One putative phosphoglucose isomerase (UniProt No. B8NBA7) was identified by Blast search of the A. flavus genome using the S. cerevisiae PGI sequence (UniProt No. P12709). The predicted PGI protein contains 553 amino acids. The phylogenetic analysis reveals that A. flavus PGI shares the highest similarity with A. oryzae PGI and the lowest similarity with C. neoformans PGI ( Figure S1A ). Analysis of the conserved domain of PGIs at the website of NCBI (https://www.ncbi.nlm.nih.gov/) shows that A. flavus PGI contains a catalytic domain that is highly conserved in fungal species ( Figure S1B ).

To evaluate the significance of PGI in A. flavus, the pgi gene was deleted by homologous recombination. The Δpgi mutant was obtained by protoplast transformation of A. flavus using A. fumigatus pyrG as the selective marker. The transformants were screened on the regeneration medium containing 1% fructose (Fru) and confirmed by PCR and Southern blotting analyses ( Figure S2 ). The revertant strain (RT) was generated by reintroducing the pgi gene and the AfpyrG gene back to the original pgi locus ( Figure S2 ).

It has been reported that pgi deletion mutants required different carbon sources in different species (Tuckman et al., 1997; Upadhyay and Shaw, 2006; Limón et al., 2011; Zhang et al., 2015). We therefore tested the requirement of the A. flavus Δpgi mutant for a range of carbon sources. The mutant was unable to grow on minimal medium (MM), in which glucose is the sole carbon source. Addition of Fru to MM could partially restore the growth while the mutant reached the best growth on MM supplemented with 1% of Fru ( Figure 1 ). Similarly, when 1% of Fru was used as the sole carbon source, the Δpgi mutant could not grow. Combinations of 0.01%–0.1% Glc with 1% of Fru partially restored the growth and 0.5% of Glc with 1% Fru supported the best growth of the Δpgi mutant ( Figure 1 ). In addition, the mutant displayed a retarded growth when 1% of gluconate, galactose (Gal), N-acetylglucosamine (GlcNAc), or glycerol (Gly) was used as the sole carbon sources, respectively ( Figure S3 ). These results suggest that PGI in A. flavus is essential for sugar metabolism.

As the mutant grew well on the medium containing 1% Fru and 0.5% Glc (MMFG), this condition was used for subsequent phenotypic analysis. When the wild-type (WT), Δpgi, and RT strains were cultured on a solid MMFG medium, there was no difference in colony diameter among three strains; however, the colony of the Δpgi mutant became white ( Figures 2A, B ). Furthermore, the number of conidia produced by the Δpgi mutant was decreased 7,000-fold or 4,000-fold as compared with that by the WT and RT strains, respectively ( Figure 2C ). In liquid MMFG medium, the Δpgi mutant exhibited a significantly delayed germination, showing only 2% conidia of the Δpgi mutant germinated in comparison to 99% or 77% of the WT or RT, respectively ( Figure S4A ). Even after 12 h of incubation, only 43% of the mutant conidia germinated ( Figure S4B ). These results demonstrate that pgi in A. flavus is required for conidiation and germination.

Since PGI catalyzes the conversion between Glc6P and Fru6P, both being the key sources of nucleotide sugars required for cell wall biosynthesis, we next explored the sensitivity of the Δpgi mutant towards cell wall stresses. Spores of the WT, Δpgi, and RT strains were cultured on MMFG medium supplemented with cell wall-disrupting agents Calcofluor White (CFW) and Congo Red (CR), cell membrane inhibitor sodium dodecyl sulfate (SDS), and antifungal drugs itraconazole (ITR), caspofungin (CAS), and amphotericin B (AMB). As shown in Figure 3A , growth of the Δpgi mutant was dramatically inhibited by CR and ITR. However, compared to the WT and RT strains, growth of the Δpgi mutant was not affected by CFW, SDS, CAS and AMB ( Figure S5 ). We further quantified cell wall components of each strain by combining hot alkali extraction, acidic hydrolysis, and HPAEC-PAD analysis (Francois, 2006). As shown in Figure 3B , the content of cell wall polysaccharides in the Δpgi mutant, particularly the glucan content, was not significantly different from the WT, which is consistent with the insensitivity to CAS. No significant difference of chitin, galactomannan (GM), and mannan was observed in the mutant in comparison to the WT and RT strains. However, defects in glycosylation/mannosylation of cell wall glycoproteins in the mutant might impact cell wall integrity and result in sensitivity to CR. This requires further investigation. Overall, the above results indicate that pgi deficiency in A. flavus affects cell wall integrity.

Fungal sclerotium is considered as a resistant structure withstanding adverse environmental conditions (Horn et al., 2009). To assess if pgi is involved in sclerotium formation, the WT, Δpgi, and RT strains were inoculated onto the MMFG medium and cultured in the dark at 37°C for 7 days. As shown in Figure 4 , sclerotia could be observed in the WT and RT strains but barely seen in the Δpgi mutant before or after the ethanol wash, suggesting that deletion of the pgi might affect the adaptation of A. flavus to the adverse environmental conditions. Indeed, this was further confirmed when we tested responses of the mutant toward temperature and osmotic stresses.

When the WT, Δpgi, and RT strains were grown on MMFG medium at 28°C, 37°C, or 42°C, all strains showed an optimal growth at 37°C but a remarkably reduced growth at 28°C or 42°C ( Figure 5A ). Compared to the colony diameter at 37°C, growth of the Δpgi mutant was reduced by 57% at 28°C ( Figure 5B ) whereas no significant difference was observed at 42°C ( Figure 5B ). These results indicate that pgi deletion affects the growth of A. flavus at a lower temperature but is not required for thermotolerance.

Fungi are known to resist a wide variety of environmental stresses (Brown et al., 2017). Previous studies have shown that pgi is associated with the response to osmotic or oxidative stresses in F. graminearum and Cryptococcus neoformans (Zhang et al., 2015; Zhou et al., 2021). Osmotic stabilizers are reported to rescue the sporulation defect of the mutant in A. nidulans (Upadhyay and Shaw, 2006). In A. flavus, as shown in Figure 6 , the Δpgi mutant displayed significant sensitivities toward osmotic stress agents (1.2 M sorbitol, 0.8 M NaCl, and 0.6 M KCl) and oxidative stress agents (5 mM H₂O₂). Altogether, these results demonstrate that deletion of the pgi in A. flavus affects not only sclerotia formation but also response to stresses, implying the important role of pgi in the adaptation of A. flavus to environmental conditions.

A. flavus is one of the main opportunistic human pathogens responsible for aspergillosis; therefore, we first evaluated the virulence of the A. flavus Δpgi mutant in our newly developed Caenorhabditis elegans-based fungal infection model (Ahamefule et al., 2020). Survival rates of glp-4 (bn2); sek-1 (km4) worms infected by A. flavus strains were counted at 24-h intervals and plotted with the Kaplan–Meier survival curve. The survival rate of the Δpgi mutant at 72 h postinfection was 66 ± 17% (compared to 51 ± 12% for the WT strain and 52 ± 23% for the RT strain) in killing assay, revealing attenuated virulence of the mutant (p < 0.0001, Figure 7A and Table S1 ). In addition, the hyphal filamentation rate of the Δpgi mutant at 24 h was 2 ± 3%, which was much lower than that of the WT and RT strains (12 ± 11% for the WT strain and 10 ± 4% for the RT strain, Table S1 ). These results demonstrate that PGI affects the virulence of A. flavus in C. elegans infection model.

To exclude the possibility that the reduced virulence of the Δpgi mutant in C. elegans was not solely attributable to the reduced assay temperature (20°C), we then evaluated virulence of the A. flavus Δpgi mutant in the G. mellonella model at 37°C. Dead G. mellonella larvae infected by A. flavus strains were counted at 24-h intervals and plotted with the Kaplan–Meier survival curve. The survival rate of the Δpgi mutant at 72 h was 87 ± 3% (compared to 30 ± 7% for the WT and 23 ± 7% for the RT), revealing significant attenuated virulence of the mutant (p < 0.0001, Figure 7B and Table S2 ). Altogether, our results demonstrate that pgi deficiency in A. flavus leads to attenuated virulence in animal models.

Uninoculated (CK) and Tween 20-inoculated larvae were included as controls to ensure that environmental conditions and physical injuries by inoculation do not affect the survival rates.

As A. flavus is also a notorious plant pathogen causing contamination of agricultural crops (Phan et al., 2021), we also accessed the infection of the A. flavus Δpgi mutant on peanut and corn seeds. After incubation at 28°C for 6 days in the dark, the seeds infected by Δpgi produced few conidia as compared with those infected by the WT or RT strain ( Figure 8A ). Counting the conidia washed from the seeds revealed that the conidia from the seeds infected by the mutant were significantly less than that infected by the WT or RT ( Figures 8B, C ). We propose that the colonization defects of the mutant are due to the nutrients in the seeds being not sufficient to support the growth of the mutant. To validate this, we used grinding powder of peanut or corn as the nutrition to incubate strains at 37°C. As shown in Figure S6 , the Δpgi mutant could not grow on peanut powder plates or only showed very limited growth on corn powder plates.

As aflatoxin B1 (AFB1) is the most fatal secondary metabolite rendering toxicity to contaminated seeds, we further measured the accumulation of AFB1 in the seeds infected by A. flavus strains using thin layer chromatography (TLC). As shown in Figure 9 , no aflatoxin B1 was detected in peanut and corn seeds infected by the Δpgi mutant whereas a large amount of AFB1 was detected in the seeds infected by the WT and RT strains. These results suggest that pgi is required for colonization and growth of A. flavus on crop seeds; therefore, targeting A. flavus PGI might be a practical strategy to reduce aflatoxin pollution.

A. flavus CA14 Δku70ΔpyrG was used as the parental strain for transformation. CA14 Δku70 was used as the wild type (WT) for phenotypic analysis. WT and RT stains were cultured at CM or MM medium, while the Δpgi mutant was cultured at CM supplemented with 1% w/v Fru or MM containing 1% w/v Fru and 0.5% w/v Glc. Five millimoles of uracil and uridine was added for the strains with pyrG auxotroph. YPD medium was used for sclerotium assay. Mycelia came from liquid medium cultivation at 37°C with shaking at 200 rpm for 2 days, then were harvested, washed with distilled water, frozen in liquid nitrogen, and ground using a mortar and pestle. The mycelium powder was stored at -80°C for DNA extraction. The spores were collected by using 0.02% v/v Tween-20 from plates with 48 h of incubation at 37°C.

The Δpgi mutant and revertant strains were constructed by homologous recombination. PCR was used to generate the upstream and downstream fragments amplified by two pairs of primers P9/P10 and P13/P14. Likewise, the AfpyrG (1.9 kb) fragment was amplified from the AfpyrG plasmid by primers P11/P12. The fusion of three purified fragments was used to assemble in pCE-Zero vector and named Afl-pgi plasmid. The pgi deletion fragment was transformed into CA14 Δku70ΔpyrG protoplasts. Transformants were screened on MM medium supplemented with 1 M sorbitol and 1% w/v fructose and then confirmed by PCR and Southern blotting. Given the destroyed growth of Δpgi on MM medium, the construction of the revertant (ΔpyrG) strain was done according to the following procedure. The fragment from the upstream flanking region to the downstream flanking region was amplified from genomic DNA of the WT strain by primers P15/P16 and transformed into the Δpgi protoplasts. Transformants were screened on MMU (containing 5 mM uracil and uridine) medium supplemented with 1 M sorbitol. A similar strategy was used for revertant strain construction, and the AfpyrG marker was reintroduced. Finally, all of the constructed strains were confirmed by PCR and Southern blotting, in which Sma I and Kpn I were used for genomic DNA digestion, and the downstream and AfpyrG regions were used as the probes. The primers used in this study are listed in Table S3 .

To test the growth phenotype of the WT, Δpgi, and RT strains, 10⁵ conidia were serially diluted and spotted on MM solid medium supplemented with different concentrations of glucose, fructose, gluconate, and other carbon sources.

The radial growth rate was measured by spotting 10⁵ conidia of the WT, Δpgi, and RT strains at the center of the plate and incubating at 37°C in the dark condition. The colony diameter was monitored every 12 h. After 10 days, conidia were washed and suspended in 5 ml of 0.02% v/v Tween-20, and the number of conidia was counted with hemocytometer.

To observe the germination of the Δpgi mutant, 10⁵ conidia of the WT, Δpgi, and RT strains were incubated in MMFG liquid medium containing coverslips. After incubation at 37°C for a specific time, coverslips were placed on the top of a glass slide and inspected by microscopy.

For sclerotium formation analysis, MMFG solid medium was used to induce sclerotia. 10⁶ conidia of the WT, Δpgi, and RT strains were inoculated on the MMFG plate at 37°C in the dark for 7 days. Three pores with 1.5-cm diameter were drilled at equal distances along the radius of each fungal colony, and the plates were then sprayed with 75% ethanol to kill and wash away conidia to aid in enumeration of sclerotia.

For the temperature assay, 10⁵ conidia of WT, Δpgi, and RT strains were inoculated onto MMFG medium and incubated at 28°C, 37°C, and 42°C for 2 days. The colony diameter was measured and the growth rate was calculated.

To examine susceptibility to various stresses, 5 μl of conidia (10⁷, 10⁶, 10⁵, and 10⁴/μl) was inoculated onto MMFG media containing specific concentrations of Congo Red (CR), Calcofluor White (CFW), sodium dodecyl sulfate (SDS), itraconazole (ITR), amphotericin B (AMB), caspofungin (CAS), sodium chloride (NaCl), potassium chloride (KCl), sorbitol, and hydrogen peroxide (H₂O₂). The plates were incubated at 37°C and photographed after a 2-day incubation.

The cell wall content analysis was performed according to previous methods with a slight modification (Francois, 2006). After 48 h of shaking at 37°C, 200 rpm in MMFG liquid medium, mycelia were harvested by filtration and then disrupted by gridding in liquid nitrogen. SDS-BME (50 mM Tris; 50 mM EDTA; 2% SDS; 1 mM TCEP) was added into the ground powder and then boiled for 40 min. The cell wall fractions were washed with Milli-Q water until bubbles disappeared, and then freeze-dried. Afterward, 10 mg of dry cell wall mass was wetted with 75 μl of 75% H₂SO₄ for 3 h at room temperature. This mixture was diluted to 2 N H₂SO₄ with 0.95 ml of Milli-Q water then boiled at 100°C for 4 h. Extra H₂SO₄ was neutralized by addition of Ba(OH)₂ solution until the pH achieves neutrality. The formed BaSO₄ was precipitated at 4°C overnight. The monosaccharide content in the supernatant was measured by HPAEC-PAD using a CarboPac PA10 anion-exchange column and equipped with an AminoTrap guard column. Elution was performed at room temperature at a flow rate of 1 ml min^-1 with 18 mM NaOH.

Pathogenicity tests for the virulence of A. flavus WT and Δpgi strains in C. elegans were performed according to previous methods used in the C. elegans–A. fumigatus infection model (Ahamefule et al., 2020). Briefly, 10⁸ conidia of the WT and Δpgi strains were used for pre-infection. Worms were transferred to killing assay medium (BHI+) at 25°C after 16 h of pre-infection. Then, survival rates of worms were recorded at 24, 48, and 72 h post-infection. The hyphal filamentation rate indicated the rate of hyphal filaments protruded from worm bodies.

For testing pathogenicity in the G. mellonella model, six instar larvae were selected and divided into five groups for untreated control, 0.02% Tween-20 control, and the WT, mutant, and RT strains. A total of 90 larvae were tested for each group. Ten microliters of 1 × 10⁶ cfu/ml conidia in 0.02% Tween-20 was injected into the hind proleg of larva with a Hamilton syringe. The survival rates of the larvae were recorded at 37°C incubation for 24, 48, and 72 h after post-infection. The larvae that could not move were defined as dead larvae, which usually had dark spots or apparent melanization.

The ability of the Δpgi mutant to infect crop seeds was conducted as described previously (Yang et al., 2018). Firstly, fresh corn seeds and peanut seeds with similar size and shape were selected. The endosperm was removed with toothpicks to prevent germination and provide infection sites. Seeds were disinfected with 0.05% of sodium hypochlorite for 3 min and 75% of ethanol for 1 min and then were rinsed three times with sterile water before putting into a 100-ml sterile flask. Seeds were inoculated with 400 μl of conidia at 10⁷/ml for 30 min and incubated at 28°C in the dark for 6 days. Humidity was maintained by wet filter paper at all times. The infected seeds were transferred into 50 ml of tubes containing 20 ml of 0.02% Tween 20. In order to release conidia from the seed surface, tubes were vigorously shaken at 200 rpm for 5 min. Then, 100 μl of spore suspension was removed for gradient dilution and spores counted using a hemocytometer. Each experiment was repeated for three times.

Aflatoxin was extracted from 500 μl of filtrate with an equal volume of chloroform. The chloroform layer was transferred into a new 1.5-ml tube and evaporated to dryness at 70°C. TLC was used to analyze aflatoxin. A solvent system consisting of acetone and chloroform (1:9, v/v) was used. Aflatoxin was visualized under ultraviolet (UV) light at 365 nm.

The Kaplan–Meier survival curve was plotted with GraphPad Prism 8 while statistical p-values for survival were calculated with the log-rank (Mantel–Cox) test. T-test was used for comparison of two different groups. A one-way ANOVA multiple-comparison test was performed for significance analysis of multiple comparisons.