M. oryzae Guy11, Js153, and the eGFP-tagged Zhong1 (a gift from Dr. W-M Wang) strains were used for rice infection. M. oryzae strains were first inoculated on CM medium (complete medium) that grew at 28^°C under a 12/12 (light/dark) condition. Spores were collected 2 weeks after inoculation, which were resuspended to 1 × 10⁵ spores ml^–1. Spores were spray-inoculated on three-leaf-stage rice seedlings. Disease symptoms were examined 5 days later. Genomic DNA was extracted from inoculated leaves for fungal accumulation examination (MoPot2; Supplementary Table 1). Disease resistance was also examined by using detached leaves from four-leaf-stage rice. 10 μl spores (1 × 10⁵ spores ml^–1) were added at two spots of each leaf, and kept in a culture dish with a wet filter. After incubation in a growth chamber at 28^°C for 24 h in dark, the leaves were kept under a 12/12 h (light/dark) light rhythm for 6 days till disease symptoms were examined. For leaf sheath inoculation, leaf sheaths were prepared from four-leaf-stage rice. 10 μl spores (1 × 10⁵ spores ml^–1) were inoculated. Fungal hyphae development was observed under a microscope at 24 and 48 h post inoculation (hpi), respectively.

Total RNA extraction was performed as described previously (Zhang et al., 2018). In brief, the inoculated plants were used for RNA extraction at 0, 24, and 48 hpi. Total RNA was extracted using TRIzol reagent (Invitrogen, United States) following the manufacturer’s protocol. RNA was resolved on a 14% denaturing 8 M urea-PAGE gel and then transferred and UV cross-linked onto a Hybond N⁺ membrane (GE Healthcare Life Science, Beijing, China) using UV light. miRNA probes were end-labeled with [γ-³²P] ATP by T4 polynucleotide kinase (New England Biolabs, Beijing, China). Expression levels were quantified using ImageJ as instructed.

For generating OsAPX1 overexpression transgenic plants, the full-length CDS was cloned into a pCAM1300 vector driven by a CaMV35S promoter. For the OsAPX1 silencing mutant, highly specific target regions from the OsAPX1 CDS were cloned to the pYLCRISPR/Cas9Pubi vector. The construct was transferred to Agrobacterium strain EHA105, which was used for transgenic rice production. For verifying the OsAPX1 silencing mutant, genomic DNA was used to examine the OsAPX1 genomic DNA sequence. For verifying OsAPX1 over-expression plants, total protein samples from the transgenic plant were used for western blot; and total RNA was used to examine the transcript level of OsAPX1 by qRT-PCR.

1 μg total RNA was reverse transcribed into cDNA by using PrimeScript RT reagent Kit (Takara, Japan). The qRT–PCR was performed in 15 μl of reaction mixture consisting 1.5 μl 10 × SYBR Green (Invitrogen, United States), 1.5 μl PCR buffer, 0.3 μl 10 mM dNTPs (Takara, Japan), 0.3 μl Taq, 0.3 μl ROX DYE2 (Vazyme, China), 1.5 μl 2 mM each primer, and 2 μl appropriate diluted cDNA. The conditions for real-time RT-PCR were as follow: 94^°C for 3 min, then 40 cycles at 94^°C for 30 s and 58^°C for 30 s followed by 72^°C 35 s for PCR amplification. Transcript levels of each gene were measured by the Applied Biosystems 7500 (Applied Biosystems, United States) according to the manufacturer’s instructions. The data were normalized to the amplification of the rice 18sRNA gene. Real-time PCR primer sequences are available in Supplementary Table 1.

Leaf tissues and leaf sheaths were dipped into 50 ml solution containing 50 mg Diaminobenzidine (DAB), 25 μl Tween-20 and 2.5 ml 200 mM Na₂HPO₄ and vacuum infiltrated for 30 min followed by staining in dark at room temperature (25^°C) overnight (10 h). The tissues were decolorized in 1:1:1 (v/v/v) acetic acid-ethanol-glycerol solution for 15–20 min at 90–95^°C and visualized afterward (ThordalChristensen et al., 1997). Decolored leaves or leaf sheath tissues were examined for H₂O₂ accumulation around the inoculating loci by using a microscope.

The free SA concentration in transgenic rice was measured as described (Liu et al., 2014). The rice tissues were homogenized in liquid nitrogen and then suspended in 90% (v/v) methanol. As an internal standard, 100 mg 3-hydroxy benzoic acid in 100% methanol was added to each sample. The SA solution was filtered and separated on a C18 analytical column using HPLC and detected using fluorescence (excitation at 305 nm, emission at 405 nm; Waters). The HPLC was programmed for isocratic conditions with a flow rate of 0.5 ml/min. The concentration of SA was quantified by area integration of the HPLC peaks.

Leaf tissues were snap-frozen in liquid nitrogen and ground into fine powder. The samples were added with 2 × SDS loading buffer, which was boiled at 100°C for 10 min. The supernatant, after 10,000 g centrifugation, was separated by 12% SDS-PAGE gels at 100 V for 1.5 h. The proteins were transferred to PVDF membrane (Bio-RAD, United States), blocked by using 5% dry milk for 30 min, which was followed by Flag antibody (Abmart, China) incubation for 2 h. The membranes were washed by using TBST buffer three times (5 min), followed by 2nd antibody incubation (Abmart, China) for 2 h. The protein signal was detected by chemoluminescence (Tanon, China).

The APX enzyme activity was examined by using a kit (Beijing Solarbio Science and Technology Co, Beijing, China) measuring the oxidation rate of ascorbic acid within 2 min with a spectrophotometer at 290 nm.

We challenged 4-week-old rice [the Japonica cultivar Nipponbare (NIP)] with the M. oryzae strain Guy11 (compatible strain) and Js153 (incompatible strain) (Supplementary Figure 1A). Compared with the mock, OsAPX1 expression was significantly induced at 24, 48, and 72 hpi after infection with both strains (Supplementary Figure 1B). At 72 hpi, OsAPX1 expression increased more than 3 folds. To examine the contribution of OsAPX1 in rice immunity against the blast disease, we first investigated whether an elevated OsAPX1 expression would lead to an altered rice blast resistance. Both OsAPX1 silencing (cas9-osapx1 #22 and #30; Supplementary Figure 2A) and over-expression (OsAPX1-OE #38 and #39; Supplementary Figures 2B,C) transgenic rice were constructed. Genotyping results indicated that cas9-osapx1 is a homozygous insertion mutant (Supplementary Figure 2A) with a T/C insertion to the 40th nucleotide of the second exon, leading to the complete silencing of OsAPX1 due to frameshifting. Cas9-osapx1 lines carried a 1-bp insertion causing protein truncation (Supplementary Figure 2A). Compared to NIP and cas9-osapx1 #22 rice, the OsAPX1-OE #38 rice were significantly higher in growth length, had longer roots (Supplementary Figure 2D), larger seed size (Supplementary Figure 2E), and greater seed weight (per 1,000 seeds) (Supplementary Figure 2F). In contrast, the cas9-osapx1 #22 rice was shorter and produced smaller seeds both in size and weight. However, neither measurable growth or developmental abnormality nor spontaneous lesions were observable on the leaf surface of the transgenic rice (Supplementary Figure 2D). There was no discernable difference in ROS accumulation between wild type and transgenic rice under normal growth conditions either (Supplementary Figure 2G), indicating that the rice innate immunity is not automatically activated without pathogen infection.

Detached leaves from transgenic plants were challenged by punching inoculation with Guy11 spores. The OsAPX1-OE #38 leaves developed less severe disease symptoms, manifested by significantly smaller necrosis size and lesser discoloration as observed on transgenic leaves when compared to leaves from control plants. Meanwhile, the lesion size on the cas9-osapx1 #22 leaves was significantly larger than that on the NIP leaves (Figure 1A). The propagation of M. oryzae on the infected leaves was quantified by qRT-PCR using primers specific to MoPot2, a M. oryzae housekeeping gene. In agreement with the resistant phenotype, less hyphae propagation was detected on leaves from the OsAPX1-OE #38 than the control, whereas the accumulation of hyphae increased on cas9-osapx1 #22 (Figure 1B), indicating that over-expression of OsAPX1 led to enhanced resistance to M. oryzae infection.

The association between OsAPX1 expression level and disease resistance was further confirmed by Guy11 spore-spray inoculation on rice leaves. From 96 hpi on, both NIP and OsAPX1-OE or cas9-osapx1 rice developed typical blast disease symptoms such as scattered lesions on the leaf surface, cell death in the center of some lesions, and chlorosis on some leaves. Specifically, more lesions developed on the cas9-osapx1 leaves, accompanied by severer chlorosis, than NIP rice. In contrast, both lesion numbers and chlorosis were significantly milder on the OsAPX1-OE (#38, #39) leaves, when compared to both cas9-osapx1 (#22, #30) and NIP rice (Figure 1C). When hyphae growth was quantified, we detected more fungal hyphae on cas9-osapx1 (#22, #30) rice than both NIP and OsAPX1-OE (#38, #39) rice, whereas OsAPX1-OE (#38, #39) had the least hyphae development among them (Figure 1D). Our results indicate that OsAPX1 over-expression led to enhanced resistance against blast fungus infection. When rice sheath epidermal tissue was challenged with an eGFP-labeled M. oryzae strain (Zhong-1), we were able to observe and assess infection progress by quantifying the rate of appressorium formation, hyphae development, and invasive hyphae spreading (Figure 1E). At 24 hpi, 27% appressorium developed visible hyphae in cas9-osapx1 #22 sheath epidermal cells, and 0% hyphae development was recorded on either OsAPX1-OE #38 or NIP. At 48 hpi, with hyphae fully developed within the infected cells in all plants, about 60% hyphae were observed to spread to the adjacent cas9-osapx1 #22 sheath epidermal cells, and about 18% to the 3rd cells; on NIP rice, only about 45% of fungi spread to the adjacent cells and about 13% to the 3rd cells; on OsAPX1-OE #38 rice, only about 20% fungi spread to adjacent cells but none of them spread to a 3rd cell (Figure 1F). Our results indicate that OsAPX1 expression level is positively related to rice ability resisting fungal hyphae development in infected tissue.

It was reported that the OsAPX1 promoter is bound by transcription factor INDETERMINATE SPIKELET1 (OsIDS1), which inhibits the expression of OsAPX1 (Cheng et al., 2021). OsIDS1 is a target of miR172a, which silences OsIDS1 by reverse-complementary sequence match (Cheng et al., 2021). To check whether the induced expression of OsAPX1 is a result of variated expression of miR172a, we examined miR172a expression by a reverse complementary DNA probe through northern blot. We found that the expression of miR172a was induced significantly at 24 and 48 hpi (Figure 2A). The expression level of OsIDS1 decreased at both 24 and 48 hpi, while the expression of OsAPX1 increased at both 24 and 48 hpi, correspondingly (Figure 2B). If miR172a expression was inhibited (miR172a-KO), the expression of OsAPX1 did not significantly change along with M. oryzae infection further (Figure 2B).

We inoculated Guy11 spores on the detached leaves of both wild type and miR172a transgenic rice (a gift from Cheng) (Cheng et al., 2021). The lesions on the miR172a-OE rice leaves were significantly smaller than the WT rice. In contrast, miR172a-KO leaves developed significantly larger lesions than WT rice. Cell death and chlorosis were also very obvious on miR172a-KO leaves (Figure 3A). Guy11 developed more fungal mass on the miR172a-KO rice, while much lesser on the miR172a-OE leaves than the WT rice (Figure 3B). When the rice was spray-inoculated with Guy11 spores, there was almost no lesions development on miR172a-OE rice leaves, whereas miR172a-KO rice exhibited much more lesions than both the WT and miR172a-OE rice (Figure 3C). Quantification of hyphae in leaves also indicated that Guy11 was spread much more in tissues of miR172a-KO rice than the WT and miR172a-OE rice (Figure 3D). Compared with mock, the OsIDS1 and OsAPX1 gene expression levels were not significantly changed after M. oryzae infection in the miR172a-KO plant, but in the miR172a-OE plant, the expression of OsIDS1 reduced significantly and OsAPX1 was induced significantly (Figures 3E,F). In summary, our results indicate that OsAPX1 is subject to a miR172a-OsIDS1 regulatory module upon M. oryzae infection.

ROS production is a hallmark of plant early defense responses, which represents a successful pathogen recognition and the activation of plant defense response (Torres, 2010; Nie et al., 2017; Qi et al., 2017; Zhang et al., 2018). We checked the ROS accumulation in both transgenic and control plants. We found that ROS accumulation in OsAPX1-OE #38 and NIP plants could be observed around infection sites at 12 and 36 hpi, then reduced at 60 and 72 hpi. OsAPX1-OE #38 rice obviously accumulated more ROS at the penetrating sites than the NIP rice, whereas it was the least observed in cas9-osapx1 #22 rice (Figure 4B). The observation that Guy11 infection led to more ROS accumulation in OsAPX1-OE #38 rice than in WT rice challenges our current knowledge that OsAPX1 eliminates, instead of, induces ROS accumulation. However, numerous groups have reported that elevated APX expression is associated with ROS accumulation (details in discussion). We, therefore, hypothesized that there must be a mechanism that ROS accumulation and scavenging coalesce.

Diphenyleneiodonium (DPI) is a flavoenzyme inhibitor that prevents the activation of NADPH oxidases necessary for ROS generation in plants (Cross and Jones, 1986; Li et al., 2020a). We found that ROS accumulation around the infection loci was dramatically reduced in OsAPX1-OE #38 rice after DPI treatment (Figure 4C), suggesting that these ROS may be produced by OsRBOHs.

OsRBOHs are crucial components in ROS accumulation upon rice M. oryzae infection (Dangol et al., 2019). OsRBOHs reduce cellular oxidation potential by catalyzing the transfer of electrons from NADPH to oxygen (O₂), which generates superoxide radicals (O^∙2–). A previous study reported that the lack of sAPX and tAPX drastically decreased the expression of H₂O₂ responsive genes in Arabidopsis under photooxidative stress (Maruta et al., 2010). Under flood stress, AtSUS1, AtPEPC, AtLDH gene expression increased in Arabidopsis plant that overexpression sponge gourd APX (LcAPX) (Chiang et al., 2017). Therefore, we hypothesized that OsAPX1 affects OsRBOHs gene expression under M. oryzae infection. To verify our hypothesis, we checked the OsRBOHs expression level in transgenic plants at 36 hpi, the time point when ROS accumulation peaked. Compared to NIP, OsRBOHs (especially OsRBOHb) expression was significantly induced in OsAPX1-OE #38 rice (green line) and significantly reduced in cas9-osapx1 #22 rice (red line) (Figure 4A). These results suggest that OsAPX1 may influence ROS production by affecting OsRBOH genes expression when rice is infected by M. oryzae.

The generation and removal of ROS are two parallel activities that maintain cellular ROS homeostasis (Li et al., 2016). We found that the expression of OsAPX1 (ROS removal) was constantly induced after M. oryzae infection, but ROS accumulation was temporal. Therefore, we speculated that ROS generation gene OsRBOHs expression is temporally induced. Indeed, the time-course experiment showed that OsRBOHs gene expression level was temporal during M. oryzae infection. At 0 hpi, OsRBOHb, OsRBOHe, and OsRBOHi expression were detectable, among which OsRBOHe is the major contributor of expressed OsRBOHs. Upon M. oryzae infection, OsRBOHb became a major contributor, which was induced at 12 hpi and peaked at 36 hpi. From 60 hpi on, expression of OsRBOH genes declined, especially OsRBOHb. It should be noticed that at the same infection period the expression of OsAPX1 was also dramatically induced (Figure 4A). The result indicates that OsAPX1 and OsRBOHb were simultaneously induced at the early stage after M. oryzae infection; at the later stage, the expression of OsAPX1 was constantly induced; however, OsRBOHb was declined. We also checked APX enzyme activity in NIP and transgenic plants. We found that APX enzyme activity was elevated both in the early and late-stage after M. oryzae infection (Supplementary Figure 3). In summary, these results showed that ROS generation activity masked ROS scavenging activity at the early stage after M. oryzae infection, which led to ROS accumulation; however, ROS scavenging activity prevailed at the later stage, which eventually led to ROS removal.

ROS are crucial signal molecules that can activate phytohormone signaling pathways such as the salicylic acid and jasmonic acid pathway (Torres, 2010; Xu et al., 2015), which play important roles in plant innate immunity (Liu et al., 2016; Hong et al., 2019). To further investigate the mechanism underlying OsAPX1-mediated disease resistance against rice blast, we examined the expression of several key SA and JA signaling pathway genes. OsPR1a and OsPR1b are important SA signaling pathway reporter genes, which are induced by many pathogen infections. Our results revealed that in OsAPX1-OE rice the expression of OsPR1a was around 3-fold and 9-fold higher than that in NIP, whereas OsPR1b expression was around 2-fold and 4-fold higher than that in NIP (Figure 5A), indicating that SA signaling pathway was activated when OsAPX1 is over-expressed. Genes involved in SA synthesis and signal transduction were also checked. As shown in Figure 5A, genes associated or directly participated in SA synthesis such as OsPAD4, OsICS1, and OsPAL1 were significantly induced in the OsAPX1-OE rice, among which OsPAD4 showed an around the 3- and 1.5-fold increase. Expression of genes involved in SA signaling transduction, such as OsNPR1 and OsWRKY45, was also increased in OsAPX1-OE rice (Figure 5A). The expressions of these genes were reduced in the cas9-osapx1 rice (Figure 5A). Taken together, our results indicate that components involved in SA synthesis, signaling transduction are induced upon OsAPX1 over-expression, which may contribute to enhanced disease resistance.

To confirm the relationship between OsAPX1 and the enhanced expression of SA signaling pathway components, we further examined in vivo SA concentration in WT and transgenic plants by HPLC analysis. Free SA concentration was about 10 μg in each gram of fresh tissue (μg/FW) in NIP and was more than 15 μg/FW in OsAPX1-OE rice, which is about 1.5 times as much as in the NIP rice. In contrast, SA content in cas9-osapx1 rice was about 1.72 μg/FW, which is much lower than in the NIP rice (Figure 5B).

We also examined the expression of OsPDF1.2, the reporter gene of the JA signaling pathway. The expression of OsPDF1.2 was not changed measurably in OsAPX1-OE or cas9-osapx1 rice (Supplementary Figure 4), indicating that the JA signaling pathway was not affected by OsAPX1. In agreement, the expression of most key components participating in JA biosynthesis (e.g., OsLOX5, and OsAOS2) and signal transduction (e.g., OsJAZ8, OsCOL1b, OsMYC2) were not significantly changed between control and both OsAPX1-OE and cas9-osapx1 rice (Supplementary Figure 4), indicating that JA signaling pathway is not a major contributor to the OsAPX1-mediated defense against M. oryzae infection.

We also tested whether OsAPX1 responds to another rice disease. R. solani causes rice sheath blight, which is one of the devastating rice diseases. We challenged both transgenic and WT rice with R. solani. Cas9-osapx1 #22 rice exhibited significantly larger lesions and server chlorosis than the NIP rice, whereas OsAPX1-OE #38 rice had smaller lesions and weaker chlorosis (Figure 6A), indicating OsAPX1 may also play a role in resistance to sheath blight. Lesion size quantification clearly demonstrated a positive relationship between OsAPX1 expression level and disease symptoms (Figure 6B). When the rice was challenged by stem-inoculated R. solani, OsAPX1-OE #38 rice exhibited significantly smaller lesions than that from the NIP rice at 15 dpi, whereas cas9-osapx1 #22 leaf sheath had significantly larger lesions than the NIP rice (Figure 6C). Lesion size quantification confirmed our judgment that the expression level of OsAPX1 is positively related to resistance to sheath blight disease (Figure 6D).

Reactive oxygen species (ROS) burst is an important defense response upon pathogen infection, in which APX is supposed to play a ROS scavenger role. It appeared intriguing to us at the very beginning that OsAPX1-OE rice accumulates significantly more H₂O₂ than the WT plants while cas9-osapx1 showed the least H₂O₂ (Figure 4B). We later found that other groups also reported similar observations. For example, it was reported that SuCMoV-infected sunflower leaves demonstrated simultaneously increased expression of APX1 and elevated H₂O₂ accumulation (Rodriguez et al., 2012). In an NPR1-silencing tomato line that is highly resistant to Botrytis cinerea, both APX activity and H₂O₂ accumulation increased (Li et al., 2020b). Applying polyamine to apricot fruits not only enhanced resistance to black spot disease but also induced transcriptional expression of PaAPX and H₂O₂ accumulation (Li et al., 2019b). R. solani-infected beans exhibited both boosted APX activity and H₂O₂ accumulation during its infection (Keshavarz-Tohid et al., 2016). All these reports recorded simultaneous increases in both APX expression and ROS accumulation in multiple species.

When we carefully investigated the origin of ROS, we found that most of the ROS accumulation was demolished in the OsAPX1-OE plant by DPI treatment (Figure 4C). DPI is a flavoenzyme inhibitor that specifically prevents the activation of NADPH oxidases required for ROS generation (Cross and Jones, 1986). RBOHs are membrane-localized NADPH-dependent oxidases that catalyze the production of superoxide from oxygen and NADPH (Kaur et al., 2014; Li et al., 2019a). Increased OsRBOHs expression level very likely led to elevated ROS production. Taken together, we are confident to conclude that OsAPX1 contributes to cellular ROS homeostasis after M. oryzae infection. Our results may also explain the observation made on other APXs. For example, N. benthamiana leaves over-expressing sugarcane APX (ScAPX6) accumulated significantly more H₂O₂ and was more resistant to Fusarium solani var. coeruleum infection (Liu F. et al., 2018).

Our study observed that OsRBOHs genes expression levels increased in the OsAPX1-OE plant. The phenomenon is similar to other studies. For example, overexpressing ScAPX6 in N. benthamiana leaves, result in NtPR1a, NtPR3, and NtEFE26 gene expression levels were increased after Fusarium solani var. coeruleum infection (Liu F. et al., 2018). Transcriptome analysis showed that lignin biosynthesis relative genes induced in Rheum austral APX overexpression line under salt stress (Shafi et al., 2015). However, the mechanism of gene expression level increase needs further study.

We found both OsAPX1 and OsRBOHs were involved in ROS homeostasis management. Both expressions were induced almost simultaneously, but the induced expression of OsRBOHs terminated shortly after M. oryzae infection while the expression of OsAPX1 was sustained. Previous studies reported that plant APX activity increases and more ROS accumulation were found under biotic stress (Rodriguez et al., 2012; Liu F. et al., 2018). It indicates that ROS generation and elimination co-exist at the same time, and ROS accumulation was temporal during the M. oryzae infection (Figure 4B). From 0 to 36 hpi, both OsRBOHs (especially OsRBOHb) and OsAPX1 are induced, during this period, the peroxidase activity is masked by the oxidase activity such that ROS homeostasis leans toward production rather than degradation. After 36 hpi, the expression of OsRBOHb gradually begins to decline but OsAPX1’s peroxidase activity remains strong, which favors ROS degradation over its production. The overlapping expression between OsRBOHs (especially OsRBOHb) and OsAPX1 corresponds very well with cellular ROS accumulation patterns (Figure 4A), manifesting a significant role played by OsRBOHb. This is further supported by the failed ROS accumulation in cas9-osapx1 rice. Although some OsRBOH members expressed normally, or even slightly increased, reduced OsRBOHb expression played a dominant role and led to failed overall ROS accumulation.

Therefore, we concluded that it is the unique expression patterns of OsRBOH and OsAPX1 that is governing the ROS rhythmic generation and elimination upon M. oryzae infection. At the early stage of M. oryzae infection, OsAPX1 and OsRBOHb co-expressed. OsAPX1’s peroxidase activity was masked by OsRBOH activity, which leads to ROS accumulation; at the later stage, OsRBOHb expression declined while OsAPX1 expression remained constantly activated, which led to ROS elimination. This hypothesis was further supported by the DPI treatment that specifically inhibits ROS. DPI treatment functionally mimics the earlier termination of RBOH activity. In OsAPX1-OE rice, DPI treatment destroys RBOH activity such that OsAPX1 scavenger activity was unmasked, demonstrated by the absence of ROS accumulation around the infection loci (Figure 4C). In contrast, when the OsRBOH activity was not offset by the DPI treatment, ROS accumulation peaked at 36 hpi before it dropped.

Our study showed that OsAPX1 can induce OsRBOHs expression after M. oryzae infection. The dynamic OsRBOHb expression was also reported, in which OsRBOHb was induced at 24 hpi but then declined at 48 hpi upon M. oryzae infection (Yang et al., 2017). It was reported that OsRBOHs expression could be regulated by other factors. For example, OsEIL1 binds OsRBOHb promoter and regulates its expression (Yang et al., 2017). OsHXK1 can regulate OsRBOHs gene expression through an unknown mechanism (Zheng et al., 2019); auxin can induce OsRBOHs expression (Zhang et al., 2019). Therefore, we speculate that the OsRBOHs expression pattern in responding to M. oryzae infection is intricately regulated, in which OsAPX1 plays an unstated role.

We previously showed that OsAPX1 proteins were induced by M. oryzae infection as early as 24 hpi (Lin et al., 2018). In this study, we further confirmed that OsAPX1 was induced at the early infecting stage at transcription level as well (Figure 2B and Supplementary Figure 1B), suggesting that OsAPX1 participates in blast resistance from a very early stage. The induced expression of OsAPX1 and other rice APXs has been reported by multiple groups. For example, Agrawal and colleagues revealed that OsAPX1/2 transcripts were up-regulated by M. oryzae infection (Agrawal et al., 2003). OsAPX1/2 were differentially expressed by some defense-related phytohormone treatments (Durner and Klessig, 1995; Chandrashekar and Umesha, 2014). OsAPX1 protein is induced by R. solani in resistant cultivars but not in susceptible cultivars (Ma et al., 2019). OsAPX7 protein was significantly induced by a necrotrophic pathogen, R. solani (Lee et al., 2006). Similarly, OsAPX8 transcription was dramatically induced when rice was infected by Xanthomonas (Jiang et al., 2016). However, the mechanism underlying these inductions is not clear to date.

Our study showed that OsAPX1 expression upon M. oryzae infection is regulated by the miR172a/OsIDS1 module. miR172a has been reported to play a role in immunity, such as in tomato (Solanum lycopersicum), in which over-expressing miR172a and miR172b enhance resistance to Phytophthora infestans by inhibiting the expression of AP2/ERF (Luan et al., 2018). Immunity-related genes expressed significantly higher in miR172b-OE Arabidopsis than in wild type after flg22 treatment (Zou et al., 2020). The above evidence supports the notion that miR172 is a positive immune regulator in multiple plants. In this study, we found that miR172a was significantly induced after M. oryzae infection (Figure 2A), accompanied by reduced expression of OsIDS1, its target gene. Most interestingly, the fluctuated expression of miR172a and OsIDS1 correspond very well with OsAPX1, their downstream target gene (Figure 2B). Our results are consistent with an observation made by Cheng and her colleagues that miR172 was induced by salt stress and contributes to salt stress through miR172a/OsIDS1 module (Cheng et al., 2021). Whether rice resistance/tolerance converges at miR172a/IDS1/OsAPX1 module is worth further exploration.

Other than the direct elimination of pathogens, ROS also serves as a signal molecule that is involved in multiple innate immunity signaling pathways, including SA and JA. SA and JA signaling pathways are an important way to respond to biotic stress (Van Camp et al., 1998; Quan et al., 2008; Torres, 2010; Xu et al., 2015; Tian et al., 2016). In the current study, we were able to show that the JA signaling pathway was not changed measurably (Supplementary Figure 4). Instead, several key components involved in both SA synthesis and signaling were upregulated in OsAPX1-OE plants but were downregulated in cas9-osapx1 plants. OsPAD4, OsPAL1, and OsICS1 are genes involved in SA synthesis, among which OsPAD4 is involved in SA regulation whereas OsPAL1 and OsICS1 are the two key components directly involved in SA synthesis. It was proposed that OsICS1 is an important factor that contributes to most of the induced SA production upon biotic challenge (Wildermuth et al., 2001; Loake and Grant, 2007). Consistently, in OsAPX1-OE plants we observed a remarkable stronger induction of OsICS1 than OsPAL1, indicating OsAPX1 contributes to the risen SA synthesis primarily through OsICS1 induction (Figure 5A). Key components involved in SA signaling such as OsNPR1 and OsWRKY45 were also differentially expressed in OsAPX1-OE plants. In OsAPX1-OE plants, both OsNPR1 and OsWRKY45 branches were up-regulated. In cas9-osapx1 plants, SA signaling pathway components behaved to an opposite profile as in the OsAPX1-OE plants, indicating a reliable connection between OsAPX1 and SA signaling pathway. The induction of OsWRKY45 was slightly stronger than the expression of OsNPR1 (Figure 5A), suggesting that OsWRKY45 is favored by OsAPX1.

We observed that OsAPX1-OE leaves had much higher SA content, whereas cas9-osapx1 leaves had much lower SA contents, compared to the WT plants (Figure 5B). Rice accumulates high basal levels of SA (8–37 μg g^–1 fresh weight) that do not change significantly upon pathogen attack (Silverman et al., 1995), which leads to the misconception that the SA signaling pathway is unrelated to rice defense against blast infection. High basal endogenous SA content does not mean that rice is insensitive to SA signaling. For example, exogenously administrated SA triggers resistance to M. oryzae in adult plants (Iwai et al., 2007). In rice mutant with constantly activated SA signaling pathway (i.e., osmpk15), rice resistance against blast disease is enhanced (Hong et al., 2019). Moreover, synthetic SA analogs such as probenazole, benzothiadiazole (BTH), and tiadinil can induce rice defense response to a wide range of pathogens, ranging from (hemi) biotrophic M. oryzae and bacterial leaf blight pathogen X. oryzae pv. oryzae (Xoo), to necrotrophic root pathogens such as Hirschmanniella oryzae (Shimono et al., 2007; De Vleesschauwer et al., 2008, 2012; Nahar et al., 2012; Xu et al., 2013).

Based on the results, we propose that rice can sense M. oryzae infection and induce the expression of a set of immune regulating factors. miR172a was induced upon M. oryzae infection. The induced expression of miR172a leads to the suppressed expression of OsIDS1, which encodes a transcription factor and enhances OsAPX1 transcription as well. Meanwhile, OsRBOHb was induced at the early stage, but it was reduced at the later stage. At the early stage, OsRBOH activity masked OsAPX activity that results in ROS generation; at the later stage, OsAPX activity was unmasked by OsRBOH activity that leads to ROS elimination. By a delicately regulated sequential expression, OsAPX1 and OsRBOHs manipulate ROS homeostasis temporally. ROS accumulates shortly after M. oryzae infection, which is at the earliest time and the most imminent frontier. After the initial ROS burst, OsAPX1 removes excessive ROS to prevent the rice from ROS toxicity (Figure 7). It would be interesting to investigate whether OsAPX1 plays a role in tolerance against abiotic stresses employing a similar mechanism.