Regulators of nitric oxide signaling triggered by host perception in a plant pathogen

Significance

Understanding belowground communications between plant roots and soil microbes is enormously important for crop production. However, in contrast to plant–beneficial microbe interactions, signaling events underpinning root–fungal pathogen interactions are poorly understood. In this study, we specifically addressed this knowledge gap in Fusarium graminearum, one of the most devastating fungal pathogens of cereal crops throughout the world. We showed that sensing of host signals by F. graminearum before physical contact with roots triggers nitric oxide (NO) production in the pathogen. We identified two proteins that physically interact to regulate NO production and virulence in this pathogen. Our results provide mechanistic insights into host-perception processes that can be manipulated to develop novel plant-protection strategies.

Abstract

The rhizosphere interaction between plant roots or pathogenic microbes is initiated by mutual exchange of signals. However, how soil pathogens sense host signals is largely unknown. Here, we studied early molecular events associated with host recognition in Fusarium graminearum, an economically important fungal pathogen that can infect both roots and heads of cereal crops. We found that host sensing prior to physical contact with plant roots radically alters the transcriptome and triggers nitric oxide (NO) production in F. graminearum. We identified an ankyrin-repeat domain containing protein (FgANK1) required for host-mediated NO production and virulence in F. graminearum. In the absence of host plant, FgANK1 resides in the cytoplasm. In response to host signals, FgANK1 translocates to the nucleus and interacts with a zinc finger transcription factor (FgZC1), also required for specific binding to the nitrate reductase (NR) promoter, NO production, and virulence in F. graminearum. Our results reveal mechanistic insights into host-recognition strategies employed by soil pathogens.

The rhizosphere, the soil region where plant roots and a diverse array of soil microbes known as the root microbiome interact, harbors the largest reservoir of microbial diversity on earth (1). The majority of plant–microbe interactions in the rhizosphere is either beneficial or commensal. Successful initiation of such interactions requires exchange of molecular signals between hosts and microbes even in the absence of any physical contact between the interacting partners. For instance, beneficial interactions of plant roots with bacterial (e.g., rhizobia) (2) or fungal (e.g., arbuscular mycorrhiza) (3) organisms start with the arrival of biologically active compounds such as flavonoids, lipochitooligosaccharides, or strigolactones secreted either by plant roots or microorganisms (4).

The rhizosphere also contains many pathogenic microbes that can pose a significant threat to plant productivity and ecosystem health. However, in a complex environment like the rhizosphere, successful colonization of plant roots by soil pathogens can be dependent on successful sensing of the host plant. Indeed, pathogenic fungi can sense the proximity of plant roots by detecting the presence of various compounds found in plant root exudates. Host sensing triggers a chemotropic response that can direct the movement of the pathogen toward plant roots. A recent example of this phenomenon is the sensing of peroxidases released by wounded tomato roots by the soil-dwelling fungal pathogen Fusarium oxysporum (5). Peroxidase chemosensing requires the synthesis of reactive oxygen species (ROS) by the NADPH oxidase B complex, the G protein-coupled receptor Ste2, and the mitogen-activated protein kinase Mpk1 in F. oxysporum (6). Similarly, after sensing diverse molecules found in root exudates, bacterial pathogens Ralstonia solanacearum and Agrobacterium tumefaciens swim toward host roots using their flagella (7–9).

These examples clearly demonstrate that belowground signal exchanges between plant roots and soil microbes are important for successful interactions. Thus, better understanding of these processes can lead to the design of new strategies to promote beneficial interactions or combat plant diseases. However, in contrast to root–beneficial microbe interactions, very little is known about host-perception and signal-exchange processes involved in plant–soil pathogen interactions (10). It is conceivable, nevertheless, that root-derived signals activate pathogenesis/virulence and developmental processes in soil pathogens. Therefore, interfering with these signal exchange processes such as modifying plant-derived signals to subvert host recognition could make plant roots “invisible” to soil pathogens.

To improve our understanding of signal exchanges between plant roots and pathogenic fungi, in this study, we investigated early molecular events triggered by host sensing in F. graminearum (Fg) using the model Fg–Brachypodium distachyon (Bd) interaction. The pathogenic fungus Fg causes some of the most economically important diseases of cereal crops and mycotoxin contaminations in food and feed products, resulting in billions of dollars of yield losses worldwide, threatening our food supply and safety (11). Although Fg is better known for its devastating effect on wheat and barley heads as well as maize stalk, recent studies have demonstrated the ability of this pathogen to act as a soil pathogen on wheat, barley, maize, soybean, as well as the model monocot Bd (12–17).

To study host sensing in Fg, we conducted an RNA-sequencing (RNA-seq) analysis to identify candidate fungal genes differentially regulated by host-derived signals prior to physical contact with host roots. Functional analyses through gene knockout analyses showed that precontact Fg genes are involved in fungal development, metabolism, as well as virulence. In addition, we found that sensing of host signals triggers nitric oxide (NO) production in the pathogen. We identified two regulators of NO production in Fg, an ankyrin-repeat containing protein (FgANK1) and a zinc finger transcription factor (TF) (FgZC1). We showed that these two proteins physically interact, and both are required for NO production and pathogen virulence. In the absence of host signals, no interaction between FgANK1 and FgZC1 and no NO production could be observed. Overall, our results reveal mechanistic insights into previously unknown components of host-sensing apparatus in an important plant pathogen.

Results

Sensing of Host Signals Reprograms the Fg Transcriptome.

To gain new insights into fungal processes potentially involved in the recognition of host-specific signals prior to physical contact with roots, we performed transcriptome (RNA-seq) analyses on the interaction between B. distachyon (Bd) and F. graminearum (Fg) at 5 d postinoculation (dpi) (Fig. 1 and Dataset S1). These analyses focused on three stages of Fg growth on minimal medium (MM) designated as 1) Fg-only (no Bd), 2) precontact (Fg grown in the presence of Bd but without any physical contact with the roots), and 3) colonization (Fg infecting Bd roots) (Fig. 1A and SI Appendix, Supplemental Methods). Principal component analysis (PCA) conducted on the transcriptome data revealed distinct differences between fungal transcriptomes from each stage (Fig. 1B). Interestingly, nearly half (320) of the total (678) differentially expressed fungal genes (DEGs) in the comparison between precontact and Fg-only (|log fold change [FC]| ≥ 1, false discovery rate [FDR] < 0.05) were not differentially regulated at the colonization stage compared to Fg-only (Fig. 1C and Dataset S2), suggesting that different fungal processes might be operational at different stages. The precontact DEGs fell into six distinct clusters, which clearly distinguished the stage-dependent transcription patterns (Fig. 1D and SI Appendix, Supplemental Results and Methods). Functional assignment and enrichment analyses for these clustered genes indicated that processes of fungal morphogenesis and development appear to be specifically regulated during the precontact stage in Fg (Fig. 1E, Datasets S3–S5, and SI Appendix, Supplemental Results). In addition, we observed that a substantial portion of precontact DEGs encode secreted proteins (Fig. 1 C and E, SI Appendix, Supplemental Results, and Datasets S3 and S6) that are involved in pathogen virulence (SI Appendix, Supplemental Results and Datasets S2 and S7–S9) and cellular nutrient metabolism (SI Appendix, Supplemental Results and Datasets S7 and S10). Deletion of the two top induced putative effectors and two putative urea transporters suggested that they may be involved in stress tolerance and virulence in Fg (SI Appendix, Figs. S1 and S2). Several precontact DEGs related to glyoxylate cycle and acetyl-CoA metabolic processes were identified in cluster 4 (with induced precontact DEGs which were not affected upon colonization) (Dataset S7), suggesting that energy remobilization processes from fatty acids are operational at this stage. In clusters 4 and 6 (both with induced precontact DEGs), precontact DEGs are enriched for nitrate, ammonium, and urea transport (Datasets S7 and S10). Besides, a large number of significantly down-regulated genes involved in carbohydrate, tryptophan, and GABA metabolic processes were present in clusters 2 and 3 (both with repressed precontact DEGs) (Datasets S7 and S10), indicating that these common nutrients are not preferentially associated with Fg processes at the precontact stage.

Fig. 1.

Transcriptomic reprogramming in F. graminearum (Fg) prior to contact with B. distachyon roots (Bd). Fg was either grown alone on MM (Fg-only) or in the presence of Bd but either with or without (colonization and precontact stages, respectively) physical contact with Bd roots. At 5 dpi, RNA was isolated from fungal mycelia of all three stages of Fg growth and subjected to RNA-seq analysis. (A) The experimental system (Fg-only, precontact and colonization) used in transcriptome analyses. (B) PCA of normalized RNA-seq data. Fragments per kilobase of transcript per million mapped reads (FPKM) values from all expressed genes were logarithmically transformed and used from Fg-only, precontact and colonization samples, each consisting of three biological replicates. x and y axes show principal component 1 (PC1) and principal component 2 (PC2) that explained 62.1% and 12.1% of the overall variance, respectively. It is estimated that a new observation from the same group would fall inside a prediction ellipse with a probability value of 0.95. (C) Expression patterns of 678 DEGs (precontact vs. Fg-only, |log₂FC|>1, FDR < 0.05). A pseudocount of 0.01 was added to the raw FPKM values for log2 transformation. Dendrograms represent distance dissimilarity across-gene expression levels and samples. Sidebars highlight key features associated with common fungal metabolic and virulence pathways. (D) Fuzzy clustering of all genes presented in B based on all six possible differential expression trends across the tested fungal growth stages. Mean values of FPKM were standardized to have a value of 0 and a SD of 1, allowing similar changes in expression. (E) The number of genes from different functional categories (Upper) or involved in putative secreted pathways (Lower) present in all six clusters.

Transcriptome analyses suggested that Fg either perceived or utilized host-derived signals/nutrients prior to physical contact with host roots and radically alters its transcriptome in anticipation of an infection. Presumably, host-derived signals altering the Fg transcriptome at the precontact stage include molecules released by Bd. To evaluate this hypothesis, we treated Fg hyphae with root exudates isolated from Bd plants grown in the presence or absence of the fungus and analyzed the expression of selected Fg precontact DEGs by RT-qPCR. Interestingly, the root exudates from Bd seedlings grown in the presence, but not in the absence of Fg, confirmed the expression of selected pathogen genes which shared similar patterns to the transcriptome (SI Appendix, Fig. S3 and Dataset S2). To partially characterize the nature of Bd root exudates in triggering transcriptional changes in Fg, we used heat-treated precontact exudates to treat Fg and found that heat treatment disabled the activity of these exudates (SI Appendix, Fig. S3), indicating that the observed transcriptional alterations in Fg are not likely caused by common nutritional compounds which would be heat insensitive. In contrast, the signal was insensitive to proteinase K treatment (SI Appendix, Fig. S3). Using untargeted metabolomics, a significant metabolic shift could be observed for plant-derived small molecules in the precontact exudate as compared to the Bd-only extracts (SI Appendix, Figs. S4A and S5 and Supplemental Results and Datasets S12–S14), which is in line with significant transcriptional changes detected here in the precontact Bd roots relative to mock-treated roots used as control (Dataset S15). Taken together, these observations are consistent with the idea that host plant senses the pathogen and secretes small molecules which then serve as host-recognition signals for Fg.

NO Biosynthesis Is Activated during the Precontact Stage in Fg.

In addition to N assimilation, genes associated with NO biosynthesis and response to nitrosative stress were significantly enriched among precontact DEGs (Dataset S7). These genes include a nitrate reductase (NR) (FG05_01947) and a flavohemoglobin NO dioxygenase (FHB) (FG05_00765) potentially involved in NO production and metabolism, respectively (18). In the model fungus Aspergillus nidulans, NR homologs are closely associated with NO homeostasis and coregulated with nitrate assimilatory genes functionally independent of N metabolite repression (18, 19). Moreover, intra- and extracellular NO directly affect the expression of conidiogenesis-related genes and fine-tune hyphae and conidia development in many fungal species (18, 20–22). Consistent with this, the conidiation regulator ortholog BrlA (FG05_01576) from cluster 4 was coregulated with the NO-associated and N transporter genes (Dataset S4).

To determine if NO is produced in Fg during the precontact stage, hyphae from Fg-alone and precontact stages were stained using 4-amino-5-methylamino-2,7-difluorofluorescein diacetate (DAF-FMDA), which detects NO (23, 24). To ensure that the staining by DAF-FMDA was specific for NO, DAF-FMDA with or without the cell-permeable NO scavenger 2-(4-car-boxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (cPTIO) (22) was applied to the edge of the mycelium growing toward the roots or to the corresponding locations in the Fg-alone plate. DAF-FMDA staining revealed predominant fluorescence signals in hyphal tips and branches of Fg during the precontact stage (Fig. 2B). In contrast, no such signal could be detected in fungal-alone or cPTIO-treated samples (Fig. 2 A and C). Furthermore, NO production was also triggered when the precontact Bd root extract was supplemented to the fungus grown in liquid MM (Fig. 2 F and H). As expected, no NO production was evident in Fg grown either in the absence of precontact Bd root exudates or in the presence of plant signals and cPTIO (Fig. 2 D, E, G, and H). Moreover, NO was not induced in Fg by heat-treated root exudates (SI Appendix, Fig. S6). Thus, endogenous NO production in Fg during the precontact stage is triggered by biologically active host-derived signals.

Fig. 2.

NO is produced in Fg in response to signals derived from Bd roots. (A and B) DAF-FMDA was applied to the edge of the Fg mycelium growing either alone (A) or toward Bd roots (B). DAF-FMDA-stained mycelia were rinsed briefly in water and visualized under a fluorescence microscopy for NO production. No NO-specific fluorescent signals could be observed when the fungus was grown alone (A) while fluorescent signals were predominantly detected in hyphal tips prior to physical contact with Bd roots (B). (C) NO scavenger cPTIO inhibited NO accumulation observed at the precontact stage. (D–G) Bd root exudates trigger NO biosynthesis in Fg. Ten microliters of 10⁶/mL of Fg spores were collected and transferred into 100 µL of liquid MM and analyzed by a fluorescence plate reader. NO was detected when Fg was treated with precontact metabolite extracts (F), but not with mock (D and E) or cPTIO treatments (E and G). Representative photos were taken at 24 h posttreatment. (H) Kinetic analysis of NO production by Fg in response to Bd root exudates. Fg grown in liquid MM in the presence or absence of Bd root extracts were treated with DAF-FMDA and/or cPTIO and NO production was assayed over a 48-h time course using a fluorescence plate reader. Data are the mean of three biological replicates with error bars representing SEM. RFU, relative fluorescence unit.

Previous studies in several fungal species have revealed roles for NO in global control of transcriptional and metabolic reprogramming and in modulating fungal morphogenesis and secondary metabolism (25–28). To test if NO is involved in transcriptional regulation in Fg during the precontact stage, the NO inducibility of NR as well as several other precontact DEGs was analyzed in Fg by RT-qPCR after the application of the NO chemical donor S-nitroso-N-acetyl-DL-penicillamine (SNAP). SNAP treatment mimicked the effect of the precontact treatment on the regulation of the selected genes (Tri5, FHB, FG05_00416, FG05_11858, and FG05_00060), which could be compromised by cPTIO (SI Appendix, Fig. S4). Among the genes potentially involved in NO biosynthesis, NR (FG05_01947) is significantly up-regulated during the precontact stage, indicating that NR might be associated with NO production in Fg. The lack of induction by SNAP may suggest that NR is upstream of the NO signal and that no positive feedback exists if NR is the source of NO in Fg. Nevertheless, the regulation of the remaining marker genes by SNAP suggests that NO may contribute to the transcriptional responses observed during the precontact stage in Fg.

An Ankyrin-Repeat Containing Protein Regulates NO Biosynthesis during the Precontact Stage in Fg.

The findings presented above suggest that NO is produced in Fg in response to host signals at the precontact stage although potential regulators acting upstream of NO production are unknown. Therefore, we next focused on the identification of regulatory genes that may be involved in host-sensing-mediated NO production in Fg. We speculate that differential regulation of Fg genes putatively involved in sensing or signal transduction could be a consequence of Bd root sensing by the fungus. Thus, we selected 19 candidates belonging to the “binding,” “signal.” and “unknown” functional categories among the top differentially regulated genes at the precontact stage and generated their independent deletion mutants. Of these, 16 mutants exhibited at least one phenotype of altered mycelium growth, stress tolerance, or virulence, while 15 showed considerably reduced virulence in root infection assays. Seven of these mutants also showed defects during wheat head infections (SI Appendix, Table S1 and Fig. S8), suggesting that shared pathogenicity processes might be used during the infection of different host tissues by Fg. Notably, we found that NO production observed at the precontact stage was completely abolished in the FG05_02877 (named here as FgANK1) mutant (SI Appendix, Fig. S9C), but not the other deletion mutants (SI Appendix, Table S1). Besides, ΔFgANK1 was no longer responsive to Bd root exudates as evidenced by unchanged expression of the fungal marker genes (Tri4, Tri5, FHB, FG05_03544, PR-1, and FG05_00060) (SI Appendix, Fig. S10A). We noticed that the deletion of FgANK1 also resulted in abnormal pigmentation and impaired growth on all tested stress conditions (Fig. 3A) but the growth rate of this mutant on potato dextrose agar (PDA) was not altered (SI Appendix, Fig. S9A). ΔFgANK1 also exhibited an aberrant hyphal morphology with significant apical hyperbranching (Fig. 3B). In addition, no disease symptom caused by ΔFG05_02877 on Bd roots or wheat heads could be observed at 10 dpi (Fig. 3 C and D). Such drastic reduction in pathogen virulence was not simply due to delayed disease development as there was no obvious disease symptom development even 3 wk after direct root inoculations with ΔFG05_02877 (SI Appendix, Fig. S9B). These phenotypes could be fully complemented by expressing GFP-FG05_02877 fusion constructs in ΔFG05_02877 (Fig. 3 A and C–E and SI Appendix, Fig. S10B). Interestingly, the mutant was responsive to external NO as SNAP treatment resulted in similar regulation patterns of the genes Tri5, FHB, FG05_00416, and FG05_00060 in ΔFG05_02877 and the complemented strain (SI Appendix, Fig. S10B). Together, these findings suggested that FgANK1-mediated NO production is responsible for the regulation of downstream genes although similarly to NR, FG05_02877 was not regulated by external NO treatment in the wild-type (WT) pathogen (SI Appendix, Figs. S9D and S10B).

Fig. 3.

FgANK1 (FG05_03877) is required for hyphal development and virulence in Fg. (A) ΔFG05_02877 shows abnormal pigmentation and increased sensitivity to cell wall (0.5 mg/mL calcofluor), oxidative (2 mM H₂O₂), and osmotic (0.7 M NaCl) stresses. These defects could be complemented with the expression of GFP fusion proteins GFP-FgANK1 in the mutant background using either native FgANK1 or the gpdA promoter. (B) ΔFG05_02877 exhibits aberrant hyphal morphology with significant apical hyperbranching relative to WT fungus. (C–E) Deletion of FG05_02877 significantly reduced Fg virulence toward Bd roots (C and E) or wheat heads at 10 dpi (D and E). Virulence-associated defects observed in ΔFG05_02877 could be restored by the expression of GFP fusion constructs Pro_FgANK1:: GFP-FgANK1 or Pro_gpdA::GFP-FgANK1 in the mutant background. (E) Shown here are representative data from one of at least two independent fungal transformants. Percentage Bd root necrosis and the number of diseased wheat spikes are averages from a minimum of 20 replicate roots and at least five wheat heads, respectively, with error bars representing the SD (one-way ANOVA with Dunnett’s multiple comparisons test, ****P < 0.001).

FgANK1 Is Localized in the Cytosol in the Absence of Host Signals but Translocates to the Nucleus in a Host-Sensing-Mediated Manner.

Although the data presented above indicates that FgANK1 is a regulator of NO production triggered by host sensing, how FgANK1 regulates this process is not clear. FgANK1 harbors three N-terminal ankyrin repeats and a C-terminal von Willebrand factor type A (VWA) domain (SI Appendix, Fig. S11B). Ankyrin domain-containing proteins perform various functions such as transcriptional regulation, cytoskeletal organization, cellular development, and differentiation, exclusively through specific protein−protein interactions mediated by the ankyrin-repeat domain (29). Similarly, the VWA conserved domain is found in both extra- and intracellular proteins participating in numerous biological events and interactions with various ligands in a wide range of taxa (30). Indeed, FgANK1 homologs are mainly present in Ascomycota, including Aspergillus species and several root-associated fungi such as F. oxysporum, Nectria hematococca, and Trichoderma virens (SI Appendix, Fig. S11A). No signal peptide or transmembrane domain could be found in FgANK1 (SI Appendix, Fig. S11B). However, we observed GFP signals exclusively localized in the cytosol when the Pro_FgANK1::GFP-FgANK1 construct was introduced into ΔFgANK1growing on MM (Fig. 4A). Interestingly, a proportion of the GFP-FgANK1 protein was translocated to the nucleus upon perception of host signals (Fig. 4B) or in the Pro_gpdA::GFP-FgANK1 strain growing in vitro (Fig. 4C). This subcellular localization of FgANK1 could be further confirmed by analyzing nuclear and cytosolic protein extracts (Fig. 4D). Together, these results suggest that FgANK1 mostly resides in the cytoplasm in the absence of the host plant. In response to host signals, however, FgANK1 translocates into the nucleus and regulates NO production.

Fig. 4.

Subcellular localization of FgANK1 in Fg. (A–C) The FgANK1 ankyrin-repeat domain containing protein accumulates in the cytoplasm during the Fg-only stage and both the cytoplasm and the nucleus of the Fg hyphae during the precontact stage. A FgANK1-GFP fusion protein expressed under the native FgANK1 promoter was exclusively localized in the cytosol when the strain was grown alone on MM (A). Cytosolic GFP signals are partially overlapped with the nuclear DAPI (4,6-diamidino-2-phenylindole) signals at the precontact stage (B), or in the overexpression strain Pro_gpdA::GFP-FgANK1 growing in vitro (C). White arrows indicate nuclear positions. Fungal membranes were stained with FM 4-64 (N-(3-Triethylammoniumpropyl)-4-(6-(4-(Diethylamino) Phenyl) Hexatrienyl) Pyridinium Dibromide). Differential interference contrast (DIC) images and florescence signals and intensities were shown in the Top and Bottom, respectively. (D) Western blot analyses of nuclear and cytosolic protein extracts confirm the cellular localization of FgANK1. Plus (+) and minus (−) represent the growth of fungal strains in the presence or absence of Bd, respectively. Cytosolic and nuclear proteins were detected using the monoclonal anti-GFP, anti-H3, and anti-GAPDH antibodies.

The Interaction Between FgANK1 and a Zinc Finger TF Regulates NO Synthesis and Pathogenesis in Fg.

Since sequence analyses did not reveal any known DNA binding motifs in FgANK1, its nuclear accumulation might involve interaction with nuclear proteins. To identify a possible interaction partner of FgANK1, we screened a yeast-two hybrid (Y2H) library generated using fungal precontact cDNAs. Several rounds of high stringency screening identified a GAL4-type Zn₂-Cys₆ zinc-finger TF (FG05_05068 named here as FgZC1) as an interaction partner of FgANK1.

To functionally characterize FgZC1, we tagged FgANK1 with blue fluorescent protein (TagBFP) and introduced into the ΔFgANK1 background. In the absence of Bd roots, we found similar cytoplasm-localized expression pattern of TagBFP-FgANK1 (SI Appendix, Fig. S12A), which fully complemented the phenotype of ΔFgANK1 (SI Appendix, Fig. S12D), indicating that TagBFP-FgANK1 functions the same as GFP-FgANK1. Further, Y2H assays and coimmunoprecipitation (co-IP) using fungal transformants coexpressing FgZC1-HA and TagBFP-FgANK1 or TagBFP-FgANK1^ΔVWA revealed that the VWA domain of FgANK1 is not required for the interaction between FgANK1 and FgZC1 (Fig. 5 A and B) while the ankyrin domain of FgANK1 is required for the nuclear interaction between FgANK1 and FgZC1. Importantly, the interaction between these two proteins was completely root precontact dependent (Fig. 5B). Interestingly, deletion of FgZC1 did not cause any defect in growth or pathogenicity toward wheat heads (SI Appendix, Fig. S8), but, similarly to the deletion of FgANK1, significantly affected NO production (Fig. 6 A, B, and F) and delayed Bd root infection (SI Appendix, Fig. S13 A and B). During the precontact stage, the selected fungal marker genes were similarly misregulated in FgANK1 and FgZC1 mutants (SI Appendix, Fig. S13C), suggesting that both FgANK1 and FgZC1 are required for their expression at the precontact stage. While the expression of both TagBFP-FgANK1 or TagBFP-FgANK1^ΔVWA driven by the native FgANK1 promoter in the ΔFgANK1 background complemented the defects in NO production (Fig. 6 C, E, and F) and root colonization (SI Appendix, Fig. S12 D–E and H), the latter failed to complement its abnormal hyphal morphology (Fig. 6E). In contrast, the expression of TagBFP-FgANK1^ΔANK fully recovered the growth phenotype but not the precontact NO accumulation and root pathogenicity phenotypes (Fig. 6 D and F and SI Appendix, Fig. S12F). A possible explanation for this is that the disruption of the ankyrin domain region of FgANK1 resulted in a loss of interaction capability, and consequently impaired NO production. There might also be a close correlation between early NO production and fungal virulence as demonstrated by the attenuated virulence phenotypes observed in FgZC1 (SI Appendix, Fig. S13 A and B) and FgANK1 mutants with either full-length or ANK repeat deletions of FgANK1 (Fig. 6 and SI Appendix, Figs. S9 and S12 F and G). The ANK repeats or VWA domain truncation of FgANK1 displayed similar cellular localization (Fig. 6 C–E) which is in line with the observation that one of ANK and VWA domains is enough to support fungal growth (SI Appendix, Fig. S12 E and F). Indeed, both ANK repeats and VWA domains are known to be associated with cytoskeletal organizers important for maintaining cell growth (29, 30). Protein function can also rely on folding or dimers formed between or within domains. The growth complementation in our study by the full length or domain truncations of FgANK1 suggest that the function of FgANK1 in fungal growth is independent of the NO-regulating role provided by the full-length protein which presumably involves protein–protein interactions involving both domains. Consistent with the nuclear localization of FgANK1, a putative nuclear localization sequence (NLS) was found between the second and third ANK repeat (SI Appendix, Fig. S11B).

Fig. 5.

FgANK1 and FgZC1 interact in a host-sensing-mediated manner in Fg. Y2H assays showing that the ankyrin-repeat domain of FgANK1 is required for its interaction with FgZC1 at the precontact stage in Fg. (A) Y2HGold yeast cells coexpressing the bait vectors pGBKT7-FgANK1, pGBKT7-FgANK1^ΔANK, and pGBKT7-FgANK1 ^ΔVWA with the prey vector pGADT7-FgZC1 in pairs were spotted in 10× dilutions on control (−Leu/−Trp) and on high stringency (−Leu/−Trp/−His/−Ade/+X-α-Gal/+aureobasidin) media. The interaction of proteins expressed by the bait and prey constructs activates α-galactosidase in the presence of X-α-Gal and thus generates blue-colored colonies. Cells expressing vector pairs pGADT7-T/pGBKT7-53 and pGADT7-T/pGBKT7-Lam were used as positive and negative controls. Self-interaction was tested by cotransforming pGBKT7-FgANK1 or pGADT7-FgZC1 with the corresponding control vectors. (B) Total proteins (input) extracted from Fg strains coexpressing FgZC1-HA with TagBFP-FgANK1 or TagBFP-FgANK1^ΔVWA were immunoprecipitated with the anti-HA antibody (IP). Western blots were developed with the anti-tRFP antibody to detect the interaction of FgZC1 with the full-length or VWA domain truncated FgANK1 fused with TagBFP. The FgANK1–FgZC1 interaction only occurs in the presence of Bd roots (precontact). Anti-GAPDH and anti-HA antibodies were used as loading controls for the input and IP, respectively.

Fig. 6.

FgZC1 and FgANK1 are indispensable for NO production triggered in Fg by Bd host signals at the precontact stage. (A–C) No NO signals could be detected by fluorescein isothiocyanate (FITC) staining in FgANK1 (A), and FgZC1 mutants (B), respectively, while a TagBFP-FgANK1 construct expressed in the ∆FgANK1 mutant background restores NO production (C) when Fg hyphae from the precontact stage were stained with DAF-FMDA. (D–F) The ankyrin-repeat domain but not the VWA domain of FgANK1 is necessary for NO production at the precontact stage as the TagBFP-FgANK1 fusion protein with a deleted VWA region (TagBFP-FgANK1^ΔVWA) restores NO production when expressed in ∆FgANK1 (D) while NO signals were completely missing in the TagBFP-FgANK1^ΔANK with a deleted ankyrin (ANK)-repeat domain expressed in the FgANK1 mutant background (E). Percentage of Fg hyphae accumulating NO at the apical branches of the indicated genotypes was quantified by observing 10 independent microscopic slides holding stained mycelium from five inoculation plates (F). Data represent averages with SE of mean (one-way ANOVA with Dunnett’s multiple comparisons test, ****P < 0.001).

Overall, these results support a model where upon perception of host signals FgANK1 and FgZC1 physically interact to regulate NO signaling and control the transcriptional responses to host recognition. Indeed, the similar regulation of NR and FgANK1 under NO treatment (SI Appendix, Figs. S7 and S10B) indicates that there might be a direct link between FgANK1 and NR. Therefore, we investigated potential binding capacity of the FgANK1-FgZC1 complex to the upstream regions of NR. Electrophoretic mobility shift assays (EMSA) showed that FgANK1 and FgZC1, only when they are both present, could specifically recognize a biotinylated DNA probe derived between −998 and −548 bp upstream of the NR gene (SI Appendix, Fig. S14 A and B). Genomic fragments from this region could be immunoprecipitated with FgANK1-GFP, with significant enrichments in the ProF_gANK1::GFP-FgANK1 strains during precontact with Bd roots (SI Appendix, Fig. S14C). In the FgANK1 overexpression strain used as control, the immunoprecipitated fragments were also significantly enriched regardless of treatment, indicating a constitutive activation of DNA binding (SI Appendix, Fig. S14C). Therefore, the FgANK1–FgZC1 complex appears to facilitate transcriptional regulation by directly binding to the NR promoter. This finding also suggests a role for NR in NO production in Fg responding to host signals. However, precontact metabolite-triggered NO production in Fg could not be fully abolished using an NR inhibitor (sodium tungstate) (SI Appendix, Fig. S14D), indicating only a partial contribution of NR to the host-mediated NO production in Fg.

Discussion

During their coevolution with plants, certain soil microbes have evolved to exploit signals released by plant roots for their benefits. Some of the best studied chemical signals associated with belowground microbe–root communications are from mutualistic interactions (31). Host sensing by soil pathogens can also be critical for root infection, but the mechanisms that soil pathogens use to sense and trace host signals remain largely unexplored. Presumably, sensing of host plant in a complex environment like the rhizosphere takes places relatively early on and even in the absence of a physical interaction between the pathogen and plant roots.

In this study, we used the Fg–Bd interaction to examine host-sensing-mediated molecular processes in the pathogenic fungus Fg. To the best of our knowledge, early responses triggered by host signals in Fg, a fungal pathogen that can infect multiple host tissues including roots, have not been previously examined. Here, we performed detailed transcriptome analyses of different fungal growth stages and identified large number of genes differentially regulated in the presence of root signals. Our discovery of the regulation of endogenous NO during the host-sensing stage sheds light on the interplay between signal recognition and developmental transitions in Fg, processes that are essential for both environmental adaptation and pathogenesis of this soil-borne fungal pathogen. NO-based regulatory protein modifications may mediate cellular responses involving a complex regulatory network underlying NO signaling. NO can also serve as a key messenger and a signaling molecule during pathogenic as well as mutualistic interactions (20, 32–37). Consequently, NO would deliver key messages from environmental signals and mediate the pathogen cellular responses through transcriptional reprogramming to facilitate lifestyle transition and niche adaptation (25–28).

Functional analyses of selected candidate precontact DEGs have led to the discovery of FgANK1, an ankyrin-repeat domain-containing protein that regulates host-sensing-mediated NO production and virulence in Fg. Our phylogenetic analyses showed that FgANK1 homologs are present in other fungal species that include other soil-borne pathogens such as F. oxysporum (SI Appendix, Fig. S11 A and B). The FgANK1 mutants were developmentally affected and incapable of infecting Bd roots and wheat heads. These observations suggest that FgANK1 is required for NO production as well as fungal development and virulence. It is possible that virulence and developmental alterations observed in the FgANK1 mutant could be due to defects in NO biosynthesis as NO has been implicated in these fungal processes in other fungi (18, 21, 22, 28, 35). It is well documented that NO is an important signaling molecule that contributes to pathogen survival within the host (38), and thus the loss-of-virulence phenotype of ΔFgANK1 could be due to the extreme sensitivity of this mutant to plant defenses or the requirement of fungal NO to interfere with plant defenses.

We showed that FgANK1 is a cytosolic protein and likely acts downstream of a membrane receptor involved in direct sensing of host metabolites (Fig. 7). Sensing of host signals by specific receptors may activate a phosphorylation cascade that in turn can mediate the translocation of FgANK1 into the nucleus although we cannot rule out the possibility that a host metabolite taken up by the fungus directly binds to the VWA domain known to function as a ligand binding domain. In addition, we identified the Zn2-Cys6 zinc finger TF FgZC1 required for NO production and fungal virulence as an interaction partner of FgANK1 in the nucleus. We show that the ankyrin-repeat domain of FgANK1 is required for the interaction with FgZC1. Interestingly, the interaction between these proteins occurs only in the presence of host signals. FgZC1 homologs can also be found in a large range of fungal species, but clearly fall into two distinct phylogenetic clusters that comprise fungal pathogens and symbionts (SI Appendix, Fig. S11C).

Fig. 7.

A model explaining the proposed roles of FgANK1 and FgZC1 in host-sensing-mediated NO biosynthesis in Fg. Host sensing triggers transcriptional reprogramming and NO production in Fg. The ankyrin-repeat domain containing protein FgANK1 resides in the cytoplasm in the absence of the host plant. Upon sensing of host signals, such as the metabolites found at root exudates, FgANK1 translocates to the nucleus and physically interacts with the zinc finger transcription factor FgZC1. The FgANK1–FgZC1 complex directly binds to the NR promoter to regulate root-triggered NO production. NO, in turn, may regulate the expression of genes involved in fungal virulence and development either directly or indirectly by further modulating the transcriptome. In addition to the regulation of NO production, FgANK1 is essential for normal fungal growth, hyphal morphology, stress resistance, and pathogenicity, whereas FgZC1 has a relatively minor role in fungal virulence. Broken arrows refer to tenuous links.

Overall, our results suggest that the FgANK1–FgZC1 protein complex acts as a master regulatory complex contributing to NO production, thus could be central for the downstream transcriptional programs (Fig. 7). Although we show that NO is produced in Fg in a host-recognition-dependent manner, and this is regulated by FgANK1 and FgZC1, the exact biosynthetic pathway by which NO is produced in Fg and other fungi is not clear. In A. nidulans, NR seems to catalyze NO production in a nitrogen metabolite repression-insensitive manner (18). In contrast, no evidence was found for an enzymatic route for NO production in Magnaporthe oryzae during hyphal growth and formation of the appressorium, implying the existence of an alternative NO synthesis pathways in this fungus (22). In this study, we obtained evidence that FgANK1 and FgZC1 facilitate a transcriptional regulation by directly binding to the NR promoter. NO production was abolished while NR was lowly expressed in FgANK1 and FgZC1 mutants, suggesting that these proteins directly affect the expression of NR that seems to at least partially contribute to host-triggered NO production. Further testing the involvement of NR in NO production in Fg requires more detailed future investigations as multiple and/or redundant pathways seem to be involved in NO production in different fungi (35, 39).

In addition to regulating NO production, FgANK1 has important roles in fungal virulence. In contrast, FgZC1, the interaction partner of FgANK1, plays a more subtle role in fungal virulence (i.e., it affects pathogen virulence on roots but has no role in wheat head infection) despite its involvement in NO production. The observation that both FgANK1 and FgZC1 mutants are deficient in NO production and pathogen virulence, suggests a causative link between these two fungal processes. Indeed, virulence was compromised in the rice blast fungus M. oryzae when rice leaves were treated with a NO inhibitor at the time of inoculation (22). However, possible effects of NO inhibitors on plant defense are unknown and in the absence of a NO-biosynthesis mutant, the link between NO and fungal virulence remains somewhat tenuous.

In addition to the enabling of the discovery of NO being an essential fungal signal following host sensing, our transcriptome analyses showing distinct gene expression profiles from host recognition to fungal colonization revealed insights into the processes involved in root infection by Fg. Our results indicate that in response to root exudate-derived signals, the fungus launches an initial “preparation-for-infection” stage presumably via NO signaling. During this stage, several virulence factors were down-regulated while fungal defense and cellular reorganization processes were induced. Suppression of genes involved in deoxynivalenol (DON) biosynthesis is particularly interesting, given that DON is a virulence factor during wheat head infection in Fg (40) and root infection in the related species Fusarium pseudograminearum (41). Tri5 expression was also suppressed by SNAP (SI Appendix, Figs. S7 and S10B), suggesting that host-sensing-mediated NO production is responsible for the suppression of toxin biosynthesis genes. It can be speculated that Fg might benefit from transcriptional suppression of these virulence factors during the precontact stage to avoid host recognition. Precontact responsive genes we identified may be closely associated with sensing of environmental signals and host recognition and preparation for infection. Indeed, independent knockout mutants we generated for five cell-surface or membrane-associated genes (FG05_02447, FG05_02961, FG05_11577, FG05_11507 , and FG05_06479) likely associated with sensing environmental cues showed reduced virulence in Bd roots (SI Appendix, Fig. S8 and Table S2).

Simultaneous up-regulation of Fg genes encoding transporters of both preferential and nonpreferential nitrogen sources at the precontact stage is particularly intriguing (Datasets S2, S7, and S10). In Fg and other fungal species, N transporters and permeases, along with many metabolic enzymes, are tightly regulated by orthologs of the N catabolite regulators AreA/Nit2 (42–46). Accordingly, strong expression of nonpreferential N assimilation genes such as nitrate and urea transporters is observed under N limitation (42, 45). As mentioned above (Dataset S4), we did not observe a changed expression of the Fg AreA homolog in our precontact DE gene set. We note that TFs such as AreA are normally up-regulated during N starvation. The precontact media we used contains ammonium nitrate and should be rich in nutrients (e.g., N compounds) because of the metabolites/exudates produced by the roots. Therefore, the function of AreAs may not be required under the experimental system used here.

In conclusion, the results presented here are consistent with a model (Fig. 7) where sensing of host-derived signals prior to host contact alters diverse processes and prepares the fungus for colonization. NO production, an early response to host sensing, requires FgANK1 in collaboration with FgZC1 in Fg. NO, in turn, can act as a versatile signal potentially involved in fungal development and virulence-associated processes by transcriptionally regulating downstream responses. Future research is required to determine if similar processes are operational in other soil-borne pathogenic fungi but the presence of FgANK1 and FgZC1 homologs in a number of pathogenic fungi and especially in Fusarium spp. suggests that this process might be conserved. The insights revealed by this work will be useful for the development of new disease control strategies that can directly or indirectly target NO production in pathogenic fungi.

Materials and Methods

The Escherichia coli strain Top10 (Life Technologies) and Saccharomyces cerevisiae strain BY4743 were used for cloning purposes. The S. cerevisiae strains Y187 and Y2HGold (Clontech) were used for yeast two-hybrid assays. The Fg CS3005 isolate was used in growth and infection assays and for generating gene deletion mutants. Fg strains were either routinely maintained on PDA (BD Difco) or grown on MM (47) at pH 7 adjusted with 2-(N-morpholino)ethanesulfonic acid (MES). For fungal stress test, growth was assayed on PDA plates supplemented with 0.7 M NaCl, 2.2 mM H₂O₂, and 0.5 mg/mL calcofluor white. Bd (Bd21-3) seeds were surface sterilized and pregerminated for 5 d prior to Fg-roots interaction assays conducted on MM. Details of materials and methods used are available in SI Appendix.

Data Availability.

RNA-seq data have been deposited in the NCBI BioProject database with accession codes PRJNA564465 and PRJNA614594. Untargeted metabolomic data associated with this study have been deposited in the Mass Spectrometry Interactive Virtual Environment (MassIVE) repository with an accession code MSV000085134.