Nematode-trapping fungus Arthrobotrys oligospora is hungry for iron-chelating agent COQ7 of nematodes

1 Introduction

Iron, the most abundant trace metal in living organisms, is essential for oxygen transport, energy metabolism, and DNA synthesis [1–4]. However, iron overload can result from a range of genetic and environmental factors, often leading to parenchymal organ damage [5, 6]. Congenital defects in iron-related genes disrupt absorption, storage, or transport, causing systemic pathologies, such as Caucasians are particularly susceptible to developing iron overload and the associated complications of hemochromatosis due to a higher prevalence of mutations in the homeostatic iron-regulator and iron-storage genes within this population [7, 8]. Among them, HFE mutations associated with susceptibility to iron overload are most common in white populations, occurring in 0.3–0.5 percent of white persons of northern European descent [9, 10]. Chelation therapy is the primary treatment for mitigating iron accumulation, particularly from repeated blood transfusions, by promoting iron excretion via urine and/or feces [11, 12]. To date, few studies have investigated the mechanisms of iron overload and chelation in non-human eukaryotes.

A heterogeneous group of fungi within the phylum Ascomycota can differentiate their mycelia into specialized trapping structures under nutrient-deficient conditions in the presence of nematodes, capturing and penetrating them as a primary nutritional source [13]. These fungi, collectively known as nematode-trapping fungi (NTFs), form a monophyletic clade within the class Orbiliomycetes and produce diverse trapping devices, including adhesive networks, adhesive knobs, and constricting rings [14]. The emergence of these trapping structures is widely considered a hallmark of the evolutionary transition from a saprophytic to a predatory lifestyle in NTFs. Nematodes, the most abundant animals on Earth, occupy a wide range of ecological niches and trophic levels within soil food webs [15]. Among them, plant-parasitic nematodes represent highly destructive agricultural pests, infecting more than 4, 000 plant species and causing annual crop losses exceeding ＄100 billion globally [16]. NTFs are regarded as effective natural antagonists that contribute to biological control of nematode populations in various ecosystems [17]. Paradoxically, despite the widespread availability of nematode prey, NTFs are neither abundant nor widely distributed in many environments [13]. It has been hypothesized that the evolution of these fungi involved a progressive degeneration of their carnivorous capabilities [18]. However, the molecular and physiological mechanisms underlying the limited predatory efficiency of NTF trapping structures remain largely unknown.

Recent studies have demonstrated that exogenous iron can induce the predominant NTF Arthrobotrys oligospora to form trapping devices even in the absence of nematode prey [19, 20]. Chemical and biochemical analyses have revealed that these trapping devices contain significantly more iron than mycelia, as indicated by the presence of iron-rich, electron-dense bodies localized within the trapping devices [19, 20]. Genomic, evolutionary, and functional analyses further suggest that during the Late Paleozoic Ice Age, NTFs lost the CCC1-mediated vacuolar iron detoxification system—a mechanism conserved in most other fungi [20]. Moreover, both the acquisition of the siderophore desferriferrichrome and the emergence of trapping structures appear to be closely linked to elevated ambient temperatures, given that intracellular iron content and composition in fungi are inversely correlated with temperature [20]. Collectively, these results suggest that NTFs are prone to iron overload at relatively high temperatures due to loss of the canonical vacuolar iron detoxification pathway, similar to iron overload observed in hemochromatosis caused by mutations in iron-regulatory and storage genes [20]. However, it remains unclear how the iron-rich trapping devices contribute to carnivorous capacity of NTFs.

2 Results

2.1 Only NTFs in Ascomycota lack the gene coq7

We hypothesize that NTFs have lost genes encoding iron chelators that are commonly retained in other fungi and nematodes, thereby driving NTFs to prey on nematodes primarily as a source of iron-chelating agents. To investigate this, we first examined 1361 fungal genomes from the phylum Ascomycota, representing 66.16% of the total fungal genomes analyzed. Notably, genomic analysis revealed that among Ascomycota, only nine fungal species, all belonging to NTFs and accounting for 0.66% of ascomycetes, lack the coq7 gene, which is essential for UQ biosynthesis [21] (Table S1). Importantly, COQ7 is a membrane-bound di-iron enzyme belonging to a unique subgroup of the ferritin superfamily and is implicated in iron binding and chelation [22–24]. We subsequently expanded the analysis to 2057 fungal genomes and conducted an extensive comparative survey, which indicated that only 25 fungal species lack the coq7 gene, including 9 NTFs from Ascomycota, 12 species from Basidiomycota, 2 from Neocallimastigomycota, and 2 from Zoopagomycota (Table S1). UQs are not only essential electron carriers in aerobic respiration, serving as redox intermediates in both aerobic bacteria and eukaryotes [25], but also contribute to regulating redox homeostasis in anaerobic microorganisms [26]. Prior studies have shown that most fungi utilize UQs with prenyl side chains of 9 or 10 isoprenoid units (UQ₉ and UQ₁₀) [27, 28]. The biosynthesis of UQ₉ and UQ₁₀ requires at least thirteen proteins, including COQ1–COQ11, YAH1, and ARH1 [28]. Among these, COQ7 is a multifunctional protein involved not only in UQ biosynthesis but also in iron chelation, functioning as a ferritin homolog [22–24].

Analysis of UQ biosynthetic genes across 2057 fungal genomes revealed that, with the exception of coq9 (56.73%, 1167/2057) and coq10 (38.89%, 800/2057), all remaining 11 genes—including coq1 (99.85%, 2054/2057), coq2 (99.71%, 2051/2057), coq3 (99.71%, 2051/2057), coq4 (99.76%, 2052/2057), coq5 (99.81%, 2053/2057), coq6 (99.81%, 2053/2057), coq7 (98.78%, 2032/2057), coq8 (99.85%, 2054/2057), coq11 (98.06%, 2017/2057), arh1 (99.90%, 2055/2057), and yah1 (99.56%, 2048/2057)—are highly conserved and present in more than 98% of fungal genomes (Fig. 1A and Table S1). These findings suggest that coq7 is one of the core genes required for UQ biosynthesis and is likely critical for maintaining metabolic function and viability in most fungi [25].

Open image in new window

2.2 UQ₁₀ inhibits trapping devices and nematicial activity of A. oligospora

We hypothesized that the absence of coq7 may disrupt UQ biosynthesis and affect iron sequestration, thereby increasing the susceptibility of NTFs to iron overload and promoting the formation of iron-rich trapping devices. Based on this, it is reasonable to infer that exogenous supplementation with UQ₁₀ could reduce trapping device formation in NTFs. To test this hypothesis, we supplemented A. oligospora with UQ₁₀ at concentrations of 10 and 50 μM under normoxic conditions. These concentrations have been reported to exert no adverse effects on human cells [29], and a solvent-only treatment was used as the control. UQ₁₀ at 50 μM significantly inhibited the formation of trapping devices in A. oligospora, whereas 10 μM had no observable effect (Fig. 2A, B). Furthermore, UQ₁₀ at 50 μM markedly impaired the nematode-capturing capacity of A. oligospora, consistent with the reduction in trapping device formation (Fig. 2A–C).

Open image in new window

2.3 Exogenous coq7 alleviates fungal trapping devices and nematicidal activity

To evaluate the role of coq7 in fungal trapping device formation and nematicidal activity in A. oligospora, the coq7 gene (819 bp) from Saccharomyces cerevisiae BJ5464 was cloned and introduced into the wild-type (WT) strain of A. oligospora to generate the E-coq7 mutant (Fig. S1A–B). Transcriptional analysis confirmed successful expression of coq7 in the E-coq7 mutant (Fig. S1C). High-performance liquid chromatography (HPLC) analysis revealed that the E-coq7 mutant exhibited a metabolic profile identical to that of the A. oligospora WT strain (Fig. S2), suggesting that exogenous coq7 expression did not affect the metabolic pathways or UQ biosynthesis in A. oligospora. Remarkably, phenotypic analysis showed that the E-coq7 mutant formed significantly fewer trapping devices than the WT strain (Fig. 2D, E). Furthermore, the E-coq7 mutant displayed significantly reduced nematicidal activity compared to the WT strain (Fig. 2D–F). Together, these results suggest that exogenous COQ7 may act in a ferritin-like manner to alleviate iron accumulation in trapping devices, thereby diminishing the nematicidal activity of A. oligospora.

2.4 Disruption of nematode coq7 did not affect formation of fungal trapping devices

Given that the trapping devices of NTFs are rich in iron and that COQ7 possesses ferritin-like properties, we hypothesized that the predator–prey interaction between NTFs and nematodes may largely result from the natural affinity between ferritin and iron. To investigate whether fungal trapping device efficacy is related to COQ7 in nematodes, the Δclk-1 mutant and WT (N2) strain of Caenorhabditis elegans were employed to assess trapping device formation and nematicidal activity of A. oligospora. The clk-1 gene encodes a 187-residue polypeptide homologous to yeast coq7, which is essential for UQ biosynthesis in yeast [30]. Notably, the Δclk-1 mutant nematodes exhibited a similar effect on fungal trapping device formation compared to the N2 strain and showed a slightly increased effect when A. oligospora was treated with 0 and 10 μM UQ₁₀ (Fig. 2A, B), suggesting that the absence of COQ7 in the Δclk-1 nematode mutant does not significantly affect trapping device formation in A. oligospora.

2.5 Disruption of nematode coq7 reduced nematicidal activity of trapping devices

Many studies have demonstrated that nematodes can induce the formation of fungal trapping devices in NTFs [31, 32]. However, whether nematode components influence the nematode-capturing ability of NTFs remains unreported. Due to the genetic defect, Δclk-1 mutant nematodes exhibit a general reduction in behavioral rates compared to the N2 nematodes [33], suggesting that Δclk-1 mutants are less capable of escaping capture by fungal trapping devices than N2 nematodes. Notably, we observed that fungal trapping devices captured nearly three times more N2 nematodes than Δclk-1 mutants (Fig. 2C). Most Δclk-1 nematodes were able to evade the traps and maintained a vibrant, curved morphology, in contrast to the stiff, immobilized N2 nematodes (Fig. 2A). The higher capture rate of N2 nematodes indicates a specific attractive interaction between fungal trapping devices and nematode COQ7. It is also possible that the lack of COQ7 in nematodes causes them to also have an iron overload similar to the NTF, resulting in repulsion between the fungal trapping devices and the Δclk-1 mutant nematodes.

To further investigate the association between nematode COQ7 and the nematicidal activity of fungal trapping devices, we examined three additional nematode mutants deficient in genes encoding Acyl-CoA oxidase (ACOX), paired-like homeobox transcription factor/defecation suppressor (DSC), and bridge-like lipid transfer protein (LPD), respectively. Notably, the acox and lpd genes are also absent from all NTF genomes (Table S2). Both the Δacox and Δlpd nematode mutants exhibited locomotor sluggishness comparable to that of the Δclk-1 mutant [34, 35]. Phenotypic analysis revealed that the Δclk-1 and Δlpd mutants had the strongest effects on inducing trapping device formation in A. oligospora, whereas the Δdsc mutant had the weakest effect (Fig. 2G, H). Interestingly, the number of Δdsc nematodes captured by the fungal trapping devices was the second highest, surpassed only by the Δacox mutant (Fig. 2I). Furthermore, the number of Δacox nematodes captured was approximately three times greater than that of the Δclk-1 mutant (Fig. 2I). These results suggest that among these nematode factors, COQ7 plays a uniquely potent role in enhancing the nematicidal activity of fungal trapping devices.

2.6 Arthrobotrys oligospora uses arthrobotrisins instead of UQ to respond to elevated oxygen levels

Because UQs are often associated with oxygen-mediated fungal metabolism, we wondered how A. oligospora that loses the gene coq7 would be different. Metabolic analysis revealed that the A. oligospora WT strain exhibited markedly different metabolic profiles under shaking versus static culture conditions (Fig. 3A). The primary distinction between these conditions is oxygen transfer and mixing: shaking enhances oxygen availability by increasing the surface area between the liquid culture and the air, leading to improved aeration for more efficient microbial growth, particularly in aerobic organisms [36, 37]. Notably, UQs were not detected in the WT strain under shaking conditions, whereas UQ₈ was present under static conditions (Fig. 3B). The absence of UQ₉ and UQ₁₀ in the WT metabolic profiles aligns with the absence of the coq7 gene in A. oligospora. Furthermore, transcriptional analysis demonstrated that coq5, coq6, and yah1 were significantly downregulated under shaking conditions compared to static conditions (Fig. 3C), consistent with the detection of UQ₈ exclusively under static conditions. Detailed metabolic profiling also revealed that arthrobotrisins, a class of chemotaxonomic meroterpenoids, accumulated significantly in the WT strain under shaking conditions but were nearly absent under static conditions (Fig. 3D). Correspondingly, transcriptional analysis showed that biosynthetic genes within the art gene cluster responsible for arthrobotrisin synthesis, including Ao283–Ao276 and Ao274–Ao273, were significantly upregulated under shaking conditions, whereas adjacent non-biosynthetic genes (Ao275 and Ao284–Ao286) exhibited no significant change (Fig. 3E).

Open image in new window

To investigate the role of arthrobotrisins in the response of A. oligospora to elevated oxygen levels, we constructed a mutant strain Δart, deficient in arthrobotrisin biosynthesis, based on a previous study [38]. We compared the metabolic profiles of the WT and Δart mutant strains under both shaking and static conditions. Principal component analysis (PCA) revealed that under static conditions, the WT and Δart strains clustered closely, whereas they did not cluster together under shaking conditions (Fig. S3). The Δart mutant exhibited nearly identical metabolic profiles under both conditions, contrasting with the pronounced differences observed in the WT strain (Fig. 3A). These results suggest that the absence of arthrobotrisin biosynthesis impairs the fungus's metabolic response to shaking conditions. To further examine the association between arthrobotrisins and oxygen levels, we cultured WT and Δart strains at four shaking speeds: 0, 60, 120, and 180 rpm, as dissolved oxygen levels positively correlate with shaking rates [39]. Metabolic analysis showed that major arthrobotrisin contents were highest in the WT strain at 120 and 180 rpm (Fig. 3F), with contents increasing progressively between 0 and 120 rpm. Conversely, UQ₈ content in the WT strain decreased as shaking speeds increased from 0 to 180 rpm (Fig. 3G). Moreover, all three functional redox states of arthrobotrisins, corresponding to the fully oxidized, partially reduced, and fully reduced forms of UQ₈, had also been isolated in A. oligospora in previous studies (Fig. S4) [40, 41]. These results indicated that the fungus A. oligospora used arthrobotrisin biosynthesis in adaption to the shaking conditions whereas utilized UQ₈ for the static conditions.

In contrast, the Δart mutant exhibited increasing UQ₈ levels over the same shaking speed ranges (Fig. 3G), indicating that the absence of arthrobotrisins may induce UQ₈ production in response to elevated oxygen. Transcriptional analysis confirmed significant upregulation of UQ biosynthetic genes in the Δart mutant relative to the WT strain (Fig. 3H). These results indicated that UQ₈ can functionally substitute for arthrobotrisins in fungal response of the Δart mutant to increased oxygen levels. Collectively, because A. oligospora preferentially used arthrobotrisins rather than UQs to cope with oxygen fluctuation, this most likely caused the fungus to lose the gene coq7.

2.7 Exogenous COQ7 exhibits adsorption effects on fungal trapping devices both in vitro and in vivo

Two experiments were conducted to evaluate the affinity of exogenous COQ7 for fungal trapping devices. In the first experiment, fluorescently labeled COQ7 and the control protein ACOX were heterologously expressed in Escherichia coli BL21. The purified proteins were then incubated with fungal trapping devices of A. oligospora WT and its mutant Δart. The results showed that COQ7 adhered to fungal trapping devices of both strains, whereas ACOX did not (Fig. 4A and Fig. S5). In the second experiment, COQ7 was fluorescently labeled within nematodes. Notably, nematode COQ7-derived fluorescence was predominantly localized on the nematode surface (Fig. 4B). Transient losses of fluorescence were observed in specific regions of the nematode body surface (Fig. S6), indicating that nematode COQ7 could be shed from the nematode. Recent studies have shown that mitochondria could be exported into surrounding environments since the observed migrasomes are filled with mitochondria directly in peripheral tissue [42–44]. It was most likely that COQ7 (CLK-1) localized in mitochondrion could be also exported from nematode into its surrounding environment. Remarkably, at sites of contact with fungal trapping devices, nematode-associated fluorescence was detected on the surface of fungal trapping devices (Fig. 4B). Collectively, these results provide evidence for nematode COQ7 secretion and surface exposure and demonstrate its affinity for fungal trapping devices.

Open image in new window

2.8 Oxygen enhances fungal trapping device formation via increasing iron levels

Numerous studies suggest that oxygen functions as a developmental morphogen in metazoan embryos, driving genomic and metabolic reprogramming during periods of oxygen fluctuation [45, 46]. However, the effects of oxygen fluctuations on fungal morphogenesis, as well as on their metabolic and genomic evolution, remain largely unexplored. Direct evidence linking hyperoxia to the induction of fungal trapping device formation remains lacking, and the roles of specific metabolites in response to hyperoxia are still poorly understood. To investigate the effect of oxygen levels on fungal phenotypes, including growth, trapping device formation, and nematicidal activity, two oxygen conditions, 21% atmospheric oxygen (normal) and approximately 10% atmospheric oxygen (low), were established for agar plate bioassays. Under low oxygen conditions, the A. oligospora WT strain exhibited smaller aerial mycelial colonies and significantly larger hollow areas compared to those grown under normal oxygen levels (Fig. 5A, B), indicating that reduced oxygen inhibited aerial mycelial growth. However, following treatment with C. elegans N2 nematodes, the fungal WT strain grown under normal oxygen produced trapping devices, whereas those grown under low oxygen did not (Fig. 5A, C). Additionally, the WT strain under normal oxygen conditions demonstrated strong nematode-trapping activity, while strains grown under low oxygen showed none (Fig. 5D). Previous studies have suggested that mycelial senescence or nutrient deficiency can readily induce trapping device formation [19, 38]. To exclude the possibility that differences in trapping devices resulted from unequal aerial mycelial growth under the two oxygen conditions, we selected cultures with comparable mycelial growth for further trapping device analysis. Specifically, aerial mycelia grown under normal oxygen were one day younger than those grown under low oxygen. Notably, younger aerial mycelia under normal oxygen developed trapping devices, whereas those under low oxygen did not. Collectively, these results indicate that elevated oxygen levels promote trapping device formation in A. oligospora WT.

Open image in new window

Since oxygen can oxidize ferrous iron (Fe²⁺) to ferric iron (Fe³⁺), resulting in electron loss and Fe³⁺ accumulation. A recent study demonstrated that A. oligospora mycelia develop trapping devices to store excess iron, as these structures contain significantly higher iron concentrations than the surrounding mycelia [20]. To evaluate the role of oxygen in regulating iron levels within fungal mycelia, we measured total free iron content under two oxygen conditions. As anticipated, mycelia grown under normal oxygen exhibited significantly higher levels of total free iron—including both Fe²⁺ and Fe³⁺—compared to those grown under low oxygen (Fig. 5E). The levels of Fe²⁺, Fe³⁺, and total free iron under normal oxygen were more than double those under low oxygen, indicating that elevated oxygen substantially increases free iron ion accumulation (Fig. 5E). Meanwhile, the Fe²⁺/Fe³⁺ ratios in mycelia remained similar under both conditions (Fig. 5F), consistent with the biological requirement to maintain redox homeostasis of the Fe²⁺/Fe³⁺ ratio. These results suggest that oxygen is a key factor inducing trapping device formation by elevating free iron content in fungal mycelia.

2.9 Arthrobotrisin deficiency causes spontaneous formation of trapping devices

All arthrobotrisin metabolites in the NTF A. oligospora are derived from a key hybrid precursor, farnesyl toluquinol, which shares a distinctive feature with ubiquinol-3 (UQ₃): the farnesyl chain is attached to a benzoquinol moiety [47]. Notably, a series of oxidation reactions are critical for converting farnesyl toluquinol into arthrobotrisins, as each arthrobotrisin incorporates at least four oxygen atoms [41, 47–50]. This oxidative incorporation of molecular oxygen also occurs during ubiquinone biosynthesis, where monooxygenase-catalyzed hydroxylation reactions introduce three oxygen atoms into the quinone structure [51]. From the perspective of the structural oxidation degree, arthrobotrisin biosynthesis requires more oxygen than ubiquinone biosynthesis. This is consistent with previous research, which indicates that arthrobotrisin production positively correlates with oxygen levels and deficiency in arthrobotrisin biosynthesis results in a significant increase in trapping device formation and nematocidal activity in the mutant Δart [41, 47–50]. Deletion of the genes involved in biosynthesis of highly oxidated arthrobotrisins, specifically the dehydrogenase gene Ao274 and the cytochrome P450 genes Ao278, Ao280, and Ao282 within the art gene cluster, markedly enhances nematode-induced trapping device formation in the mutant strains [41, 47, 49, 50]. However, the underlying mechanism by which arthrobotrisin biosynthesis suppresses nematode-induced trapping device formation in A. oligospora remains unclear.

The Δart mutant and WT strains were cultivated on plates under normal (21% atmospheric oxygen) and low oxygen conditions (approximately 10% atmospheric oxygen). Notably, the Δart mutant exhibited significantly larger hollow areas at the colony center compared to the WT strain under normal oxygen conditions, whereas both strains showed similar colony growth under low oxygen levels (Fig. 5G, H). This suggests that the arthrobotrisins improve aerial mycelial formation under normal oxygen conditions, consistent with the role of arthrobotrisins in fungal adaptation to elevated oxygen levels. Interestingly, the Δart mutant formed trapping devices without nematode treatment under both oxygen conditions (Fig. 5I, J), while the WT strain did not, indicating that endogenous arthrobotrisin biosynthesis inhibits spontaneous formation of fungal trapping devices. Both Δart and WT strains treated with nematodes produced significantly more trapping devices under normal oxygen conditions compared to low oxygen conditions (Fig. 5I, J), consistent with the above result that elevated oxygen promotes fungal trapping device formation. Importantly, the increase in trapping device numbers attributable to elevated oxygen was substantially greater in the Δart mutant than in the WT strain (Fig. 5I, J). Furthermore, the Δart mutant exhibited a more pronounced increase in both trapping device formation and nematicidal activity under both oxygen conditions compared to the WT strain (Fig. 5J). Collectively, these results indicate that arthrobotrisin biosynthesis functions to inhibit oxygen-induced trapping device formation in A. oligospora.

2.10 The mutant Δart is much hungrier for UQ₁₀ and COQ7 than WT

Since the Δart mutant, which lacks arthrobotrisins, formed significantly more trapping devices and exhibited stronger nematicidal activity than the WT strain, we hypothesized that the Δart mutant may have a higher demand for UQ₁₀ and COQ7 compared to the WT. To test this hypothesis, the Δart mutant was supplemented with UQ₁₀ at concentrations of 10, 50, and 500 μM under normal oxygen conditions. Two controls were included: a solvent-only control and an untreated control. Because the Δart mutant can spontaneously develop trapping devices, trapping device formation was induced both in the presence and absence of nematodes. In both scenarios, at all tested UQ₁₀ concentrations, the Δart mutant produced fewer trapping devices than the controls (Fig. 6A–C), with the inhibitory effect of UQ₁₀ on trapping device formation displaying a clear dose-dependent pattern. Notably, the Δart mutant at 10 μM UQ₁₀ without nematode induction exhibited a slight reduction in trapping device formation (Fig. 6A, B), whereas the WT strain treated with 10 μM UQ₁₀ in the presence of nematode did not (Fig. 2A, B). These results suggest that the Δart mutant has a higher requirement for UQ₁₀ than the WT strain in inhibiting trapping device formation.

Open image in new window

Interestingly, the trapping devices of the Δart mutant treated with 500 μM UQ₁₀ still exhibited strong nematicidal activity, albeit with a slight reduction compared to other treatments with lower concentrations of UQ₁₀ (Fig. 6D). These results suggest that UQ₁₀ supplementation may not fully satisfy the fungal requirements for iron sequestration despite its inhibitory effect on trapping device development. Subsequently, the nematode mutant Δclk-1, along with its WT nematode strain N2, was used to assess trapping device formation and nematicidal activity of the Δart fungal mutant. The Δclk-1 nematodes induced trapping device formation in the Δart mutant at levels comparable to those induced by N2 nematodes (Fig. 6E, F), confirming that the coq7 mutation in nematodes does not affect fungal trapping device formation. As expected, with or without UQ₁₀ supplementation, the Δart mutant, like the WT fungal strain, captured significantly more N2 nematodes than Δclk-1 nematodes (Fig. 6G). Notably, the Δart mutant captured 3.1 times more N2 nematodes than Δclk-1 nematodes, whereas the WT fungal strain captured 2.6 times more N2 nematodes than Δclk-1 nematodes, indicating a substantially higher demand for N2 nematodes by the Δart mutant compared to the WT strain (Figs. 6G and 2C). These results suggest that the trapping devices of the Δart mutant have a higher requirement for COQ7 than those of the WT fungal strain.

Three additional nematode mutants, Δacox, Δdsc, and Δlpd, were employed to evaluate the role of COQ7 in trapping device formation and nematicidal activity of the fungal Δart mutant. Compared to the nematode WT strain N2, the Δacox mutant exhibited a slightly stronger effect on inducing trapping device formation in the fungal Δart mutant, whereas the Δdsc, Δlpd, and Δclk-1 mutants displayed slightly weaker effects (Fig. 6H, I). However, no significant differences in the induction of fungal trapping devices by the Δart fungal mutant were observed among the nematode mutants and N2. Notably, the number of Δclk-1 nematodes captured by the fungal Δart mutant was the lowest among all nematode mutants tested (Fig. 6J), confirming that nematode COQ7 plays a crucial role in the nematicidal activity of fungal trapping devices, likely due to an attractive interaction between iron-rich trapping devices and the ferritin-like COQ7 protein.

2.11 Occurrence of art retainment and coq7 loss in NTFs during evolution

All eukaryotic organisms must maintain oxygen homeostasis, as oxygen is essential for biological processes but also poses significant risks to cellular function. Traditionally, it has been assumed that biological defenses against hyperoxia are less robust than those against hypoxia, since eukaryotes may not have encountered supraphysiological oxygen levels during evolution [52]. However, geological evidence suggests that rising atmospheric oxygen levels were a key driver in the origin and evolution of aerobic life [53, 54]. Global oxygen levels gradually increased from approximately 1.78% in the early Ordovician period (around 477 million years ago [Mya]) to about 19.88% in the middle Silurian period (around 427 Mya) [55]. At various points in history, atmospheric oxygen levels surpassed the current 21%, reaching approximately 30% during the Carboniferous (around 300 Mya) and Cretaceous (around 100 Mya) periods [55]. To date, limited research has addressed the role of oxygen in the evolution of metabolism, genomes, and morphology in NTFs.

We hypothesized that the loss of coq7 and the utilization of arthrobotrisins in NTFs under elevated oxygen conditions may be linked to the origin and evolution of trapping devices and their predatory behavior. To investigate this, we analyzed 21 genomes, including nine NTFs (Orbiliomycetes), four fungi from its sister lineage (Pezizomycetes), five fungi from Eurotiomycetes (all harboring the art gene cluster), two fungi from Saccharomycetes (which lack the art gene cluster), and one fungus from Taphrinomycetes (also lacking the art gene cluster) as an outgroup. Using this dataset, we constructed a time-calibrated phylogenetic tree with five calibration points employing MCMCtree. Notably, the loss of the coq7 gene, together with the loss of ccc1, occurred between 413 and 258 Mya, following the acquisition of the art gene cluster by the ancestor of NTFs approximately 464 Mya (Fig. 7). This finding suggests that acquisition of the art gene cluster may have functioned better in fungal adaptation to fluctuating oxygen levels, rendering the involvement of coq7 in fungal responses to elevated oxygen redundant or less essential, resulting in the loss of the coq7 gene in NTFs.

Open image in new window

2.12 Temperature and oxygen are key factors driving genomic changes in NTFs

To investigate the evolutionary forces driving the acquisition of the art gene cluster and the loss of coq7 in fungi, we mapped three biogeochemical parameters (atmospheric O₂ levels [55], average global temperature [56], and ironstone occurrences [57]) onto a time-calibrated evolutionary tree. Previous studies have shown that A. oligospora is a temperature-sensitive NTF that fails to grow above 30 ℃ [20, 58]. Notably, the formation of its trapping devices is positively temperature-dependent, which correlates with the inverse relationship between iron levels and temperature in fungal mycelia [20]. Geological data indicate that around 464 Mya, when the NTF ancestor acquired the art gene cluster, atmospheric oxygen levels gradually increased from approximately 1.78% in the early Ordovician (around 477 Mya) to 19.88% in the middle Silurian (around 422 Mya) (Fig. 7). During the same period, Earth's temperatures underwent a pronounced V-shaped fluctuation, with average global temperatures rapidly dropping to 12 ℃ and subsequently rising to 25 ℃ between 506 and 420 Mya. A similar, though less pronounced, trend was observed in ironstone levels during the Ordovician (Fig. 7). These findings suggest that the art gene cluster may have facilitated fungal adaptation to gradually increasing oxygen levels, consistent with the above finding that the art gene cluster functions under elevated oxygen conditions.

Interestingly, species with fruiting bodies in the class Pezizomycetes, which share a common ancestor with NTFs, lost the art gene cluster rather than the coq7 gene around 413 Mya, when atmospheric oxygen levels briefly dropped between 380 and 420 Mya (Fig. 7). This is consistent with the above finding that the art gene cluster and coq7 gene overlap in their function in fungal response to elevated oxygen levels. Additionally, fungal species in Orbiliomycetes and Sordariomycetes lost some genes (Ao273, Ao274, Ao276, Ao277, Ao279, Ao281) within the art gene cluster—or the entire cluster—as oxygen concentrations gradually declined from 37.2% to present-day levels after 263 Mya (Fig. 7). This correlation between oxygen levels and the evolutionary dynamics of the art gene cluster suggests that environmental oxygen concentration played a crucial role in the gain and loss of genes within this art gene cluster. Further analysis of gene gains and losses related to stress responses revealed an enrichment of oxidative stress-related genes at nodes 2, 3, 4, 5, 6, and 11, corresponding to periods of rising oxygen levels (Fig. S7A–B and Table S3).

2.13 Distribution of art analogues in Aspergillus and Penicillium

To better understand the distribution patterns of the art gene cluster within the fungal kingdom, we conducted a comprehensive survey of art gene cluster homologs across 2057 fungal genomes, using the A. oligospora art gene cluster as a reference. For further analysis, we retained only those hits that contained Ao283, the core biosynthetic gene of the art cluster, along with at least four additional art cluster genes, including Ao273, Ao274, Ao276, Ao277, Ao278, Ao279, Ao280, Ao281 and Ao282. To identify potentially pseudogenized versions of the cluster in fungal genomes lacking a complete set, we performed BLAT searches. This analysis revealed art gene cluster analogues in 103 fungal genomes, representing 86 species across 33 fungal genera. These fungal genomes were distributed among several fungal classes, including Eurotiomycetes (13.2%, 52/395), Sordariomycetes (6.6%, 32/488), Dothideomycetes (5.7%, 10/174), Lecanoromycetes (27.3%, 3/11), and Orbiliomycetes (66.7%, 6/9) (Fig. S8 and Table S4).

Global fungal data from the GlobalFungi database [59] indicates that 67% of the genera (22 of 33) containing the art gene cluster are dominant fungal genera, capable of surviving across a broad range of environmental conditions and, in some habitats, becoming the predominant group (Table S5). Moreover, the majority of species (87%, 73 out of 86) belonged to these dominant genera (Table S6). Further analysis of gene turnover within the art cluster across the 86 fungal species revealed that only species in Arthrobotrys, along with some in Aspergillus and Penicillium, retained an intact analogue of the cluster. In contrast, other species have lost one or more key biosynthetic genes (Fig. 8). The high occurrence of the art gene cluster in dominant fungal genera, together with its intact preservation in Arthrobotrys, Aspergillus, and Penicillium, suggests that this gene cluster may contribute to the ecological success and environmental adaptability of these fungi. This is particularly notable given that Aspergillus and Penicillium are among the most dominant fungal genera in modern ecosystems.

Open image in new window

3 Discussion

Phenotype, defined as the observable characteristics of an organism, arises from the complex interaction between its genetic blueprint (genotype) and environmental factors. Genetic defects, including gene loss or deletions, represent critical perturbations to this genetic blueprint and constitute fundamental causes of phenotypic diversity, influencing both normal variation and disease pathogenesis [60, 61]. Previous studies have demonstrated that the trapping devices of the NTF A. oligospora contain numerous iron-rich electron-dense bodies, suggesting that these trapping devices serve as an alternative iron sequestration strategy [20, 62]. This adaptation likely compensates for the loss of the CCC1-mediated vacuolar iron detoxification mechanism, which is otherwise conserved in most fungi [63]. Notably, the CCC1-mediated vacuolar iron storage mechanism was lost in NTFs during the Late Paleozoic Ice Age, and the subsequent formation of fungal trapping devices emerged during the Permian–Triassic mass extinction which is correlated with the significantly elevated global temperatures [20]. In particular, the common desferriferrichrome-mediated iron storage pathway, which is widely distributed in fungi, emerged in NTFs during the elevated temperatures of the Cambrian period [20, 64].

This study investigates how the genetic defect—specifically the loss of the coq7 gene in NTFs—interacts with the environmental factor oxygen to facilitate the fungal iron-rich trapping devices to capture nematodes. Genomic analyses indicate that among the Ascomycota, only nine NTF species lack the essential coq7 gene, which encodes COQ7, a membrane-bound di-iron protein belonging to a unique subset of the ferritin family and responsible for UQ biosynthesis [22–24]. Time-calibrated evolutionary analyses suggest that the last common ancestor of NTFs within the Orbiliomycetes lost both the ccc1 and coq7 genes during the Late Paleozoic Ice Age, a period characterized by cold "superoligotrophy, " marked by extremely low nutrient availability due to cold stress [65]. Previous research has shown that mutation in the clk-1 gene of C. elegans resulted in extended lifespan and generally slowed developmental rates [66]. Thus, the loss of the coq7 gene in Orbiliomycetes may have conferred a survival advantage during the prolonged cold "superoligotrophy" period.

We observed that the absence of the coq7 gene in NTFs renders these fungi sensitive to elevated oxygen levels, as increased oxygen availability promotes iron accumulation, potentially leading to iron overload. Phenotypic analyses revealed that the formation of trapping devices and their nematicidal activities were significantly higher in the predominant NTF A. oligospora under normoxic conditions compared to hypoxic conditions, suggesting that oxygen is a key driver of trapping device development. Moreover, UQ₁₀ markedly inhibited trapping device formation. The introduction of an exogenous coq7 gene also significantly suppressed the nematode-induced formation of trapping devices in A. oligospora. Importantly, fungal trapping devices exhibited a significantly enhanced ability to capture COQ7-expressing nematodes compared to COQ7-deficient nematode mutant Δclk-1. Previous studies suggest that nematodes are typically iron-deprived organisms [67, 68]. For example, C. elegans can detect pyoverdine, an iron-chelating siderophore secreted by Pseudomonas aeruginosa, and preferentially feeds on the bacterial prey [69]. Our further experiments indicated that the fluorescently labeled COQ7 protein from a genetic engineering Escherichia coli could adhere to the fungal trapping devices, while the ACOX control protein could not. Further in vivo experiments suggested that nematode COQ7 can be secreted and exposed to the nematode surface, while also indicating the affinity of the nematode COQ7 for the fungal trapping device.

Since the insertion of an exogenous coq7 gene in A. oligospora significantly reduces the formation of fungal iron-rich trapping devices, which is consistent with the previously reported function of COQ7 as an iron chelator [22–24]. Therefore, the deletion of coq7 in nematodes should also lead to an increase in free iron content in the mutant Δclk-1 nematodes. Most likely, iron-rich fungal trapping devices may repel iron-rich nematodes, resulting in the failure of A. oligospora to capture the mutant Δclk-1 nematodes without COQ7. In the predator–prey interaction between NTFs and nematodes, the ferritin-like COQ7 protein in nematodes may readily bind to iron-rich fungal trapping devices, potentially triggering fungal predation. This observation supports the dual function of fungal trapping devices: attracting and capturing nematodes [70].

Remarkably, we found that A. oligospora utilizes arthrobotrisins instead of UQ₈ in response to elevated oxygen levels, as UQ₈ is employed only under hypoxic conditions (Fig. 9). Arthrobotrisins that serve as key chemotaxonomic markers for A. oligospora possess a triprenyl-phenol backbone structurally similar to that of UQ₃. Triprenyl-phenols and their oxygenated derivatives are common fungal meroterpenoids derived from phenolic precursors [71]. Notable examples include the potent antifungal yanuthones produced by Aspergillus and Penicillium, as well as the phytotoxic macrophorins from Macrophoma, a pathogen responsible for apple fruit rot [72]. However, the biosynthetic pathway of these triprenyl-phenol metabolites has been elucidated only in A. oligospora, due to its relatively simple metabolism and reduced number of biosynthetic gene clusters [41, 47–50]. Previous studies have shown that disruption of the arthrobotrisin biosynthetic pathway significantly enhances the formation of trapping devices and nematicidal activity, but impairs fungal colonization in field soils [48, 50].

Open image in new window

In this study, mutation analysis revealed that the Δart mutant lacking arthrobotrisin biosynthesis lost the metabolic response observed in WT caused by elevated oxygen levels and instead significantly promoted the UQ₈ accumulation. Notably, unlike WT, which requires nematode induction to produce traps, the Δart mutant was capable of spontaneously forming trapping devices without nematode induction under both hypoxic and hyperoxic conditions. These results suggest that arthrobotrisin biosynthesis inhibits trapping device formation by modulating oxygen-mediated metabolism in A. oligospora. Like the fungal WT strain, the Δart mutant exhibited significantly enhanced nematicidal activity against COQ7-expressing nematode strains compared to COQ7-deficient nematode mutants. These findings suggested that the NTF A. oligospora utilizes built-in arthrobotrisin biosynthesis to inhibit oxygen-mediated formation of its own trapping devices, confirming that these trapping devices are physiologically a phenotypic manifestation of iron overload in the fungus. The limited nematicidal activity of these trapping devices is consistent with the therapeutic rationale underlying iron chelation strategies [73], which help explain why NTFs often form more trapping devices but exhibit reduced viability.

Interestingly, fungal species in the class Pezizomycetes, which share a common ancestor with Orbiliomycetes, lost the art gene cluster approximately 413 million years ago (Mya), whereas NTFs within Orbiliomycetes lost the coq7 gene. This divergence in gene loss strategies suggests that coq7 and the art gene cluster may fulfill analogous roles in adapting to fluctuating oxygen conditions, likely due to their structural and functional similarities. Phylogenetic analysis revealed that only the genus Arthrobotrys retains an intact art gene cluster, while other NTFs have partially or completely lost this cluster. This pattern corresponds with the widespread ecological distribution of A. oligospora, the most globally dispersed NTF.

In summary, this study investigates how the loss of coq7 in A. oligospora reshapes fungal metabolic responses to elevated oxygen levels, resulting in the formation of trapping devices that selectively target COQ7-expressing nematodes (Fig. 9). Namely, A. oligospora employs two distinct strategies to mitigate iron overload: one involving the acquisition of exogenous iron-chelating agents from nematodes, and the other through endogenous metabolic adaptation. Collectively, these findings further confirm that the formation of fungal trapping devices is a phenotype for iron overload due to a synergistic effect of genetic and environmental factors, highlight the utility of NTFs as a eukaryotic model for studying eukaryotic iron dysregulation or iron overload-related cellular responses.

4 The experimental section

4.1 Gene loss analysis in NTFs

We selected 21 representative genomes including 9 nematode-trapping fungi (Orbiliomycetes), 4 fungi from its sister lineage (Pezizomycetes), 5 fungi (all of these contain art gene cluster) evenly sampled across Eurotiomycetes, 2 fungi (without art gene cluster) from Saccharomycetes as well as 1 fungus (without art gene cluster) from Taphrinomycetes as an outgroup. All-versus-all BLASTP results of 207, 036 protein sequences from the 21 fungal genomes were input into OrthoFinder (v.2.5.2) [74] for gene clustering, in which the MCL algorithm was enabled by setting the inflation factor to 1.5. Orthologous gene table were analyzed in Count software [75] to investigate gene repertoire evolution and reconstruct gene family histories for both total and secreted genes. Gene familyies represented in fewer than three species were excluded from the rate estimation to ensure robustness. Evolutionary rate modeling was performed using a gain–loss-duplication framework, incorporating a Poisson distribution at the root and allowing for lineage-specific rate variation. The optimization procedure was conducted over 100 iterations with a likelihood convergence threshold set at 0.1. Gene content (gain and loss) was inferred using Dollo parsimony, complemented by posterior probability estimation under a birth-and-death process. Each inferred gene loss was validated by conducting BLASTp searches.

4.2 Phylogenetic tree construction of fungal genomes based on BUSCO genes

We systematically collected 2517 publicly available fungal genomes from the JGI MycoCosm database (Access: 2023-6-15) and NCBI genome database. To ensure data quality and completeness, we evaluated all genomes using the Benchmarking Universal Single-Copy Orthologs (BUSCO) [76] tool and retained only those with a completeness score higher than 90%. A total of 2057 high-quality fungal genomes were retained, with Ascomycota (1361 genomes, 66.2%) and Basidiomycota (540 genomes, 26.3%) dominating the dataset. The identified orthologous sequences from each genome were extracted from the respective BUSCO output directories. For each ortholog group, multiple sequence alignments were performed. These alignments were rigorously inspected for quality and subsequently concatenated into a comprehensive alignment matrix. This concatenated dataset provided a holistic view of the conserved genetic information across the sampled genomes. Maximum likelihood (ML) tree with 1000 bootstrap iterations were built.

4.3 Distribution pattern of UQs biosynthesis genes in 1361 Ascomycota genomes

At least 13 proteins (COQ1–COQ11, YAH1, and ARH1) are necessary for the synthesis of UQs. To elucidate the distribution patterns of these proteins in the fungal kingdom, protein sequences of COQ1–COQ11, YAH1, and ARH1 from Saccharomyces cerevisiae were used as queries. These sequences were used to search for orthologs via BLASTp (v.2.11.0 +) against a custom database containing all predicted protein sequences from 1, 361 high-quality fungal genomes. Our criteria for BLASTp were a bit scores greater than 50, minimum alignment length of 50 amino acids and minimum sequence similarity of 30%.

4.4 Distribution pattern of UQs biosynthesis genes in 2057 fungal genomes

The method is consistent with that described above. Presence and absence of each gene were mapped to the phylogenetic tree (containing 2057 fungal genomes) to visualize the distribution of UQs biosynthesis genes in fungal kingdom.

4.5 Fungal strain and culture conditions

The WT and its Δart mutant (ΔAo283) of A. oligospora YMF1.3170 was obtained from the State Key Laboratory for Conservation and Utilization of Bio-Resources & Key Laboratory for Microbial Resources of the Ministry of Education. Δart mutant was constructed according to a previous study [38]. PDA (potato 200 g/L, glucose 20 g/L, agar 15 g/L), water agar (15 g/L agar) media were used for analyzing mycelial growth and related phenotypic traits. For the flask culture, strain was initially cultured on PDA medium at 28 ℃ for 10 days to obtain conidia, and then the conidia were inoculated in 250 mL of liquid PDB medium (potato 200 g/L, and 20 g of glucose per liter) with a final concentration of 1 × 10⁵ conidia/mL under shaking (180 rpm) and static conditions (0 rpm) at 28 ℃.

4.6 Acquisition and cultivation of nematode mutant strains

The C. elegans N2 and mutant strains (Δclk-1, Δacox, Δlpd and Δdsc) used in this study were obtained from the State Key Laboratory for Conservation and Utilization of Bio-Resources & Key Laboratory for Microbial Resources of the Ministry of Education. Mutants were constructed in previous studies [34, 35]. Both N2 and mutant strains were requested through the standard organization strain request process. Upon arrival, the nematodes were maintained at 20 ℃ on Nematode Growth Medium (NGM) plates seeded with Escherichia coli OP50, as per standard cultivation protocols. The mutation status of the strains was confirmed by PCR amplification, as detailed in the strain descriptions provided by the repository. Nematode populations were synchronized. In brief, gravid adults were transferred onto fresh NGM plates after growing for about 2 days at 20 ℃, and eggs were collected after allowing the adults to lay eggs. The eggs were then allowed to hatch overnight at 20 ℃, and synchronized populations of L1 larvae were isolated. These larvae were grown to the desired developmental stage (L4) before being used for experimental purposes.

4.7 Construction of coq7 heterologous expression strain (E-coq7) using A. oligospora

To construct a heterologous expression plasmid, the PUC19-1300-D-HYB (PUC19) plasmid was amplified using fusion polymerase chain reaction (PCR) with Phanta Max SuperFidelity DNA Polymerase (Vazyme, Nanjing, China), following the manufacturer's instructions. The primers used in this study are listed in Table S7. The heterologous fragment was amplified from Saccharomyces cerevisiae BJ5464 and purified. The coq7 gene was inserted into the PUC19 plasmid, which contains the strong PtrpC promoter and TtrpC terminator from Aspergillus niger. The coq7-plasmid and blank plasmid transformed into A. oligospora protoplasts, respectively. After 2 − 4 days of incubation at 28 ℃, transformation colonies were selected and transferred to new PDA plates. Genomic DNA from putative transformants was extracted after 5 days of incubation at 28 ℃, and PCR was performed to confirm the integration of the target gene into the genome. The mutant was identified and verified by PCR and sequencing analysis. The WT strain and E-coq7 mutant were cultured on PDA medium at 28 ℃ for 6 days. Vegetative hyphae were harvested from 9-cm Petri dishes, which were then incubated at 28 ℃ for 3 and 5 days. Following induction with nematode, the hyphae were collected and immediately frozen in liquid nitrogen. Total RNA was extracted from all samples using the AxyPrep Multisource RNA Miniprep Kit (Axygen, Jiangsu, China). This extracted RNA was reverse transcribed into cDNA using the FastQuant RT Kit with gDNase (Takara, Kusatsu, Japan). The cDNA was used as a template to analyze the expression of the coq7 gene, with cDNA from S. cerevisiae BJ5464 serving as the positive control template.

4.8 Prey preference assay of A. oligospora on different nematode mutants

To assess the selective preference of A. oligospora toward different nematode mutants, nematocidal activity assays were performed using WT (N2) and mutant strains (Δclk-1, Δacox, Δlpd, and Δdsc). Fungal mycelia were cultured on standard 9 cm PDA plates at 28 ℃ for 4 days until mature trapping devices had developed. Approximately 400 synchronized L4-stage nematodes from each mutant were separately suspended in sterile water at equal densities. Three hundred microliters of each suspension were then transferred to independent Petri dishes containing fungal cultures under identical growth conditions. Plates were gently swirled to evenly distribute nematodes, and all dishes were incubated in parallel within the same incubator at 28 ℃ to ensure uniform environmental conditions. Captured nematodes were monitored at 24 h post-inoculation under a stereomicroscope. For each sample, 6 randomly selected visual fields were examined, and nematodes were considered captured when clearly immobilized within trap rings. Nematode mortality was assessed based on the absence of movement in response to physical stimulation with a fine needle, with straight body posture used as a supporting criterion [19]. All experiments were repeated in triplicate for each mutant.

4.9 Fungal colony growth under normal oxygen and low oxygen conditions

Anaerobic Gas Generator is a disposable oxygen-absorbing and carbon dioxide-generating agent for use in anaerobic jars (C-1, Mitsubishi Gas Chemical Company, Inc.). The mutant strain ∆art and the WT were cultured on PDA medium within anaerobic bags. One sachet of anaerobic gas generator is used for a 2.5-L anaerobic jar, and the sachet is discarded after 30 min of reaction to create a low-oxygen environment (approximately 10% oxygen). Smart Sensor (Model: AR8100) oxygen detector was used to detect the oxygen concentration inside the anaerobic jars. Both WT and mutant strains were cultured on PDA at 28 ℃ in normal oxygen and low oxygen conditions, and their growth rates and colony morphology were observed. Hyphal morphology was examined using 5-day-old PDA cultures grown at 28 ℃.

4.10 Trap formation and nematicidal activity assays under normal oxygen and low oxygen conditions

WT and mutant strains were cultured on 9 cm agar plates under normal oxygen and low oxygen at 28 ℃ for 3 − 4 days, and then about 400 nematodes (C. elegans) were introduced to each culture. After 12 h, traps and captured nematodes per plate were observed under a microscope and counted at specific time-points. Three hundred microliters of the nematode suspension were then transferred to each Petri dish and gently mixed. All dishes were incubated at 28 ℃. Nematode mortality was assessed at 12h exposure periods using a normal binocular microscope, and four visions (1cm²/per) in each Petri dish were picked at random for counting. Nematodes were considered to be dead when their bodies were straight, and they failed to move on physical stimuli with a fine needle. Nematicidal activity was evaluated according to the mean corrected percentage of dead nematodes. All tests were repeated three times, and the data obtained were statistically analyzed.

4.11 LC–MS/MS sample extraction and preparation

The WT and mutant strain were inoculated in 250-ml PDB flasks cultured at 28 ℃ under shaking(180rpm, 120rpm, 60rpm) and static conditions (0 rpm) for 7 days prior to analysis by high performance liquid chromatography with diode array detection MS (UPLC-DAD/MS). The fermentation broths were extracted with an equal volume of ethyl acetate, and the organic layers were evaporated and dissolved in 1 ml of menthol and filtered through a 0.22 μm filter membrane. All samples were subsequently analyzed using an HPLC system coupled with a Q Exactive Focus Orbitrap mass spectrometer (Thermo Fisher) and equipped with an Agilent Zorbax ODS 4.6 by 250 mm column (Agilent, Santa Clara, CA) in the electrospray ionization (ESI) mode. The data were acquired in both positive and negative ion modes. To ensure the accuracy of compound identification and quantification, QC samples were injected periodically throughout the analysis to monitor instrument stability and data reproducibility. The metabolites extracted from WT and mutant strains A. oligospora were analyzed by LC–MS following the established methods in the literature.

4.12 LC–MS/MS data processing and analysis

Raw MS data files generated from the LC–MS/MS system were processed using Thermo Scientific Xcalibur software for data acquisition and processing. Raw MS data were also imported into Compound Discoverer (version 3.3) software for comprehensive data analysis. Peak detection was performed using the built-in algorithm. The raw data were processed to identify and align chromatographic peaks across all samples using a reference sample. A list of features, including accurate mass, retention time, and ion intensities, was generated for each sample. Feature Finder module was used to extract relevant features from the chromatograms. The extracted features were matched to compounds using the mzCloud databases within Compound Discoverer. The software utilized both the exact mass of detected ions and their fragment ion patterns to propose possible compound identities. The ion intensities of identified compounds were normalized against total ion count (TIC). Quantitative analysis was performed to determine the relative abundance of each metabolite in the different sample groups. Statistical analysis was conducted using the software's built-in tools, including principal component analysis (PCA) and hierarchical clustering, to visualize group separation and identify significantly different compounds across mutant Δart under shaking and static conditions.

4.13 Quantitative real-time PCR analysis

The WT and mutant strain were inoculated in 250-ml PDB flasks cultured at 28 ℃ under shaking(180 rpm) and static conditions (0 rpm) for 7 days. After induction, hyphae were collected and frozen immediately in liquid nitrogen. Total RNA was extracted from the mycelia of each sample using an RNA extraction kit (Axygen, Jiangsu, China) and then reverse transcribed into cDNA using the PrimeScriptHRT reagent kit (with genomic DNA (gDNA); TaKaRa, Kusatsu, Japan). The resulting cDNA was used as a template for RT-qPCR to analyze the mRNA expression of Ao273-286 with specific paired primers (Table S8) and the LightCycler 480 with SYBR green I master mix (Roche, Basel, Switzerland). β-Tubulin was used as an internal standard. All RT-qPCR experiments were performed in triplicates. The relative transcript level (RTL) of each gene was calculated as the difference of the cycle threshold (CT) value for the gene β-Tubulin and Ao273-286 under static condition and shaking condition.

4.14 RNA-seq data analysis

Raw reads from both WT and Δart samples were examined using FastQC. To obtain clean data, fastp (v.4.10.0) [77] was used with default parameters to filter adaptor sequences and remove low-quality reads. The reference genome and gene model annotation files for A. oligospora ATCC 24927 were downloaded from NCBI's Genome Browser. Genome indexing was performed using 'hisat2-build' in Hisat2 (v.2.2.1) [78], incorporating splice sites and exon information. The filtered paired-end reads were aligned to the reference genome using Hisat2. Gene abundance for each sample was estimated from the aligned BAM files. Differential expression analysis was conducted. Resulting p-values were adjusted using the Benjamini–Hochberg approach to control the false discovery rate.

4.15 Assay for iron contents

The WT and mutant strain were inoculated in 250 mL potato dextrose broth (PDB) flasks and cultured at 28 ℃ (180 rpm or 0 rpm) for 7 days. The fermentation broth was extracted with the equal volume of ethyl acetate, the mass of mycelium was weighed and recorded, then dissolved in 30 mL of distilled water and homogenized to obtain a well-ground mixture. Subsequently, 1 mL of the mixture was centrifuged at 12, 000 rpm for 5 min to remove the supernatant for the following experiments. An iron assay kit (#ab83366, Abcam) was used for evaluation of the intracellular ferrous iron level (Fe²⁺ and Fe³⁺) in fungi. First, mycelia (20 − 80 mg) were collected in phosphate-buffered saline (PBS) and homogenized in iron assay buffer using a dounce homogenizer sitting on ice with 50 − 100 passes. Iron reducer was then added into the collected supernatant, thoroughly mixed, and incubated. Finally, an iron probe was added, mixed, and incubated for 1 h. The resulting solution was immediately measured on a microplate reader at OD = 593 nm.

4.16 Identification homologue of art gene cluster in the fungi kingdom

Biosynthetic gene clusters for each genome were annotated by antiSMASH 7.0 (v.7.0.0) [79] via command line version with default parameters, except for the argument 'taxon fungi'. In total, 84, 476 secondary metabolism gene clusters were annotated from 2057 fungal genomes. A local database was generated based on all the gene clusters using the 'cblaster makedb' module (v.1.3.18) [80]. To identify art gene cluster homologous in fungal kingdom, the search function of cblaster was used to query amino acid sequences of 14 genes in art gene cluster (Ao273–Ao286) against the local database constructed above. Only resultant hits containing Ao283, the core biosynthetic gene of art, as well as at least four other genes in this cluster, were considered as putative art homologues. We found 103 fungal genomes containing art gene clusters homologues. We further conducted additional analyses to strengthen the functional annotation of the core biosynthetic genes within the art-like clusters identified in different species. The predicted protein sequences of the core genes in each art-like cluster were submitted to the NCBI Conserved Domain Database (CDD). This analysis revealed that all identified core genes share the same conserved domain architecture, including PksD, KR_FAS_SDR_x, PKS_PP and hot_dog superfamily, consistent with the functional modules typically found in Ao283 (Table S9). The distribution pattern of presence and absence of art gene cluster was mapped to the phylogenetic tree (containing 2057 fungal genomes) to visualize the distribution of the gene cluster in the fungi kingdom.

4.17 Gene members of art gene cluster among various fungal species

The 103 fungal genomes containing the art gene cluster were used for this analysis. When a species had multiple genomes containing PKS-TPS gene cluster, only the genome with the highest number of gene members in this gene cluster was retained as the representative of that species. We identified 86 fungal species containing the PKS-TPS gene cluster. Evolutionary relationships of these 86 fungal species were extracted from the phylogenetic tree, which included 2057 fungal genomes. A heatmap was generated to illustrate the presence, absence, or expansion (gene copies > 2) of these genes across these fungal species.

4.18 Phylogenetic tree construction for nematode-trapping fungi in Ascomycota

In the section of gene loss analysis, 1256 sets of 1:1 orthologous protein encoding genes were identified among the 21 fungi species. Amino acid sequences of each orthogroups were aligned with with the 'auto' parameter, and PAL2NAL (v14) [81] was used under default parameters to create a codon alignment from MAFFT-aligned amino acids. The aligned orthogroups amino acid sequences and its codon alignment were concatenated and used to construct an amino acid (AA) and a nucleotide (NT) data matrix, respectively. Based on AA data matrix, we determined the best model using the "-m TEST" parameter, which automatically estimates the best fitting model of substitutions according to Bayesian Information Criterion values (BIC) using IQ-TREE2 (v. 2.2.2.7) [82]. A maximum likelihood (ML) tree with 1000 bootstrap iterations was built based on the best fitting model.

4.19 Function annotations of homologous groups

To assign function annotations to each homologous group (OG), we selected a representative sequence for each group. First, we ordered the sequences within each OG by their lengths. Next, we employed the UCLUST algorithm to cluster the sequences within each OG based on 30% amino acid sequence similarity. The sequence with the highest abundance in each OG was designated as the representative sequence. In cases of equal abundance, the longest sequence was chosen as the representative. For each representative sequence, gene function and gene ontology (GO) were predicted using the PANNZER2 web server [83].

4.20 Estimating divergence times of nematode-trapping fungi in Ascomycota

To estimate the divergence times, fourfold degenerate (4D) sites were extracted using get4foldSites (https://github.com/brunonevado/get4foldSites) and concatenated into a single supermatrix. The Bayesian relaxed molecular clock approach was employed to estimate species divergence time using MCMCTREE in PAML (v.4.10.0) [84] (clock = independent rates, burn-in = 10, 000, sample number = 100, 000, sample frequency = 10, model = JC69) based on 278, 923 fourfold degenerate sites from 1256 single copy orthologous genes and calibrated with one fossil calibration point (A extinct NTF group with ring trapping structure: > 100 Mya) [85]. To compensate for the availability of limited direct fossil evidence, five phylogenetically deep nodes were calibrated using rigorously validated secondary calibration points derived from highly supported, multi-evidence-based divergence time estimates: node1 (498–571 Mya) [86], node 4 (357–459 Mya) [87], node 12 (216–267 Mya) [88], node 15 (108–181 Mya) [88] and node 17 (64–135 Mya) [88]. We used Tracer (v.1.7.1) [89] to visually check convergence of parameters across replicate MCMC chains in report of the MCMC runs. All the resulting effective sample sizes (ESS) ranged from 291 to 1306, suggesting that the parameters across replicate MCMC chains reached convergence (The effective sample size (ESS) should be higher than 200).

4.21 Dynamic landscape of gene members in art gene cluster during the evolutionary timeline

Based on the time-calibrated phylogenetic tree, we mapped three representative biogeochemical properties—atmospheric O₂ [55], average global temperature [56], and the number of ironstone occurrences [57] during the Phanerozoic eon onto the tree. Gene members of art gene cluster (Ao273–Ao286) and UQs synthesis-related genes of 21 genomes in phylogenetic tree were also mapped onto the tree to display their loss and gain at different nodes.

4.22 Expression, purification, and feeding of recombinant fluorescently labeled COQ7-EGFP/ACOX-EGFP fusion proteins

The complete coding sequences of acox and coq7 from Saccharomyces cerevisiae were amplified by PCR using gene-specific primers and cDNA as the template. The purified PCR products were individually cloned into the pCold II expression vector, which contains an N-terminal His tag and a cold-shock promoter, to generate C-terminal fusions with enhanced green fluorescent protein (EGFP). The resulting vector designated pCold II-His-ACOX-EGFP and pCold II-His-COQ7-EGFP, were verified by DNA sequencing to confirm the accuracy of the insert sequences and reading frames. The validated recombinant plasmids were transformed into the E. coli expression strain BL21 (DE3). A single colony was inoculated into LB medium containing ampicillin (100 μg/mL) and cultured at 37 ℃ with shaking at 200 rpm until the OD₆₀₀ reached 0.5–0.8. Protein expression was induced by adding isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM, followed by incubation at 16 ℃ with shaking at 200 rpm for 16 h. After induction, bacterial cells were harvested by centrifugation at 4 ℃ and 8000 rpm for 5 min. The pellet was resuspended in ice-cold lysis buffer and lysed by sonication on ice. Cell debris was removed by centrifugation at 4 ℃ and 8000 rpm for 30 min, and the supernatant was collected as the crude extract. The supernatant was applied to a Ni–NTA affinity column pre-equilibrated with lysis buffer. After sample loading, the column was washed with 10 column volumes of wash buffer to remove nonspecifically bound proteins. Bound proteins were eluted with 4 column volumes of elution buffer. Eluted fractions containing the target proteins were pooled and concentrated using an Amicon Ultra-15 centrifugal filter device. During concentration, the buffer was exchanged three times with desalting buffer to remove residual imidazole completely. The purity, molecular mass, and specificity of the purified proteins were evaluated by SDS–PAGE, and protein concentrations were quantified using the BCA assay. Fluorescence activity was confirmed by fluorescence microscopy to verify the structural integrity and functionality of EGFP. For feeding assays, spores of A. oligospora were spread onto water agar plates and incubated for 3–4 days to allow trap formation. Purified protein solution was then added to the medium, and the plates were immediately examined under a microscope.

4.23 Construction of C. elegans strains expressing CLK-1: EGFP fusion protein

Genomic DNA from C. elegans WT (N2) was used as template to amplify a fragment containing approximately 2.0 kb of the upstream promoter region and the complete coding sequence (CDS) of the clk-1 gene via PCR with gene-specific primers. Homology arms (15–25 bp) complementary to the ends of the linearized vector were added to the 5′ ends of the primers. The vector pCFJ97-75, which carries the dominant selection marker rol-6, the EGFP reporter gene, and an ampicillin resistance gene, was linearized by inverse PCR. The clk-1 promoter fragment was seamlessly inserted into the linearized vector using homologous recombination, resulting in the fusion expression construct p[clk-1p: : CLK-1:: EGFP; rol-6]. The ligation product was transformed into E. coli DH5α competent cells and plated onto LB agar containing ampicillin. Positive clones were selected and confirmed by colony PCR and sequencing to verify the correct insertion, orientation, and in-frame fusion. The validated reporter plasmid was mixed with the co-injection marker plasmid pCFJ104, which expresses a red fluorescent protein specifically in body wall muscle cells to facilitate visual screening under a fluorescence microscope. The final injection mixture contained the reporter vector at 50 µg/mL and pCFJ104 at 1 µg/mL, diluted in sterile TE buffer or standard nematode injection buffer. This DNA mixture was microinjected into the gonad of young adult N2 hermaphrodites. Injected P0 animals were transferred to fresh NGM plates and maintained at 20 ℃. After 3–4 days, F1 progeny were screened under a stereomicroscope equipped with fluorescence optics, and individuals exhibiting both red fluorescence in the body wall muscle and broad green fluorescence were selected.

4.24 Statistical analysis

Statistical analyses and related plots were performed using the R software environment (version 4.4.2). Significance tests of all quantitative measurements in experimental groups were determined using two-sided unpaired Student's t-test. Differential gene expression analysis was performed through Wald test, following normalization of RNA-seq data. P-values were adjusted using the Bonferroni correction method. Statistical significance thresholds were defined as P < 0.05, ^**P < 0.01, and *^**P < 0.001, with corresponding asterisks denoting these significance levels in all graphical representations. Data was obtained from three biological replicates unless stated otherwise. Graphical data are presented as mean ± s.d.

Notes

Author Contributions

Qunfu Wu and Jiao Zhou conceived and designed the experiments; Jiao Zhou, Donglou Wang, Songhan Xue, Ling Li, Li Wu and Junxian Yan performed the experiments; Qunfu Wu and Jiao Zhou analyzed the data; Qunfu Wu and Jiao Zhou wrote the paper. All authors reviewed the results and approved the final version of the manuscript.

Funding

This work was sponsored by the Central Guidance Fund for Local Science and Technology Development (grant no. 202407AB110021), Projects 202201BF070001-012 from "Double tops" Program from Yunnan Province and Yunnan University, 202201AT070172 from Yunnan Province and Scientific Research Fund of Education Department of Yunnan Province and Graduate Research Innovation Fund of Yunnan University (KC-24249015).

Data availability

Our sequences files of transcriptome are accessible from the National Center for Biotechnology Information (NCBI) under BioProject accession: PRJNA1195634 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1195634).

Declarations

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.