Hydrogen peroxide as a multifaceted regulator of the Atg4 protease and autophagy in the pathogenic fungus Alternaria alternata

Autophagy-related protease AaAtg4 was previously identified as a key regulator in the pathogenic fungus Alternaria alternata, orchestrating a complex interplay among autophagy, oxidative stress resistance, iron homeostasis, and ACT toxin biosynthesis. The underlying mechanisms of AaAtg4 in relation to oxidative stress response remain unknown. Genetic and biochemical analyses. In this study, we examined the effect of hydrogen peroxide (H₂O₂) on AaAtg4. Functioning as a cysteine protease, AaAtg4 directly interacts with the AaAtg8 ubiquitin-like protein and is indispensable for AaAtg8 processing and autophagosome formation, with its enzymatic activity modulated by oxidative cues. H₂O₂ differentially impacts AaAtg4 activity, phosphorylation, binding with AaAtg8, AaAtg8 lipidation/delipidation, and autophagy. H₂O₂ has biphasic effects on AaAtg4. Moderate H₂O₂ levels enhance AaAtg4 activity and autophagy, whereas excessive H₂O₂ suppresses both, revealing a threshold-dependent redox regulation. Furthermore, AaAtg4 interacts with the stress-responsive mitogen-activated protein kinase AaHog1, which modulates its phosphorylation under conditions less conducive to autophagy and thus, reinforces a dynamic signaling axis. These findings indicate that H₂O₂ has multifaceted effects on AaAtg4 in a dosage-dependent or threshold-specific manner. These regulatory mechanisms may position AaAtg4 as a central integrator of cellular stress responses and secondary metabolism, thereby advancing our understanding of fungal pathogenicity and environmental adaptation.

Autophagy plays a critical role in cell survival in all eukaryotes under starvation or other stress conditions. The system degrades damaged organelles, misfolded proteins, and many other unnecessary components, contributing to cellular homeostasis (Mizushima and Komatsu, 2011). In response to starvation or stress, autophagy begins with the formation of a phagophore (a cup-shaped isolation membrane). The membrane elongates and encircles the material, forming an autophagosome (a double-membrane structure). Components prone to degradation are selectively or nonselectively enclosed in an autophagosome, which is then fused with a vacuole (a lysosome in mammals). Hydrolytic enzymes in the vacuole degrade the components for recycling. Autophagy is controlled by dozens of autophagy-related (Atg) proteins. Six central functional units, each composed of several Atg proteins, are required for autophagy. They are the Atg1-kinase complex, the class III phosphatidylinositol 3-kinase (PtdIns3K) complex, the Atg9 membrane complex, the Atg2-Atg18 complex, the Atg12 ubiquitin-like protein conjugation system, and the Atg8 ubiquitin-like protein conjugation system (Yang and Klionsky, 2010; Mizushima et al., 2011; Noda and Inagaki, 2015).

Atg4 is a cysteine protease that plays a central role in the Atg8 ubiquitin-like protein conjugation system (Kirisako et al., 2000; Ichimura et al., 2000). Atg4 directly interacts with Atg8, regulating its processing and deconjugation. In addition, Atg4 is involved in the assembly of the pre-autophagosomal structure (PAS) (Kirisako et al., 2000). Unlike animals, yeast has only a single copy of each of Atg4 and Atg8 (Kirisako et al., 2000). Atg4 removes arginine in the C-terminus of Atg8 to expose its glycine residue, which can be attached to phosphatidylethanolamine (PE) to form an Atg8-PE conjugate. The Atg8-PE is then attached to autophagosomal membranes. This conjugation process (also called lipidation) is crucial for the phagophore elongation and autophagosome formation (Reggiori and Klionsky, 2013). During the late stage of autophagy, Atg8 is released from Atg8-PE, which is anchored to the outer membrane of autophagosomes by Atg4, a process known as deconjugation or delipidation. This process recycles Atg8 for the nascent conjugation reaction (Nakatogawa et al., 2012; Hirata et al., 2017). Both Atg8 conjugation and deconjugation reactions are regulated by Atg4, which is critical for autophagosome formation. Thus, impairment of Atg4 blocks autophagosome formation and the initiation of autophagy. Atg4 activity is known to be regulated by redox status. In yeasts, humans, and plants, the Atg4 protein is active in its reduced form and becomes inactive in its oxidized form (Scherz-Shouval et al., 2007; Pérez-Pérez et al., 2012; Woo et al., 2014). The yeast Atg4 forms a single disulfide bond, which is required for its activation in response to ambient redox status and is mediated by the reducing oxidoreductase thioredoxin (Pérez-Pérez et al., 2014).

Atg1 is a serine/threonine kinase essential for autophagy initiation by recruiting several Atg proteins to the PAS. In yeast, Atg1 has been shown to phosphorylate Atg4 at the site of autophagosome biogenesis, thereby inhibiting Atg4 protease activity and facilitating autophagosome formation (Sánchez-Wandelmer et al., 2017). Atg1 interacts with Atg4 at the PAS, a crucial site for spatial regulation of Atg4 activity, ensuring proper Atg8 processing during autophagosome formation. Atg4 remains active in the cytoplasm but is suppressed at the autophagosome formation site. Both Atg1 and Atg4 are well characterized in yeast; however, their biochemical functions in response to redox status in phytopathogenic fungi remain unknown. Autophagy-mediated degradation of peroxisomes (pexophagy) confers cellular resistance to peroxides in the phytopathogenic fungus Alternaria alternata (Wu et al., 2022).

The tangerine pathotype of A. alternata is a highly specialized fungal strain that causes Alternaria brown spot in citrus, particularly tangerines (Kohmoto et al., 1993). The pathogen produces an ACT host-selective toxin to kill plant cells before penetration. The production of ACT by the tangerine pathotype is absolutely required for fungal pathogenicity (Tsuge et al., 2013). Additionally, the ability to mitigate the toxicity of reactive oxygen species (ROS) is essential for full virulence (Chung, 2012). To survive in harsh oxidative environments, A. alternata relies on well-regulated mechanisms to detoxify ROS. The A. alternata pathogen generates low-level H 2 O 2 via an NADPH oxidase (Nox), which in turn serves as a signal to activate gene expression and nuclear localization of the stress-responsive transcription regulator Yap1 and the high-osmolarity glycerol (Hog1) mitogen-activated protein kinase (Yang and Chung, 2012, Yang and Chung, 2013). Both Yap1 and Hog1 are highly sensitive to the ambient redox status and activate the expression of genes related to antioxidants, such as catalase, peroxidase, superoxide dismutase, Major Facilitator Superfamily (MFS) transporters, as well as thioredoxin and glutaredoxin systems (Lin et al., 2009, Lin et al., 2011; Yang et al., 2016; Ma et al., 2019). In addition to oxidative stress, the A. alternata Hog1 (AaHog1) is involved in cellular resistance to osmotic stress, fungicides, and xenobiotics (Lin and Chung, 2010; Chen et al., 2017). The role of AaHog1 in autophagy remains unknown. In yeast, loss of Hog1 impairs autophagy under osmotic stress, establishing a connection between Hog1 and autophagy (Prick et al., 2006). Hog1 is also involved in mitophagy in yeast (Mao et al., 2011). Suppressing the LeHog1 phosphorylation lowers the expression of LeAtg8 in the mushroom Lentinula edodes, indicating the involvement of Hog1 in autophagy induced by H 2 O 2 (Yan et al., 2023).

We hypothesized that the oxidant H 2 O 2 would trigger autophagy in A. alternata. This study investigated the impact of H 2 O 2 -mediated redox stress on AaAtg4 and autophagy. AaAtg4 responded to H 2 O 2 in a dosage-dependent or threshold-specific manner. H 2 O 2 exhibited biphasic effects on AaAtg4 phosphorylation, AaAtg8-PE formation, and autophagy, while suppressing AaAtg4 activity and its binding to AaAtg8. Unlike yeast, AaAtg1 did not phosphorylate AaAtg4; however, AaHog1 deficiency caused hyperphosphorylation under autophagy-suppressive conditions.

Fungal strains and growth conidiation

The wild-type EV-MIL31 strain of A. alternata (Fr.) Keissler was isolated from a diseased Minneola tangelo (Citrus paradisi Macfad. x C. reticulata Blanco) leaf (Lin et al., 2009) and used throughout this study. The Δ AaAtg1 strain was created by the split-marker approach to delete the AaAtg1 gene encoding a serine/threonine kinase in the EV-MIL31 strain (Choo et al., 2026). The Δ AaAtg4 strain was created by the split-marker approach to delete the AaAtg4 (Lu et al., 2026, in revision)1. All other strains used in this study were modified from the EV-MIL31 strain, with their characteristics (Supplementary Table 1). Fungal strains were grown on potato dextrose agar (PDA, Difco, Sparks, MD, U.S.A.) or minimum medium (MM) (Chung et al., 2002) under constant fluorescent light at 28 °C for 3 to 5 days. For DNA and RNA isolation, fungal strains were cultured in potato dextrose broth (PDB, Difco) for 2 days. For protein isolation, strains were subjected to a medium shift and cultivated for 24 h. The mycelium was harvested by filtration through Miracloth, washed with water, and transferred into PDB, MM, or MM-N (MM lacking nitrogen), or into PDB supplemented with varying concentrations of H 2 O 2, then incubated on a shaker for different periods. For fluorescence microscopy, fungal strains were cultured for 8 h, as fluorescence signals were difficult to detect in cultures incubated for fewer than 6 h. For western blot analysis, strains were grown in MM-N or treated with H 2 O 2 for 4 h prior to protein extraction. Longer incubation times (>4 h) led to protein degradation, especially under H 2 O 2 treatment.

Miscellaneous procedures for the manipulation of nucleic acids

Fungal genomic DNA was obtained using a Genomic DNA Mini Purification kit (BioKit, Taipei, Taiwan). Fungal RNA was isolated using TRI reagent (Sigma-Aldrich, St. Louis, MO, USA) or the PureLink RNA Kit (Invitrogen, MA, U.S.A.) according to the manufacturer’s protocols. Complementary DNA (cDNA) was synthesized from the isolated RNA using iScript Reverse Transcriptase (Bio-Rad, Hercules, CA, U.S.A.). DNA cleavage with restriction endonucleases, electrophoresis, DNA ligation, and the transformation of bacterial cells (Escherichia coli DH5α) were performed according to standard methods. Targeted gene deletion using the split-marker approach and fungal transformation were conducted as described previously (Chung et al., 2002; You et al., 2009). Plasmids pNeoGPE1 and pHygGPE1 were modified from pBarGPE1 (Pall and Brunelli, 1993) by replacing the bialaphos-resistance cassette (bar) with a neomycin (Neo)- or a hygromycin (Hyg)-resistance cassette, respectively, and used to express a gene under the control of the Aspergillus nidulans gpdA promoter. Fungal protoplasts were prepared according to the procedures described by Chung et al. (2002), and fungal transformations were performed employing the calcium chloride and polyethylene glycol-mediated method (You et al., 2009). A detailed overview of the plasmid constructs used in this study is provided in Supplementary Figure 1.

Western blot and phos-tag analyses

To assess autophagy formation, AaAtg8 cDNA was translationally fused with the C-terminus of GFP, and the resulting fragment (GFP-AaAtg8) was cloned into a pCB1532 plasmid, which carries a sulfonylurea-resistance (Sur) gene, and expressed under the control of the AaAtg8 promoter. The resulting construct was transformed into wild-type and Δ AaAtg4 protoplasts, yielding WT/GFP-AaAtg8 and Δ AaAtg4 /GFP-AaAtg8, respectively. To test AaAtg4 protease activity, AaAtg8 cDNA was translationally fused with the N-terminus of GFP. The AaAtg8-GFP fragment was cloned into pCB1532 and transformed into wild-type protoplasts to yield a WT/AaAtg8-GFP strain, which was used for assaying AaAtg4 protease activity. A medium shift was conducted to evaluate autophagy: fungi were grown in PDB for 24 h, and the mycelium was collected, washed, transferred to PDB or MM-N with or without H 2 O 2, and incubated for 4–6 h. Proteins were isolated using lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, and 0.1% Triton X-100) amended with 1% protease inhibitor cocktail (Biokit, Toufen City, Miaoli, Taiwan), separated by SDS-PAGE, and transferred to PVDF membranes as described (Wu et al., 2022). Following transfer, proteins on the PVDF membrane were stained with Ponceau S (Sigma-Aldrich, cat # P3504), photographed, and subsequently incubated overnight at 4 °C with a rabbit anti-GFP antibody (1:5000), an anti-HA antibody, or an anti-actin antibody. Because the proteins on the PVDF membrane were both stained and probed with an antibody, Ponceau S staining served as a loading control. Membranes were treated with an HRP-conjugated goat anti-rabbit IgG (1:10,000, Jackson ImmunoResearch, West Grove, PA, U.S.A.) and LumiFlash Prime Chemiluminescent Substrate, HRP (Visual Protein, Taipei, Taiwan) for chemiluminescence signals, which were acquired by the ChemiDoc MP imaging system (Bio-Rad, Hercules, CA, U.S.A.) and ImageLab software (version 6.1.0). The intensity of the band was determined using ImageJ software (US National Institutes of Health, Bethesda, MD, U.S.A.) (https://imagej.net/ij/).

For phos-tag analysis, AaAtg4 was cloned into a pNeoGPE1-HA plasmid carrying a linker HA-tag sequence (5’-atgggcagctacccatacgatgttccagattacgct-3’) and a G418-resistant cassette, and the resulting construct was transformed into wild-type and Δ AaAtg1 protoplasts to yield WT/HA-AaAtg4 and Δ AaAtg1 /HA-AaAtg4, respectively. Because Δ AaHog1 fails to produce protoplasts (Lin and Chung, 2010), the Δ AaHog1 /HA-AaAtg4 strain was created by deleting AaHog1 and replacing it with a hygromycin-resistant cassette in the WT/HA-AaAtg4 strain by the split marker method. Similarly, the Δ AaHog1 /GFP-AaAtg8 strain was created by deleting AaHog1 in the WT/GFP-AaAtg8 strain. Primers used to generate split-marker fragments and replace AaHog1 are summarized in Supplementary Table 2. Proteins were extracted from fungal strains using the lysis buffer amended with protease inhibitors (1.0 mM Na 3 VO 4 and 1.0 mM NaF). Proteins, after being mixed with 1 mM Zn (NO 3) 2 at a 10:1 (v/v) ratio, were combined with 5x loading dye, heated to 95 °C for 5 min, and centrifuged at 8,150 x g (Eppendorf 5424, Hamburg, Germany) for 3 min to remove cell lysate. Zn-Phos-tag-SDS-PAGE gels were prepared by adding 10 mM Zn (NO 3) 2 and 5 mM Phos-tag acrylamide AAL-107 (Fujifilm Wako, Chuo-Ku, Osaka, Japan) into a 30% acrylamide:bis-acrylamide (37.5:1, v/v) solution. Proteins were separated using a freshly prepared running buffer containing 0.5 M Tris base, 0.5 M MOPS, 10% SDS, and 0.5 M NaHSO 3. Proteins were transferred to polyvinylidene difluoride (PVDF) membranes, washed with 10 mM EDTA, and incubated with an anti-HA (Sigma-Aldrich) or anti-Hog1 antibody (Santa Cruz Biotechnology, Dallas, TX, U.S.A.). Unless otherwise indicated, proteins were extracted from 4-h cultures following the medium shift.

Yeast two-hybrid assay

Y2H was carried out by the GAL4-based Matchmaker Yeast Two-Hybrid System (Takara Bio, San Jose, CA, U.S.A.). AaAtg1, AaAtg4, or AaAtg8 was amplified with the primers listed in Supplementary Table 2, fused with the GAL4 DNA-activation domain (AD), and cloned into the pGADT7 plasmid as prey. AaAtg4 or AaAtg8 open reading frame was fused with the GAL4 DNA binding domain (BD) and cloned into the pGBKT7 vector as bait. Plasmid constructs were propagated in Escherichia coli and transformed into the yeast Y187 or Y2HGold strain. Yeast transformation and mating procedures followed the manufacturer’s protocol (Takara Bio). Positive (Tp53; the asterisk denotes mating) and Negative (TLam) controls were also included to ensure the reliability of the kit. Each experimental pairing contained four independent yeast transformants. All yeast strains, after serial dilution, were cultured on stringent quadruple-dropout SD/–Leu/–Trp/–Ade/–His agar plates amended with X-α-Gal (40 µg/ml) and Aureobasidin A (200 ng/ml) (QDO/X/A).

Co-immunoprecipitation

For Co-IP experiments, plasmids pNeoGPE1-HA-AaAtg4 and pCB1532-GFP-AaAtg8 were individually or co-transformed into wild-type protoplasts, resulting in three different fungal strains: WT/HA-AaAtg4, WT/GFP-AaAtg8, and WT/HA-AaAtg4+GFP-AaAtg8. Fungal strains were cultured in PDB for 24 h. Proteins were extracted from three strains using RIPA buffer amended with Ser/Thr phosphatase inhibitors (1.0 mM Na 3 VO 4 and 1.0 mM/l NaF). Immunoprecipitation was performed using anti-HA magnetic beads (Sigma-Aldrich) as per the manufacturer’s recommendation. Proteins were eluted from magnetic beads using 2× Laemmli buffer (Bio-Rad) and analyzed by Western blots. Anti-HA (Sigma-Aldrich), anti-GFP, or anti-actin (Millipore Sigma, Burlington, MA, U.S.A.) antibodies were used to detect proteins.

Bimolecular fluorescence complementation assay

The N-terminal half of GFP (GFP N) was fused with AaAtg8 and AaAtg1 and cloned into pHygroGPE1 to yield a GFP N -AaAtg8 plasmid and a GFP N -AaAtg1 plasmid, respectively. The C-terminal half of GFP (GFP C) was fused to AaAtg4 and cloned into pNeoGPE1 to yield a GFP C -AaAtg4 plasmid. Plasmids were co-transformed into wild-type protoplasts, and transformants were selected on RMM amended with 200 μg/ml hygromycin and 300 μg/ml geneticin (G418), confirmed by PCR with gene-specific primers, and observed by a ZOE Fluorescent Cell Imager (Bio-Rad) microscope. A fungal strain carrying a truncated GFP N and a truncated GFP C was constructed as a negative control.

AaAtg8 lipidation assay

To determine the attachment of PE to AaAtg8, a process known as lipidation (Ichimura et al., 2000), during autophagy, an HA tag was added to AaAtg8, which was then cloned into a pNeoGPE1 plasmid and introduced into wild-type protoplasts to yield a WT/HA-AaAtg8 strain. Fungal proteins were separated by SDS-PAGE containing 6 M urea as described by Hirata et al. (2017), and HA-AaAtg8-PE was detected by Western blots using an anti-HA antibody. The HA-AaAtg8-PE protein with a higher hydrophobicity would migrate faster than the HA-AaAtg8 in the SDS-PAGE gel (Mizushima et al., 2011).

Fluorescence microscopy

The wild-type strain co-expressing GFP-AaAtg4 and mCherry-AaAtg8 was examined using a ZOE Fluorescent Cell Imager (Bio-Rad) microscope with the appropriate fluorescence filter set. Green fluorescence of sGFP was excited at 488 nm and detected at 510 nm. Red fluorescence from mCherry was examined at 543/560–615 nm (Choo et al., 2023). Vacuoles were stained with 100 μM 7-amino-4-chloromethylcoumarin (CMAC, Fisher Scientific) and visualized with a 353/466 nm filter set. The intensity of green fluorescence was measured by ImageJ.

Statistical analysis

Statistical evaluation was performed using one-way ANOVA, and treatment differences were assessed with Tukey’s HSD post hoc test, applying a significance threshold of p < 0.05.

AaAtg4 is required for autophagy formation in response to starvation and H 2 O 2

Autophagy was evaluated by transforming a plasmid carrying a synthetic GFP translationally fused with the N-terminus of an AaAtg8 protein into the wild-type and the Δ AaAtg4 strains, and the resultant transformants (WT/GFP-AaAtg8 and Δ AaAtg4 /GFP-AaAtg8) were examined by fluorescence microscopy. To determine the optimal window for capturing cellular changes, a time-course study was conducted to monitor hyphae at 2, 4, 6, and 8 h. The results revealed that 8 h was the peak time point for visualizing both vacuoles and the accumulation of fluorescent puncta (data not shown). WT/GFP-AaAtg8 and Δ AaAtg4 /GFP-AaAtg8, after being shifted to PDB, produced nearly uniformly green fluorescence in the hyphae (Figure 1A). After shifting to MM-N (minimal medium without nitrogen) or PDB containing 10 mM H 2 O 2 for 8 h, green fluorescence in the WT/GFP-AaAtg8 hyphae became bright spots, indicating the formation of autophagosomes (A), aligned well with the vacuoles, indicating the formation of autophagic vacuoles (AV). In contrast, green fluorescence excluding vacuoles was observed in the cytoplasm of Δ AaAtg4 /GFP-AaAtg8 after shifting to MM-N or H 2 O 2. Western blot analyses were performed to detect protein degradation by releasing free GFP from GFP-AaAtg8 during autophagy (Figure 1B). No free GFP was detected in the WT/GFP-AaAtg8 strain after being shifted to PDB for 4 h. However, it showed 71% and 55% free GFP after shifting to MM-N or H 2 O 2, respectively. No free GFP was detected in the Δ AaAtg4 /GFP-AaAtg8 strain after shifting to either of the conditions.

AaAtg4 interacts with AaAtg8 but not AaAtg1

Y2H assays were conducted to determine if AaAtg4 would interact with AaAtg1 and AaAtg8. Yeast cells harboring AaAtg4, either as bait or prey, resulted in successful growth with a blue colony when co-cultured with cells harboring AaAtg8 on the QDO/X/A medium containing aureobasidin A and X-α-Gal (Figure 2A). Yeast cells harboring AaAtg4-BK (bait) were co-cultured with AaAtg1-AD (prey) and failed to form a colony, however. Yeast cells, after pairing p53 with T (positive control), formed a colony that was intensely blue. All negative controls showed no sign of yeast growth. A direct interaction between AaAtg4 and AaAtg8 was confirmed by the Co-IP assay, as well as the BiFC assay using split GFP. For the Co-IP assay, plasmids pNeoGPE1-HA-AaAtg4 and pCB1532-GFP-AaAtg8 were individually or co-transformed into wild-type protoplasts to yield three different fungal strains. Proteins were purified from the three strains and detected with anti-HA, anti-GFP, or anti-actin antibodies. The result revealed that AaAtg4 directly interacted with AaAtg8 (Figure 2B). For the BiFC assay, a GFP N -AaAtg8 plasmid (in pHygroGPE1) was constructed to carry a full-length AaAtg8 fused with the C-terminus-lacking GFP. A GFP C -AaAtg4 plasmid carried a full-length AaAtg4 fused with the N-terminus-lacking GFP. Co-transformation of GFP N -AaAtg8 and GFP C -AaAtg4 plasmids into protoplasts prepared from the wild type yielded fungal strains that emitted green fluorescence (Figure 2C). Co-transformation of GFP N -AaAtg1 and GFP C -AaAtg4 (or truncated GFP N and GFP C) into the wild type failed to yield transformants exhibiting green fluorescence.

H 2 O 2 affects AaAtg4 protease cleavage of the AaAtg8 C-terminus

To test the AaAtg4 protease activity, GFP was translationally fused with the C-terminus of AaAtg8 and transformed into wild-type and Δ AaAtg4 protoplasts to yield WT/AaAtg8-GFP and Δ AaAtg4 /AaAtg8-GFP, respectively. Western blot analyses revealed that AaAtg8-GFP was converted entirely to free GFP in the WT/AaAtg8-GFP strain after shifting to PDB, MM-N, or 10 mM H 2 O 2 (Figure 3A). However, no free GFP was detected in the Δ AaAtg4 /AaAtg8-GFP strain after shifting to any of the conditions. When the WT/AaAtg8-GFP strain was transferred to a higher concentration (20 or 30 mM) of H 2 O 2 for 4 h, low-level AaAtg8-GFP was detected (Figure 3B).

H 2 O 2 suppresses the AaAtg4/AaAtg8 interaction but promotes AaAtg8 lipidation and autophagy

The BiFC strain carrying both GFP N -AaAtg8 and GFP C -AaAtg4 constructs was used to test the effect of H 2 O 2 on the binding between AaAtg4 and AaAtg8 (Figure 4A). After shifting to PDB or MM-N, the strain emitted strong green fluorescence. However, after shifting to PDB amended with H 2 O 2, the strain emitted weak green fluorescence. The intensity of green fluorescence decreased significantly as the H 2 O 2 concentration increased (Supplementary Figure 2). Further Western blot tests were conducted to investigate the effect of H 2 O 2 on the formation of AaAtg8-PE. No HA-Atg8-PE was detected in the WT/HA-Atg8 strain carrying an HA tag fused with the N-terminus of AaAtg8 after shifting to PDB for 2 h (Figure 4B) or 4 h (data not shown), revealing that most AaAtg8 proteins were not conjugated with PE even in the presence of H 2 O 2. Low-level HA-Atg8-PE was detected in the strain after shifting to PDB amended with 5, 10, or 30 mM H 2 O 2. A high level of HA-AaAtg8-PE was detected after shifting to PDB amended with 20 mM H 2 O 2. The occurrence of autophagy was assessed by releasing free GFP in the WT/GFP-AaAtg8 strain (Figure 4C), showing that free GFP reached the highest level in the strain after shifting to PDB amended with 10 mM H 2 O 2 for 4 h and that the free GFP level decreased after shifting to 20 and 30 mM H 2 O 2 in a concentration-dependent manner. No HA-AaAtg8-PE was detected in the Δ AaAtg4 /HA-AaAtg8 strain in PDB-amended H 2 O 2, indicating the requirement of AaAtg4 in the formation of AaAtg8-PE (data not shown). All experiments were repeated three times, showing similar patterns.

H 2 O 2 impacts AaAtg4 phosphorylation

The WT/HA-AaAtg4 strain was used to assess AaAtg4 phosphorylation in response to starvation and H 2 O 2. Western blot analyses revealed that the strain accumulated more AaAtg4 protein after shifting to MM-N than after shifting to PDB or to PDB amended with 30 mM H 2 O 2 (Figure 5A). Phos-tag analyses showed that the level of AaAtg4 phosphorylation was higher after the strain was shifted to MM-N than to PDB or PDB amended with 30 mM H 2 O 2, which correlated with autophagy formation as assessed by releasing free GFP from GFP-AaAtg8 (Figure 5B). Further tests revealed that H 2 O 2 at 10 mM, but not at 20 or 30 mM, could increase the level of the AaAtg4 protein and its phosphorylation (Figure 5C). AaAtg4 phosphorylation was unaffected in the Δ AaAtg1 /HA-AaAtg4 strain with the AaAtg1 deletion (Figure 5D).

AaAtg4 interacts with the mitogen-activated protein kinase AaHog1

Co-IP assays revealed that AaAtg4 interacted with AaHog1 (Figure 6A). Phos-tag assays revealed that AaHog1 was phosphorylated in response to starvation and 30 mM H 2 O 2 (Figure 6B). H 2 O 2 elevated AaHog1 phosphorylation in a concentration-dependent manner, showing that increasing concentrations of H 2 O 2 led to elevated AaHog1 phosphorylation (Figure 6C). Phos-tag assays in the Δ AaHog1 mutant revealed that AaAtg4 was phosphorylated to high levels in the absence of H 2 O 2, in the presence of 30 mM H 2 O 2, but barely phosphorylated in the presence of 10 mM H 2 O 2 (Figure 6D), patterns that were opposite to those of wild type (Figure 5C). Western blot analysis revealed that the Δ AaHog1 mutant expressing GFP-AaAtg8 showed free GFP formation in response to starvation and H 2 O 2 at efficiencies similar to the wild-type strain (Figure 6E).

AaAtg4 and AaAtg8 are transported into vacuoles during autophagy maturation

To determine the localization of AaAtg4 corresponding with AaAtg8 during autophagy, a fungal strain (WT/GFP-AaAtg4+mCherry-AaAtg8) was created by co-transforming a pNeoGPE1 plasmid carrying a GFP fused with AaAtg4 and a pHygroGPE1 plasmid carrying an mCherry protein fused with AaAtg8 into wild-type protoplasts. Fluorescence microscopy analyses revealed that the strain, after shifting from PDB to PDB for 6 or 24 h, emitted uniform green and red fluorescence, excluding vacuoles (Figure 7), indicating that AaAtg4 and AaAtg8 were localized in the cytosol during non-autophagic conditions. The vacuoles were identified by staining hyphae with CMAC. After the strain was shifted from PDB to MM-N for 24 h, but not 6 h, uniform green or red fluorescence in the cytoplasm, including vacuoles, was observed, indicating that some of AaAtg4 and AaAtg8 were transported into vacuoles during the late stage of autophagy.

Atg4 is a cysteine protease essential for autophagosome formation through the modification and recycling of Atg8 (Nakatogawa et al., 2012). We have found that a redox regulatory mechanism tightly controls AaAtg4 activity and autophagosome formation. In yeast, Atg4 activation involves a redox-sensitive disulfide bond mediated by thioredoxin (Pérez-Pérez et al., 2014, Pérez-Pérez et al., 2016).

AaAtg4 is activated, highly phosphorylated, and interacts with AaAtg8 under starvation. AaAtg4 is required for the processing and activation of AaAtg8, enabling autophagosome formation and cargo degradation. As shown in Figure 8A, AaAtg4 functions to remove arginine in the C-terminus of Atg8 to expose its glycine (G) residue, which can be attached to PE to form an Atg8-PE conjugate. At the late stage, AaAtg4 functions to release AaAtg8 from the membrane of autophagosomes. Without AaAtg4, the autophagic machinery fails to respond to starvation or oxidative stress, as evidenced by both microscopy and biochemical assays. AaAtg4 specifically interacts with AaAtg8, but not AaAtg1, suggesting that its role in autophagy is linked to AaAtg8 processing and recycling rather than initiation. AaAtg4 likely acts as a cysteine protease, exposing a glycine residue on AaAtg8 for lipidation and aiding autophagosome membrane expansion. BiFC assays reveal a strong interaction between AaAtg4 and AaAtg8, which is disrupted in a dose-dependent manner by H 2 O 2, possibly due to disulfide bond formation in AaAtg4. In yeast, Atg1 phosphorylates Atg4 (Sánchez-Wandelmer et al., 2017), reducing its ability to cleave Atg8 and preserving Atg8-PE for autophagosome formation. However, AaAtg4 does not interact with AaAtg1, suggesting that it acts downstream of, or independently of, the initiation complex. Its phosphorylation is unaffected by AaAtg1 deletion and does not correlate with AaAtg8-PE levels. The functional impact of AaAtg4 phosphorylation is still unclear. Beyond phosphorylation, AaAtg4 activity may be modulated by other post-translational modifications such as O-GlcNAcylation (Jo et al., 2016), S-nitrosylation (Li et al., 2017), and ubiquitination (Kuang et al., 2012; Sun et al., 2024). Moreover, various kinases can phosphorylate Atg4 (Ni et al., 2018; Pengo et al., 2018; Li et al., 2021).

Autophagy is activated by redox via Atg4 modification in mammals and yeast (Scherz-Shouval et al., 2007; Pérez-Pérez et al., 2014; Qiao et al., 2015). In this study, we examined how H 2 O 2 differentially impacts AaAtg4 phosphorylation, its proteolytic activity, the interaction with AaAtg8, the formation of AaAtg8-PE, and autophagy (Figure 8B). AaAtg4 activity is finely tuned by redox conditions, with H 2 O 2 acting as a modulator that activates at low concentrations and inhibits at high concentrations. The presence of moderate H 2 O 2 (10 mM) enhances AaAtg4 activity, thereby promoting cleavage. In contrast, high H 2 O 2 inhibits AaAtg4 activity, likely due to oxidative damage or modification of its catalytic cysteine, which can encourage the formation of a disulfide bond (Scherz-Shouval et al., 2007; Zhou et al., 2022). Different concentrations of H 2 O 2 affect the formation of AaAtg8-PE. H 2 O 2 acts as a trigger for AaAtg8 lipidation in A. alternata, likely by inducing oxidative stress that activates autophagy pathways. H 2 O 2 at 20 mM appears to be the optimal concentration for inducing AaAtg8-PE formation, while both lower and higher concentrations are less effective. This suggests a dose-dependent response, with a peak at 20 mM, possibly reflecting a balance between stress induction and cellular tolerance. The requirement for H 2 O 2 in AaAtg8 lipidation underscores the central role of oxidative stress in initiating autophagy. Moreover, H 2 O 2 has a dose-dependent effect on AaAtg4 accumulation and phosphorylation, as well as autophagy. H 2 O 2 enhances AaAtg4 levels and phosphorylation and promotes autophagy at moderate levels (10 mM), but inhibits it at higher concentrations. While 10 mM H 2 O 2 optimally induces autophagy, the peak AaAtg8-PE level occurs at 20 mM, indicating that AaAtg8-PE accumulation does not directly correlate with autophagy activation. Moreover, 10 mM H 2 O 2 increased both AaAtg4 abundance and phosphorylation but did not promote AaAtg8–PE accumulation. Collectively, these results reveal that H 2 O 2 exerts distinct effects on different autophagic processes, highlighting the complexity of autophagy regulation. High levels of H 2 O 2 may suppress AaAtg4 activity, allowing the accumulation of AaAtg8-PE. The autophagy process is susceptible to oxidative stress: mild stress triggers autophagy as a protective response, whereas excessive stress suppresses it, suggesting threshold-dependent regulation. Time is also a critical factor influencing autophagy, given its dynamic nature. In this study, our primary objective was to elucidate the impact of H 2 O 2 on autophagy, with particular emphasis on Atg4 modification. Accordingly, time-course experiments were conducted to determine the optimal time points for fluorescence imaging and Western blot analysis. The reported results reflect these optimized conditions.

AaAtg4 may be involved in stress signaling through its interaction with AaHog1, as demonstrated by Co-IP assays. AaHog1, phosphorylated under starvation and oxidative stress (30 mM H 2 O 2), appears to regulate AaAtg4 phosphorylation, suppressing it under certain conditions. This regulation is dose-dependent and specific to oxidative, not reductive, stress. In Δ AaHog1 mutants, AaAtg4 becomes hyperphosphorylated under conditions less conducive to autophagy, suggesting that AaHog1 acts as a negative regulator. Despite their interaction, AaHog1 is not essential for autophagy marker processing, indicating that autophagy can proceed independently or via compensatory pathways in its absence. AaHog1 may influence autophagy indirectly or under specific conditions. Additionally, the MoAtg4 homolog is linked to the mitogen-activated protein kinase-mediated pathway in the rice blast fungus, Magnaporthe oryzae. Phosphorylation of MoAtg4 by the mitogen-activated MoKK1 kinase involved in cell wall integrity abolishes the C-terminal cleavage of MoAtg8 (Guo et al., 2024).

Under non-autophagic conditions, both AaAtg4 and AaAtg8 proteins remain in the cytoplasm and are not targeted to vacuoles. During the late stage of autophagy, some of AaAtg4 and AaAtg8 are relocated from the cytosol into vacuoles, likely for degradation or recycling. The co-localization and vacuolar transport of AaAtg4 with AaAtg8 during autophagy imply a functional interaction between these proteins in the autophagic process. Since AaAtg4 is a cysteine protease involved in Atg8 processing, its vacuolar transport may reflect a regulatory mechanism or turnover during autophagy progression.

This study reveals that the addition of H 2 O 2 can disrupt intracellular redox balance and that AaAtg4 plays a multifaceted and indispensable role in A. alternata, bridging autophagy regulation, redox sensitivity, and ACT toxin biosynthesis. Unlike AaAtg8, which is dispensable for ACT production, AaAtg4 deletion disrupts toxin maturation, leading to the accumulation of biosynthetic intermediates and altered virulence profiles. Mechanistically, AaAtg4 functions as a cysteine protease essential for AaAtg8 processing and autophagosome formation, with its activity finely tuned by oxidative and reductive signals. H 2 O 2 differentially modulates AaAtg4 activity, phosphorylation, and AaAtg8 binding, thereby regulating AaAtg8 lipidation/delipidation and autophagy in a concentration-dependent manner. Interestingly, the threshold required for AaAtg4 phosphorylation did not correlate with that for AaAtg8–PE formation or GFP–AaAtg8 turnover. Moderate H 2 O 2 levels enhance AaAtg4 function and autophagy, whereas excessive oxidative stress inhibits both, highlighting threshold-dependent regulation of redox balance. The interplay between AaAtg4 and AaHog1 further highlights a stress-responsive signaling axis in which AaHog1 modulates AaAtg4 phosphorylation under oxidative conditions. Collectively, these findings position AaAtg4 as a central node that integrates autophagy, redox homeostasis, H 2 O 2 resistance, and secondary metabolism, offering new insights into fungal stress adaptation and virulence.