracted in 500 μl of Sorenson’s phosphate buffer (pH 7.0). Samples were vortexed for 30 s followed by centrifugation at 9000 × g for 15 min. A 100 μl aliquot of the supernatant was carefully transferred to a 96-well plate and GFP fluorescence was measured at an excitation wavelength of 485 nm and an emission wavelength of 535 nm using a fluorometer (BioTek, Synergy4, Winooski, VT, United States) along with the appropriate control (buffer only blank).

Homogenized maize kernel tissue (∼20–70 mg) was extracted with ethyl acetate/acetone (1:1)/0.1 % formic acid (1 ml) for 24 h at room temperature. The extracts were filtered through cotton plugs and the filtrates were concentrated under nitrogen to dryness. Each extract was re-dissolved in acetonitrile (1 mg/ml), filtered through a 0.22 μm Spin-X centrifuge tube filter, and analyzed on a Waters Acquity UPLC system (40% MeOH in water, BEH C18 1.7μm, 2.1 × 50 mm column) using fluorescence detection (Ex = 365 nm, Em = 440 nm). Samples were diluted 10-fold if the aflatoxin signal saturated the detector. Analytical standards (Sigma–Aldrich, St. Louis, MO, United States) were used to identify and quantify aflatoxins: aflatoxin B1 (AFB1, retention time = 4.60 min.); aflatoxin B2 (AFB2, retention time = 3.55 min.). Aflatoxin contents were expressed in ng/mg fresh weight of homogenized maize kernels.

The data used to identify potential maize genes involved in resistance to A. flavus infection and aflatoxin contamination was obtained from a set of RNA-seq data of the A. flavus–maize interaction publically available and located at NCBI [SRP082421] (Musungu et al., 2016). Additionally, data from the PiZeam interactome (Musungu et al., 2015) was utilized to detect additional gene targets by analyzing the overlap between the networks. For mining of the RNA-seq data, a two-pronged approach was utilized with PiZeaM, a Z. mays–A. flavus interactome (Musungu et al., 2015), and a published A. flavus/Z. mays gene co-expression network (Musungu et al., 2016). First, the RNA-seq data was analyzed as described in Musungu et al. (2016) and large regulatory networks inferred. The ‘R’ statistical language was then used to mine the networks for ZmPRms. After filtering by correlation strength and overlap, ZmPRms was predicted to be regulating multiple Z. mays and A. flavus genes. The sub-network was further analyzed using the PiZeaM interactome (Musungu et al., 2015) to identify protein–protein interactions in Z. mays, potentially revealing downstream targets of ZmPRms. This approach resulted in the identification of additional targets that would have been missed if differential expression had been used as the only metric cutoff to identify partners of ZmPRms.

RNA was isolated from individual homogenized maize kernels using the ‘Spectrum^TM Plant Total RNA kit’ (Sigma–Aldrich, St. Louis, MO, United States) and cDNA was synthesized using iScript^TM cDNA synthesis kit (Bio-Rad). Quantitative RT-PCR (qRT-PCR) was performed using SYBR green I chemistry and iCycler iQ5 Multicolor real-time PCR detection system (Bio-Rad). The thermocycling conditions included a pre-incubation at 95°C for 3 min, dye activation at 95°C for 10 s, primer annealing at 55°C for 30 s, elongation at 55°C for 50 s followed by a dissociation curve between 65°C and 95°C for 30 min (with 0.5°C increments). The primers used for qRT-PCR are presented in Supplementary Table S1. Gene expression was normalized by ΔΔC_T analysis (Livak and Schmittgen, 2001) to Zea mays ribosomal structural gene GRMZM2G024838 expression (Shu et al., 2015) utilizing the gene expression analysis software package of the Bio-Rad iQ5.

A 700 bp upstream promoter region of the maize PRms gene was analyzed using the ‘PlantPAN 2.0’ web tool (Chow et al., 2016) to identify putative transcription factor–binding site motifs present in the promoter region.

For all experiments, typically three to five biological replicates were used per treatment. The data presented here are mean ± SE. Statistical significance between control and RNAi lines were determined by Student’s t-test. Significant difference between treatment and control was analyzed at ^∗∗P ≤ 0.05 and/or ^∗P ≤ 0.1 as indicated in the Figure legends.

PCR screening of genomic DNA isolated from the leaves of seven independent transgenic events of ZmPRms–RNAi plants were initially confirmed for the presence of an 1156 bp product spanning the 27 Zein promoters to the PR10 intron region of the ZmPRms–RNAi cassette (Figure 2A). Genomic DNA from both ZmPRms–RNAi and empty vector control plants amplified a region of 433 bp specific to the plant selection marker gene ‘Bar’ (Figure 2B). Based on the screening of fungal growth in the ZmPRms–RNAi silenced lines, three independent ZmPRms–RNAi lines that showed higher fungal growth were selected for further investigation to understand the impact of seed-specific ZmPRms silencing on A. flavus growth and aflatoxin production. Expression of the native ZmPRms gene in transgene positive individual seeds of RNAi lines and empty vector control plants inoculated with the AF70-GFP strain were examined by qRT-PCR. Seed-specific RNAi-mediated silencing of the native ZmPRms gene resulted in a ∼90–99% downregulation in expression of the gene (Figure 2C) in seeds of infected ZmPRms–RNAi lines compared to seeds of infected control lines. Line 1-5 showed the lowest relative expression (0.001) followed by the line 4–5 (0.104) and line 3–5 (0.100). Transgenic ZmPRms–RNAi plants were comparable to the wild type (WT) or empty vector transformed control plants with respect to plant morphology (Figure 2D).

Fungal growth in the seeds from ZmPRms–RNAi plant lines compared to control plants was qualitatively analyzed by fluorescence microscopy of longitudinally sectioned seeds (Figure 3A). Imaging of seed from the three ZmPRms–RNAi lines showed a significantly higher degree of GFP fluorescence and spread compared to seed from an empty vector control. Visual evidence of increased GFP fluorescence, primarily in the scutellar and adjacent endosperm tissue of the seed-correlated well with the data obtained from absolute quantification of GFP fluorescence. An increase of ∼250–350% in GFP fluorescence was observed in the ZmPRms–RNAi lines compared to the control with lines 1–5 showing the greatest increase followed by lines 4–5 and 3–5 (Figure 3B).

Silencing of the PRms gene in maize significantly affected aflatoxin content in the kernels of silenced lines as compared to the control. In general, the quantities of aflatoxin B1 were several folds higher than those of B2 in both control and silenced lines. A 4.6–7.4-fold increase (significant) in aflatoxin B1 content (ranging between 58 and 94 ng/mg FW) was observed in the ZmPRms–RNAi kernels (vs. control; Figure 4A) and a 4.2–6.0-fold increase (significant) in aflatoxin B2 content (ranging between 3.5 and 5 ng/mg FW) were observed in the ZmPRms–RNAi kernels (vs. control; Figure 4B).

Analysis of the RNA-seq data resulting from the A. flavus–maize pathogenic interaction has previously identified ZmPRms as a kernel PR protein possibly involved in resistance against A. flavus (Musungu et al., 2016). Further gene regulatory network analysis (in the current study) of the previously published RNA-seq data identified downstream gene targets possibly regulated by ZmPRms during the A. flavus–maize interaction (Figure 5, Table 1 and Supplementary Figures S1–S3). These included candidates associated with pathogen responses in plants, LRR and NB-ARC domain-containing gene (GRMZM2G060714), RabGAP/TBC-domain containing gene (GRMZM2G156320), Rab28 defense-related gene (GRMZM2G472236), and F-Box gene (GRMZM2G008528). Several other genes belonging to diverse biochemical functions, namely anthocyanidin 3-O-glucosyltransferase (GRMZM2G165390), adenine nucleotide alpha hydrolases-like gene (GRMZM2G151425), inositol monophosphatase (GRMZM2G036007), plasma-membrane choline transporter (GRMZM2G330453), choline transporter (GRMZM2G330453), tRNA-His guanylyltransferase (GRMZM2G158901), and an ATP-binding microtubule motor family gene (GRMZM5G878823), were also identified as ZmPRms-regulated genes. Other transcription factor-related candidates included Myb (GRMZM2G003406) and ereb44 (GRMZM5G806839). Several other candidates referred to ‘uncharacterized’ (unknown biological function; Table 1) also appeared to be regulated by ZmPRms.

Real-time expression analyses were performed on several of the aforementioned maize genes identified from the gene network analysis, that are possibly regulated by ZmPRms during the A. flavus–maize interaction (Table 1). Several disease resistance/biotic stress response–related genes, namely the LRR and NB-ARC domain-containing gene, RabGAP/TBC-domain containing gene, and F-Box gene, were downregulated in the ZmPRms–RNAi lines (Figure 6). Other genes associated with diverse biochemical functions including an adenine nucleotide alpha hydrolase-like gene, inositol monophosphatase, plasma-membrane choline transporter, and a tRNA-His guanylyltransferase were also downregulated in the ZmPRms silenced lines. Genes that were upregulated in the ZmPRms–RNAi lines included ereb44 (transcription factor), an ATP-binding microtubule motor family gene, and a Rab28 defense-related gene among which the Rab28 gene showed the highest upregulation (∼6–10-fold) in the ZmPRms silenced lines.

Based on the physiological responses of the PRms gene in maize, we analyzed the upstream promoter region to identify any stress-related motifs that might help us better understand the biological function of this gene. Analysis of the 700 bp upstream promoter region of the ZmPRms gene revealed the presence of several transcription factor–binding motif sites associated with both biotic and abiotic stress responses in plants. These include MYBPLANT, MybSANT, WRKY, ANAERO1CONSENSUS motifs, and several others. Details of the locations, consensus sequences, and physiological functions of these cis elements are presented in Table 2.