Fungal strains and plasmids used in this study are listed in Supplementary Table S1. The A. flavus strain PTSΔku70ΔpyrG (Chang et al., 2010), a uracil auxotroph, was purchased from the Fungal Genetics Stock Center (School of Biological Sciences, University of Missouri, Kansas City, MO, USA) and used for gene deletion (Yang et al., 2016). Wild-type (WT) A. flavus and the transformants generated in this study were grown on yeast extract–sucrose (YES) agar, yeast extract–glucose agar, or yeast extract–glucose agar with trace elements, uracil, and uridine (Yang et al., 2016). YES medium, potato dextrose agar (PDA, Bai and Shaner, 1996), Czapek agar (CA, Becton Dickinson), and glucose minimal medium (GMM, Shimizu and Keller, 2001) were used for mycelial growth assays, sporulation analysis, and AF analysis. All experiments included four replicate plates and were performed at least three times with similar results, and the error was expressed as the standard deviation.

The NmrA sequence (EED47920) from A. flavus was originally identified in the NCBI Protein database using BLAST. To verify the existence and the size of the exon in nmrA, we cloned the coding sequence into the pET-28a(+) expression vector and sequenced it. The obtained sequence was compared with that reported in the NCBI database. The protein domains were analyzed using InterPro¹. A phylogenetic tree based on NmrA sequences from A. clavatus (UniProt: A1C544), A. flavus (UniProt: B8NR87), A. fumigatus (UniProt: Q4WEG4), A. nidulans (UniProt: Q5AU62), A. niger (G3XN01), A. oryzae (GenBank: XP_001822655.2), and A. terreus (UniProt: Q0CAL7) was constructed with DNAMAN 6.0 software using the neighbor-joining method.

To create an nmrA deletion mutant, we constructed an AP-pyrG-BP vector by inserting upstream and downstream flanking sequences of the nmrA gene on either side of the pyrG gene. The upstream and downstream flanking sequences of nmrA were amplified from genomic DNA of A. flavus WT with primer pairs nmrA-P1/nmrA-P2 and nmrA-P3/nmrA-P4, respectively (Supplementary Table S2). The nmrA deletion mutant (ΔnmrA) was complemented with a full-length nmrA gene, a 2869 bp fragment, spanning from 1069 bp upstream of the WT A. flavus nmrA translation initiation codon to 681 bp downstream of the translation termination codon, to confirm that the phenotype of the ΔnmrA mutant was due to deletion of the nmrA gene. The full-length nmrA gene was amplified from genomic DNA of A. flavus WT using the primer pair nmrA-CF/nmrA-CR (Supplementary Table S2), and it contained a constitutive promoter, which controlled nmrA gene expression. The complementation plasmid (pPTR I-nmrA) was constructed using the backbone of the chromosomal integrating shuttle vector pPTR I DNA (Takara, Japan, Kubodera et al., 2000), which contained the same restriction enzyme recognition sites as the full-length nmrA gene and could integrate into A. flavus genome randomly. Six out of 15 pyrithiamine resistance transformants were selected for their wild-type growth phenotype in the presence of GMM supplementing ammonium as the sole nitrogen source. We concluded that these transformants had integrated an intact copy of the A. flavus nmrA gene into the genome after the A. flavus nmrA gene in this plasmid was sequenced to ensure flawlessness of the sequence.

Mycelial growth assays were performed on YES medium, PDA, and GMM supplemented with 50 mM glutamine, ammonium, proline, alanine, or sodium nitrite (NaNO₂). For the stress assays, rapamycin, methyl methanesulfonate (MMS), NaCl, and sorbitol were added to YES medium at the concentrations indicated in the figure legends (Yang et al., 2016). Each plate was inoculated with 1 μL of conidial suspension (4 × 10⁴ conidia/mL) and incubated at 28°C for 4–7 days in the dark. The assays were carried out in triplicate and were repeated three times.

Glucose minimal medium agar and PDA were point-inoculated with 1 μL of conidial suspension (4 × 10⁴ conidia/mL) and incubated at 28°C for 5 days in the dark. Three plugs were removed from each plate and the conidia produced by A. flavus were suspended in a solution of 0.05% Tween 20 and 7% dimethylsulfoxide (DMSO), and counted in a hemocytometer. Conidial suspensions were viewed with an inverted microscope (Leica Microsystems, Germany).

For analysis of AF production, 1 mL of spore suspension (1 × 10⁶ spores/mL) was inoculated in YES, PDA or GMM supplemented with 50 mM glutamine, ammonium, proline, alanine, or sodium nitrite and incubated in the dark at 28°C for 7 days. AFs were extracted from 500 μL of culture filtrate with an equal volume of chloroform. The chloroform layer was transferred to a new 1.5 mL tube and evaporated to dryness at 70°C. Thin-layer chromatography was used to identify AFs. A solvent system consisting of acetone and chloroform (1:9, v/v) was used, and the plates were observed under UV light at 365 nm. For quantitative analysis of AF production, GeneTools image analysis software was used.

Sclerotia formation was measured as previously described (Amaike and Keller, 2011). Briefly, GMM supplemented with the indicated nitrogen sources, 2% agar, and 2% sorbitol was overlaid with 1 × 10⁴ spores/plate. Cultures were grown at 37°C in complete darkness for 7–10 days. Plates were sprayed with 70% ethanol to kill and wash away conidia and exposed sclerotia. The sclerotia were collected from 10 mm cores and counted in triplicate. Experiments were repeated three times.

Mature live peanut seeds were used to measure the pathogenicity of the WT, ΔnmrA, and complementation (ΔnmrA::nmrA) strains (Duran et al., 2007; Amaike and Keller, 2011). Plates were cultured in the dark at 28°C for 5 days, and the filter paper was moistened daily. Inoculated peanut cotyledons were processed to count conidia and extract AFs, and the methodology of counting conidia and extracting AFs was the same as above described.

Aspergillus flavus mycelia from the WT, ΔnmrA, and ΔnmrA::nmrA strains were harvested after 48 h of growth. RNA was extracted with TRIzol reagent (Biomarker Technologies, Beijing, China), and then a Nano Drop 2000 and Agilent 2100 were used to evaluate the quality of RNA after total RNA extraction and DNase I treatment. TransScript^® All-in-One First-Strand cDNA Synthesis SuperMix was used to synthesize cDNA, and RT-qPCR was performed on a PikoReal^TM Real-Time PCR System (Thermo Scientific Inc.) with PikoReal^TM 2.2 software using TransStart Top Green qPCR SuperMix (TransGen Biotech, Beijing, China). The RT-qPCR conditions were as follows: 95°C for 7 min and 40 cycles of 95°C for 5 s and 60°C for 30 s. The A. flavus actin gene was used as an reference gene to normalize the expression data. The relative expression of genes was calculated using the 2^-ΔΔCt method, and standard deviation was calculated from three biological replicates (Livak and Schmittgen, 2001). The gene-specific primers are shown in Supplementarey Table S3.

In A. nidulans, NmrA can bind nicotinamide dinucleotides and may have a redox-sensing function, since NmrA is bound with NAD⁺ and NADP⁺ preferentially to NADH and NADPH. In addition, NmrA interacts directly with AreA zinc fingers (Stammers et al., 2001; Kotaka et al., 2008; Zhao et al., 2010). The A. flavus nmrA ORF consists of 1,119 bp with one introns, and encodes a putative NMR regulator NmrA with 351 amino acids. A. flavus NmrA might function similarly to its ortholog in A. nidulans, since the two protein sequences have a similarity of 88.07%. We downloaded the NmrA protein sequences of seven representative Aspergillus species (A. clavatus, A. flavus, A. fumigatus, A. nidulans, A. niger, A. oryzae, and A. terreus) from FungiDB² and aligned using DNAMAN software version 6.0.3.99 (Figure 1). The sequence similarity was as high as 94.81%, demonstrating a highly conserved nature of NmrA in Aspergillus. For instance, the NmrA protein sequences in A. flavus and A. oryzae are only differed at one amino acid between the proteins of, reflecting the evolutionary similarity between them. We also identified a canonical Rossmann fold motif in A. flavus NmrA.

A targeted gene deletion strategy was employed to determine the roles of nmrA in A. flavus (Supplementary Figure S1A). The nmrA gene was successfully replaced with the pyrG cassette (Supplementary Figure S1B), indicating that the nmrA deletion mutant (ΔnmrA) was successfully constructed. The ΔnmrA strain was complemented by introducing a construct consisting of the nmrA open reading frame in the pPTR I vector, which generated the complementary strain ΔnmrA::nmrA. The ΔnmrA and ΔnmrA::nmrA strains were verified by RT-PCR and RT-qPCR (Supplementary Figures S1B,C). The growth rate of the ΔnmrA strain consistently matched that of the WT and ΔnmrA::nmrA strains on YES medium and PDA, but the growth rate was slower than that of the WT and ΔnmrA::nmrA strains on GMM supplemented with glutamine, ammonium, proline, alanine, or sodium nitrite as the sole nitrogen source (the usual nitrogen source in GMM is sodium nitrate, NaNO₃) (Figures 2A,B). The edges of the mycelial colonies of the ΔnmrA strain were more irregular on glutamine, ammonium, and proline than on the other nitrogen sources (Figure 2A). It should be mentioned that A. flavus strains could not grow normally on sodium nitrate because of a deficiency of the nitrate reductase gene niaD. Therefore, sodium nitrite was used in place of sodium nitrate; however, the deletion strain showed retarded growth on this nitrogen source as well. These results suggested that nmrA could regulate nitrogen utilization and metabolism.

There was little difference in conidia production among strains when they were germinated on PDA, but the ΔnmrA strain produced more spores than the WT and ΔnmrA::nmrA strains when germination occurred on GMM with ammonium (Figures 2A and 3A). The ΔnmrA mutant also produced more conidiophores than the WT and ΔnmrA::nmrA strains on GMM with ammonium (Figure 3B). To gain further insight into the role of NmrA in conidiation, we performed RT-qPCR analysis. The results showed that transcript levels of the conidiation-related genes abaA (AFLA_029620) and brlA (AFLA_082850) were higher in ΔnmrA than in WT A. flavus and ΔnmrA::nmrA when the strains were germinated on GMM with ammonium (Kato et al., 2003; Son et al., 2013), while there was little difference among strains when germination occurred on PDA (Figure 3C). Collectively, these results indicated that nmrA was likely involved in the regulation of conidiation in A. flavus.

Next, we assessed the production of AFs by the WT, ΔnmrA, and ΔnmrA::nmrA strains on the media described in Figure 2. Like the growth rate, AF production was similar among the WT, ΔnmrA, and ΔnmrA::nmrA strains when they were grown on YES or PDA plates (Figure 4). However, there were significant decrease in AF production among the strains when they were germinated on GMM supplemented with glutamine and alanine (Figures 4A,B). We concluded that nmrA might participate in the regulation of AF biosynthesis in A. flavus.

We found that in addition to suppressing AF biosynthesis, nmrA deletion resulted in a concomitant increase in sclerotia production. The ΔnmrA strain displayed a more compact and intensive distribution of sclerotia on the plate than the WT and ΔnmrA::nmrA strains (Figure 5A). To some extent, sclerotia production by the ΔnmrA mutant was greater on glutamine than on ammonium (Figure 5B), indicating that it could be affected by the quality of the nitrogen source. In summary, the effect of the nmrA gene on sclerotia production by A. flavus was modulated by the nitrogen source.

Although Wilson examined the role of three NmrA orthologs in Magnaporthe oryzae during infection of rice (Wilson et al., 2010), there has been no report on the role of NmrA in infection by A. flavus. To dissect the role of NmrA in invasive growth, we assessed the growth of A. flavus on peanut seeds. Figure 6A shows that the ΔnmrA strain was crippled in its ability to colonize and sporulate on host seeds compared with WT A. flavus and ΔnmrA::nmrA. Deletion of nmrA also led to statistically significant reductions in conidia production and AF biosynthesis (Figures 6B,C), which resulted in the decrease in virulence. These data suggested that nmrA deletion reduced the virulence of A. flavus.

To explore the potential roles of NmrA in responses to environmental stressors, we tested fungal sensitivity to the target of rapamycin (TOR) inhibitor rapamycin, the alkylating agent MMS, and the osmotic stressors NaCl and sorbitol. Previous research has shown that TOR-mediated repression of nitrogen catabolic genes and virulence occurs through MeaB-dependent and MeaB-independent mechanisms in Fusarium oxysporum (López-Berges et al., 2010), so it seemed possible that an interaction between NmrA and TOR might be involved in nitrogen metabolism regulation in A. flavus. The assay results showed that the ΔnmrA mutant was more sensitive to rapamycin and MMS than WT A. flavus or ΔnmrA::nmrA. However, the ΔnmrA mutant was not more sensitive to the osmotic stressors (Figures 7A,B). Overall, these results indicated that NmrA is associated with the TOR pathway.

The potential interactive partners of NmrA were identified using the SMART analysis service³. The identified partners included the GATA transcriptional activator AreA (AFLA_049870), the bZIP transcription factor MeaB (AFLA_031790), the GATA transcription factor AreB (AFLA_136100), the siderophore transcription factor SreA (AFLA_132440), the nitrate reductase NiaD (AFLA_018810), and the C6 transcription factor NirA (AFLA_093040) (Figure 8A). Nitrite reductase NiiA (AFLA_018800) was also analyzed. The possible interactions were analyzed by real-time qRT-PCR. The transcript levels of areA, areB, nmrA, meaB, sreA, and nirA on ammonium were higher than those on nitrite, suggesting that these genes are likely involved in NMR. The finding that the areA expression level on ammonium exceeded that on nitrite was unexpected and was probably correlated with nitrate reductase deficiency and the relative richness of the nitrogen sources. However, transcript levels of niaD and niiA on ammonium were lower than those on nitrite (Figure 8B). Furthermore, on ammonium, all transcript levels except that of sreA were higher in the ΔnmrA strain than in the WT and ΔnmrA::nmrA strains. In contrast, on nitrite, areA, meaB, sreA, and nirA transcript levels were downregulated in the ΔnmrA mutant, while areB, niaD, and niiA transcript levels were upregulated, which was consistent with the postulated reverse roles of areA and areB. Importantly, the finding that the areA, areB, and meaB transcript levels on ammonium were higher in the ΔnmrA mutant than in the WT and ΔnmrA::nmrA strains indicates that the roles of these genes in nitrogen repression might be related to nmrA. Moreover, the observation that niaD and niiA transcript levels on nitrite were significantly higher in the ΔnmrA mutant than in WT A. flavus and ΔnmrA::nmrA demonstrates that NmrA strictly repressed transcriptional activation of genes encoding enzymes required for utilization of less favored nitrogen sources. In conclusion, these findings shed light on the complexity and sophistication of the regulatory mechanisms of NMR.