Whole plants of N. nipponica were collected with adhering soil from Nankoku-shi, Kochi Prefecture, Japan, in June 2013. Underground parts were washed thoroughly under running tap water to remove soil particles and checked for mycorrhizal colonization. Mycorrhizal parts of three individuals were cut into pieces 1–2 cm long, from which longitudinal sections including the stele (200–500 μm in length) were prepared. Pelotons visualized in the sections were observed under a compound microscope (BX51, Olympus, Tokyo, Japan). After the epidermis was removed from the sections, they were transferred to distilled water and incubated at 4 °C for 1 h, dried on sterilized filter paper for 1 min, and placed in 1/6 nutrient dextrose yeast agar (1/6 NDY) medium (Masuhara and Katsuya 1992). The dishes were incubated for 22–36 days at 4 °C in the dark. To establish the fungal strain in pure culture, a single hypha was isolated from each section and incubated on a plate of potato agar [PA; 1000 mL of filtrate from 200 g of boiled potato and 20 g of agar (Wako Pure Chemical Industries, Ltd., Osaka, Japan)] at 17 °C in the dark. Established cultures were maintained on PA at 17 °C in the dark and transferred to fresh medium every month. Two strains, T469 and T470, were obtained, both of which were incubated on potato dextrose agar [PDA (BD Difco, Franklin Lakes, New Jersey, USA)] and deposited in the Research Center of Genetic Resources, NARO (hereafter, T469 and T470 are designated MAFF425704 and MAFF425705, respectively).

Hyphae of MAFF425704 growing on cellophane over PA were embedded in 1.5% water agar, from which solidified pieces (2–5 mm²) containing hyphae were cut. For fixation, these agar blocks were placed in 2% paraformaldehyde and 2.5% glutaraldehyde in 25 mM phosphate buffer (PB) adjusted to pH 7.0 at room temperature for 2 h. After being washed three times in 25 mM PB, the agar blocks were postfixed with 1% osmium tetroxide in 25 mM PB at 5 °C overnight. The blocks were dehydrated in an ethanol series (10%, 30%, 50%, 70%, and 90% for 15 min per step, followed by 95% once and 100% twice for 20 min each time) and embedded in Agar Low Viscosity Resin (Agar Scientific, Essex, UK). Ultrathin sections (50–100 nm) were prepared with a diamond knife set on an RMC MT-XL ultramicrotome (RMC Products, Tucson, Arizona, USA). Sections were stained with 2% (w/v) uranyl acetate and lead citrate (Reynolds 1963) and observed with a transmission electron microscope (HT7700, Hitachi, Tokyo, Japan) at an accelerating voltage of 80 kV. A three-dimensional model of the septal pore cap of strain MAFF425704 was reconstructed from the serial TEM section images using paper clay. The overall structure, including the thickness of the pore cap and the size and distribution of perforations, was determined based on the TEM images.

To investigate the optimal temperature for mycelial growth, a single 4-mm agar block of each isolate (MAFF425704 and MAFF425705) was cut from a precultured colony and grown on PDA for six days at 20 °C in the dark. The blocks were subsequently placed at the center of new PDA medium (9 cm in diameter) and incubated for three days at 17 °C in the dark. These plates were then incubated at one of seven temperatures (5 °C, 10 °C, 15 °C, 20 °C, 25 °C, 30 °C, and 35 °C) in the dark. For each incubation temperature, the position of the edge of the mycelial colony was observed and recorded every 24 or 72 h for one month, by which time the colonies had grown to the edge of the inner plate. The growth rate was calculated from the average of 12 replicates. A one-way analysis of variance (ANOVA) followed by a Tukey–Kramer honest significant difference (HSD) test was performed to compare growth rates across temperatures using R 4.3.2 (R Core Team).

To observe the number of nuclei within the hyphae, mycelia from pure cultures on PDA were fluorescently stained after plating onto glass and mounted in 10 ppm 4,6-diamidino-2-phenylindole (DAPI) and 200 mg/mL Calcofluor White in the dark at 25 °C for 10 min. Nuclei were observed under a fluorescence microscope (Axioscope 5, Carl Zeiss, Switzerland) using an excitation filter (330–385 nm) and absorbing filter (420 nm).

Morphological characteristics of the fungal colonies on different culture media were examined. A single 4-mm agar block, precultured on PDA in the dark for six days, was placed on each of the following media: CMA, OA, 1/6 NDY supplemented with 100 ppm streptomycin and 50 ppm tetracycline, PA, PDA, and MA; cultures were incubated at 20 °C in the dark. After 14 days, colony color and morphological characters of aerial hyphae were observed.

Fruitbody induction was further examined. Mycelial disks (5 mm in diameter) taken from the edges of 20-day-old colonies on PDA culture were inoculated into sterilized cedar (Cryptomeria japonica) and evergreen Japanese oak (Quercus spp.) sawdust in Petri dishes 9 cm in diameter. The moisture content of the sawdust substrate was adjusted to 65% by mixing sawdust and bran at a ratio of 3:1. These dishes were incubated at 22 °C in the dark for 50 days, and the development of the colonies was observed under a stereomicroscope, optical microscope, and scanning electron microscope (Miniscope TM-1000, Hitachi, Tokyo, Japan). To stimulate fruitbody production, portions of the culture medium 2 cm in diameter were transferred to fresh sawdust medium in 650 mL bottles and 1.0 kg plastic bags. After 50 days, the culture medium was flooded with tap water for 1 h, the water was decanted, and the medium was irradiated and then cultured again at 20 °C for 30 days.

In Basidiomycota, higher-level taxonomy at the order or family level has often been resolved by multigene phylogenetic approaches (e.g., Binder et al. 2010; Hodkinson et al. 2014; Sjökvist et al. 2014; Johnston et al. 2019). Therefore, to provide greater resolving power in pinpointing the phylogenetic position of uAX, two additional commonly used loci, rpb2 and tef1-α, were sequenced (Matheny et al. 2007; Binder et al. 2010).

PCR amplification was conducted using the four types of uAX DNA samples previously analyzed by Nomura et al. (2013) from N. nipponica mycorrhiza (type I: M293; type II: M1051; type III: M38; type IV: M1277; Suppl. material 6: table S1), with the primer pairs fRPB2-5F/bRPB2-7R2 for rpb2 (Liu et al. 1999; Matheny et al. 2007) and EF1-983F/EF1-1567R for tef1-α (Rehner and Buckley 2005). PCR products that were difficult to sequence directly were cloned using the pGEM-T Vector System II (Promega, Madison, Wisconsin, USA). Twelve positive clones per PCR product were randomly chosen for sequencing. PCR products were purified using ExoSAP-IT (Affymetrix, Santa Clara, CA, USA) and sequenced on an ABI3130xl automated capillary sequencer (Applied Biosystems, Foster, CA, USA) with the BigDye Terminator 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster, CA, USA), following the manufacturer’s instructions. The following internal primers were used for sequencing in addition to the PCR primers for rpb2: bRPB2-6F and gRPB2-6R (Liu et al. 1999; Matheny 2005).

To identify the two established fungal isolates (MAFF425704 and MAFF425705), PCR amplification and sequencing of ITS and 28S rDNA were performed using the ITS5/ITS4 and NL1/NL4 primer pairs, respectively (White et al. 1990). Additionally, sequences of the tef1-α and rpb2 regions for MAFF425704 were obtained by the aforementioned procedure.

To determine the phylogenetic placement of uAX, sequence data representative of the main Basidiomycota lineages were selected. Sequences of 28S and 5.8S rDNA, rpb2, and tef1-α were obtained mainly from the Assembling the Fungal Tree of Life (AFToL) database (Suppl. material 6: table S1). Phylogenetic analyses were performed using a combined dataset of 28S, 5.8S, rpb2, and tef1-α sequences (four-locus dataset), as well as each locus independently, under maximum likelihood (ML) criteria and Bayesian inference.

For the four-locus dataset, two or three representative sequences of all existing Basidiomycota orders (He et al. 2019; Varga et al. 2019; James et al. 2020) were downloaded from the International Nucleotide Sequence Database (INSD, DDBJ/EMBL/GenBank) (Suppl. material 6: table S1). In particular, all families of Cantharellales and Phallomycetidae (Gomphales, Hysterangiales, and Geastrales) were included because the results of Nomura et al. (2013) suggested that these groups had a close affinity to uAX. Three sequences of Ascomycota (Morchella galilaea, Saccharomyces cerevisiae, and Tuber magnatum) were chosen as outgroups. Each dataset (28S, 5.8S, rpb2, and tef1-α) was aligned separately using the MAFFT 7 online server (https://mafft.cbrc.jp/alignment/software/) (Katoh and Standley 2013), with poorly aligned sites identified using Gblocks 0.91b (Castresana 2000). Individual alignments for the four regions were then combined into a single dataset.

Maximum likelihood analyses were performed using IQ-TREE 2.3.6 (Minh et al. 2020). The TIM3+I+G4, TIM3u+G4, TIM3+F+I+G4, and GTR+I+G4 models of nucleotide evolution were selected as the best substitution models for 28S, 5.8S, tef1-α, and rpb2, respectively, following evaluation with the Bayesian information criterion (BIC) by ModelFinder (Kalyaanamoorthy et al. 2017) implemented in IQ-TREE. Shimodaira–Hasegawa approximate likelihood ratio test (SH-aLRT) and ultrafast bootstrap values were obtained with 10,000 replicates each and used as measures of branch support. The final tree was visualized with the Interactive Tree Of Life (iTOL) (Letunic and Bork 2006).

Bayesian analyses were conducted in MrBayes 3.2.7a (Ronquist et al. 2012). The GTR+G model was applied for 28S, GTR+I+G for 5.8S and rpb2, and SYM+I+G for tef1-α. State frequencies, substitution matrices, shape parameters, and the proportion of invariable sites were estimated for each locus. Four independent Markov chain Monte Carlo (MCMC) chains were run, sampling every 1000^th tree until the average standard deviation of split frequencies (ASDSF) became <0.01. The log files of MrBayes were further analyzed using Tracer 1.7 (Rambaut et al. 2018) to check the effective sample sizes, which were always >200, indicating sufficient independent sampling to estimate the posterior distribution of each parameter. The first 10% of the sampled trees were discarded as burn-in. The remaining trees were used to construct a 50% majority-rule consensus tree, which was visualized with iTOL.

The five closest matching sequences for each of the four types of uAX were retrieved by BLAST searches (Altschul et al. 1994) using the 28S, rpb2, and tef1-α loci. These sequences were then added to each dataset for further analysis (accessed on 14 October 2025; Suppl. material 7: table S2). In addition, four additional major novel lineages within Agaricomycetes (GS28, GS29, GS30, and branch 5) were included in the 28S dataset; these lineages had been identified from soil DNA samples using metabarcoding and phylogenetic approaches (Suppl. material 8: table S3; Tedersoo et al. 2017). For the protein-coding genes (rpb2 and tef1-α), nucleotide sequences were translated into amino acid sequences to confirm correct alignment and minimize potential codon misalignment. Amino acid alignments were then used to guide the corresponding nucleotide alignments. Introns were excluded for both loci. Gblocks trimming was not applied to rpb2 and tef1-α to preserve codon positions for partitioned codon-based phylogenetic analyses. Partitioned evolutionary models were applied to rpb2 and tef1-α by separating the first, second, and third codon positions of each gene for ML analyses (Suppl. material 9: table S4). In Bayesian analysis, codon-partitioned analyses were initially performed for tef1-α and rpb2; however, because of poor convergence of the MCMC chains, the final Bayesian analyses were conducted using locus-based partitions.

The nuclear ribosomal DNA (nrDNA) ITS region, a fungal DNA barcoding region, is suitable for grouping samples based on sequence similarity and inferring the ecology and distribution of target fungal taxa (Nilsson et al. 2019). In assigning taxonomic groups, a threshold value of 90% sequence identity is used for genus, 85% for family, 80% for order, and 75% for class (Tedersoo et al. 2014; Nilsson et al. 2019). ITS divergence was investigated within the four uAX types, as well as between uAX and the eight sequences of Tremellodendropsidales, the most likely sister lineage of uAX. These eight sequences included those from Tremellodendropsis tuberosa (KU900857–KU900859 and KU900861; Berbee et al. 2016) and environmental sequences belonging to Tremellodendropsidales (KY687729 and KY687667; Tedersoo et al. 2017; JX316510 and JX316518; Truong et al. 2017).

To explore the ecology and distribution of close relatives of uAX, massBLASTer analysis was conducted on ITS sequences of the four types of uAX in the UNITE database, including INSD and high-throughput sequencing (HTS) data (Abarenkov et al. 2023). Data for 30 representative sequences for each species hypothesis (SH) were selected from the BLAST results for each of the four types of uAX, and information on their distribution, origin, and other geographic characteristics was downloaded (accessed on 15 October 2025). Preliminary phylogenetic analyses using Tremellodendropsidales, Trechisporales, or both as outgroups indicated that high genetic divergence between ingroup and outgroup taxa resulted in low resolution in terms of both tree length and branch support. Therefore, unrooted phylogenetic analysis was performed with the retrieved sequences (Suppl. material 10: table S5). All alignment files used in this study have been deposited in Figshare (http://doi.org/10.6084/m9.figshare.32592672) and are also provided as Suppl. materials 11–15.

Two culture strains (MAFF425704 and MAFF425705) were successfully isolated from the mycorrhiza of N. nipponica (Fig. 1). Both strains were obtained from the base of the inflorescence that developed from the underground tuber of a single individual; other plant parts failed to yield fungal material. The ITS and 28S rDNA sequences of the two strains were 100% identical to each other and matched uAX type I (Nomura et al. 2013) with 100% similarity.

The aligned sequence lengths, as well as the numbers of OTUs, phylogenetically informative sites, variable sites, and non-variable sites in the four-locus combined and individual datasets (28S, 5.8S, rpb2, and tef1-α) used for phylogenetic analyses, are shown in Suppl. material 9: table S4.

In the four-locus phylogeny, Sebacinales branched after Tremellomycetes and prior to Dacrymycetales and Unilacrymales in the ML analysis (ML aLRT = 98%/ultrafast bootstrap value = 99%), whereas in the Bayesian inference (BI) tree, Sebacinales was placed outside Dacrymycetales and Unilacrymales with maximum support (Bayesian posterior probabilities = 100) (Fig. 2). The monophyly of the four types of uAX was strongly supported by ML aLRT, ultrafast bootstrap values, and Bayesian posterior probabilities (93/100/100) (Fig. 2). Individual phylogenetic analyses of each of the three regions (28S, tef1-α, and rpb2) exhibited some discrepancies in branching order among Basidiomycota orders, especially in the tef1-α dataset (Suppl. materials 1–3: figs S1–S3). Nevertheless, the monophyly of the four uAX lineages was strongly supported (28S: 100/100/100; rpb2: 100/100/96; and tef1-α: 97/99/100). Furthermore, whereas ITS sequence similarity among the four types of uAX was 89–96%, similarity between each uAX type and the eight Tremellodendropsidales sequences ranged from 63% to 76%. In the combined analysis of the four loci and the 28S rDNA dataset, which had the densest taxon sampling, the four uAX types exhibited a sister relationship with the Tremellodendropsidales clade with strong support (four loci: 90/96/98; 28S rDNA: 88/97/92) (Fig. 2; Suppl. material 1: fig. S1). Together, these two clades formed a sister relationship with Trechisporales and Sistotremastrales in the four-locus phylogeny (Fig. 2; Suppl. material 1: fig. S1).

Phylogenetic relationships inferred from the BLAST searches using ITS sequences from the four uAX types as references are shown in Fig. 3, and collection localities of the retrieved sequences are shown in Suppl. material 10: table S5. In the ITS phylogeny, four clades corresponding to each of the four uAX types were identified (Fig. 3A). The clade containing uAX type I mainly consisted of sequences derived from forest soils, as well as those isolated from Zea mays L. and Allanblackia stuhlmannii (Engl.) Engl. root samples and leaf litter (Fig. 3A; Suppl. material 10: table S5). This clade primarily comprised sequences from temperate to tropical biomes (Fig. 3B). In contrast, ITS sequences from uAX type II exhibited >99% similarity with a fungal isolate obtained from fine, live roots of Cerasus sargentii (Rehder) H.Ohba collected in Japan. This clade had the fewest OTUs but included sequences derived from forest soils collected in Estonia and South Africa (Fig. 3). The clade containing uAX type III comprised sequences from forest soils collected across Europe, Africa, and Central America, although the uAX type III sequences were sister to those derived from mycorrhizae of the Japanese endemic chlorophyllous orchid Neottia makinoana (Ohwi) Szlach. Finally, the clade containing uAX type IV included sequences collected globally from diverse substrates spanning forest and open woodland soils and leaf litter, as well as Cistus and Quercus ectomycorrhizae and orchid mycorrhizae.

Longitudinal and transverse serial sections revealed that MAFF425704 had a dolipore septum associated with multi-pored (perforate) septal pore caps (SPCs) (Fig. 4), the margins of which were bent outwards (black arrows in Fig. 4B, F). Perforations of the SPCs in MAFF425704 were approximately 300 nm in diameter. The SPCs ranged from 75 to 90 nm in thickness and were constructed of three layers: two outer layers consisting of electron-dense lines (white arrows in Fig. 4C, D) and an inner layer that formed a less electron-dense band.

Both strains (MAFF425704 and MAFF425705) cultured on PDA developed well at 10–30 °C, with maximum growth rates observed at 25 °C. Growth terminated at 5 °C and 35 °C (Suppl. material 4: fig. S4). Despite attempts to induce fruitbody production in MAFF425704, these structures failed to develop during the experiments. Nevertheless, uAX formed more than 50 sclerotia-like structures in a Petri dish during incubation (Suppl. material 5: fig. S5). These sclerotia doubled in size, turned brown, and produced flower-like druse and prismatic structures on the surface.

The ITS sequence retrieval and environmental phylogenetic analyses indicate that uAX fungi have a ubiquitous, worldwide distribution across tropical, temperate, and cold biomes (Fig. 3). Rather than being restricted to specific mycorrhizal hosts, uAX lineages associate with diverse substrates and plant roots, including non-mycorrhizal or arbuscular mycorrhizal hosts (Zea mays and Cerasus sargentii), ectomycorrhizae (Quercus and Cistus), and orchid species (Neottia nipponica and N. makinoana). These findings suggest that uAX fungi possess versatile ecological roles—likely acting as saprotrophs, endophytes, or opportunists—rather than obligate ectomycorrhizal symbionts. The physiological temperature optimum of 25 °C recorded for strain MAFF425704 aligns well with the thermal environment of its host N. nipponica (Gale et al. 2021) and corresponds with the predominantly temperate-to-tropical distribution of uAX type I sequences.

In the culture experiments, uAX formed brown sclerotia-like structures covered with flower-like druse and prismatic crystals (Suppl. material 5: fig. S5). Similar structures have been observed in the wood-decaying Trichaptum abietinum (Dicks.) Ryvarden in Hymenochaetales (Connolly et al. 1996), the ascomycetous pathogenic fungus Sclerotinia sclerotiorum (Lib.) de Bary (Uloth et al. 2015), and Rhizoctonia solani J.G.Kühn in Cantharellales growing on deadwood and living stems (Yang et al. 1993). Given that these structures in other fungi are known to be formed of calcium oxalate (Gadd et al. 2014), it is highly likely that uAX produces similar substances. However, it remains unclear what biological or ecological advantages this confers for the growth, development, and survival of uAX.

Since all molecular and morphological evidence presented here demonstrates that uAX is not affiliated with any known order of Basidiomycota, a new fungal order is proposed as follows. The type species, for which cultured strains were successfully obtained, is further described.