Introduction
The digestive tract of ruminants contains a microbiome that enables conversion of plant material to nutrients for its host. This digestion of feed is critical to ruminant health, and to their performance in farming. Driven by this agricultural importance, major advances have been made in understanding the diversity and function of bacterial and archaeal microbiome members (Seshadri et al.
2018, Solden et al. 2018, Mizrahi et al. 2021), but eukaryote members, anaerobic gut fungi (AGF, phylum Neocallimastigomycota), and protozoa, are less well studied, despite their importance in digestion.
AGF activity increases feed digestion, weight gain, and milk production in vivo in ruminants, as reviewed (Gruninger et al. 2014), and increases digestibility of feed by 13%–53% in rumen fluid in vitro cultures (Lee et al. 2000, Ma et al.
2020). The effective AGF digestive activity is thought to be enabled by a broad arsenal of digestive carbohydrate active enzymes as well as penetrative fungal structures that open up plant material for further digestion (Akin and Borneman 1990, Solomon et al. 2016, Haitjema et al.
2017).
Genomics-based approaches have revealed a wealth of AGF diversity, but isolation, cultivation, and functional characterization of AGF is falling behind. Amplicon sequencing using the ITS region or 28S rRNA (LSU) gene markers uncovered the existence of a broad variety of AGF genera in ruminants and hindgut fermenters (Liggenstoffer et al. 2010, Nicholson et al.
2010, Kittelmann et al. 2012, Edwards et al. 2017, Meili et al.
2023). More recently, AGF were also found to be present more broadly in herbivores, including ostrich (Vinzelj et al. 2025) and tortoises (Pratt et al.
2024). Obtaining cultured isolates of these fungi allows for direct functional studies of properties such as metabolism or lignocellulose degradation capacity, and exploration of their capabilities for biotechnology applications. Isolation efforts over the last decade have resulted in the cultivation of more than a dozen new fungal genera from the digestive tracts of domestic and wild herbivores (Dagar et al.
2015, Hanafy, Johnson et al. 2020, Hanafy, Lanjekar et al. 2020, Pratt et al.
2024, Vinzelj et al. 2025), as reviewed by (Hanafy et al. 2022).
However, there are numerous uncultivated anaerobic fungal lineages in both wild and domesticated herbivores (Meili et al. 2024), and for many AGF genera, currently only one species has so far been brought into culture (Hanafy et al. 2022).
Evidently, a substantial diversity of AGF exists in nature that has so far not been cultured. Furthermore, considerable challenges exist in long-term preservation and transport of AGF isolates (Nagpal et al. 2012, Elshahed et al.
2022, Vinzelj et al. 2022). Together, this results in a limited availability of a diversity of AGF strains, hampering functional studies of AGF.
With increasing interest in the AGF for their biotechnological and agricultural applications, and engagement of a growing global community of researchers with this field, it is of increasing importance to report on, and ensure the availability, of broadly accessible strategies to obtain isolates of AGF into culture and assess their identity. In bringing AGF into culture, the diversity of fungi isolated from a sample is thought to be affected by interplay between the diversity and evenness in the sample’s microbiome as well as the employed enrichment and isolation conditions. AGF are found along the ruminant digestive tract (Davies et al.
1993). While AGF communities in faecal samples are less diverse and distinct in composition than in rumen-derived samples, as summarized in (Meili et al. 2024), faecal samples are far more easily obtained for a wide range of animals.
AGF enrichment from stool samples, followed by isolation of fungal strains, is therefore commonly practised in the field and has resulted in recovery of a range of novel genera. While some herbivores harbour a very diverse AGF community in such samples, other host AGF communities seem to be dominated by one or few species (Meili et al. 2023).
A high relative abundance of an AGF species in such samples, and a low evenness of the AGF community, has been demonstrated to strongly correlate to successful isolation of the AGF (Hanafy, Johnson et al. 2020), as also suggested by Stabel et al (2020). Also, the composition of culture media has been reported to affect which AGF are enriched, e.g. the carbon source used affects diversity (Griffith et al.
2009) and the inclusion of rumen fluid selectively enriches for Neocallimastix and Piromyces genera (Joshi et al. 2022).
Selection or characterization of any obtained AGF isolates may include microscopy or functional testing, but unambiguous identification requires DNA-based analysis as key morphological properties such as number of zoospore flagella (mono or poly-flagellate), thallus development (mono or polycentric), or rhizoid morphology (bulbous or filamentous) are commonly shared by taxa, and other distinctive morphological characteristics are subtle and be easy to miss in initial non-axenic isolate cultures (Hanafy et al. 2022). DNA-based methods such as Sanger sequencing of ribosomal ITS and 28S LSU regions or amplicon sequencing are commonly employed (e.g.
Wang et al. 2017, Hanafy, Lanjekar et al. 2020, Joshi et al.
2022, Young et al. 2022), but these may be costly in terms of time or reagents. Historically, DNA-based assessments of the ITS or 28S LSU region via PCR-denaturing gradient gel electrophoresis (PCR-DGGE) or -RFLP has been used to delineate similar groups of isolates, as reviewed by Edwards et al.
(2017), methods that may be fast and economical but were not able to confirm isolate identity. However, their use for discrimination between Orpinomyces species (Dagar et al. 2011) and between up to six distinct genera has been demonstrated (Fliegerová et al.
2006, Dagar et al. 2014; Hausner et al. 2000), suggesting potential for developing this technique via integration with genetic information of the broad range of AGF genera that has become available over the last decade.
In this study, we aimed to isolate a set of AGF species representing different genera and test a combination of identification methods to facilitate fast and economical isolate identification. We hypothesized that integrating different functional and molecular characterization methods (Fig. 1) would benefit assessment of the diversity of culturable AGF isolates obtained from faecal samples.
Zoo-housed herbivores giraffe, nyala, and zebra were selected as hosts as these animals represent different gut systems, with nyala and giraffe being ruminants and zebra a hindgut fermenter. The zebra AGF community has been reported to be rich in Khoyollomyces and the giraffe’s in Feramyces, with faecal samples of these animals therefore likely to harbour distinct AGF (Hooker et al. 2023, Meili et al.
2023). The nyala mycobiome has to our knowledge not been investigated so far. From these samples, we successfully obtained AGF isolates representing Neocallimastix, Feramyces, Khoyollomyces, and Piromyces/NY08 genera.
Isolation success was critically affected by vitamin and antibiotic supplementation of the cultivation media. Based on genetic information of the D1/D2 LSU region of 21 established AGF genera, we developed and validated a PCR-RFLP method that was instrumental in time and cost-effective identification of fungal isolates.
Materials and methods
Faecal samples
Faecal samples were obtained from nyala antelope (Tragelaphus angasii), giraffe (Giraffa camelopardalis), and Grevy’s zebra (Equus grevyi) housed in Edinburgh Zoo, Scotland, United Kingdom. Samples were collected freshly after deposition, by trained zoo personnel following protocols for entry to animal enclosures approved by Edinburgh Zoo. Faecal samples were transported to the laboratory and stored at −20°C until use.
Fungal enrichment, isolation, and growth
Fungal cultivations and enrichments, unless specified otherwise, were performed in 6 ml medium C (as described by Theodorou et al. (2005) but replacing bactocasitone with tryptone) in 15 ml Hungate tubes containing a CO2 atmosphere and incubated at 39°C for 3–4 days.
For initial AGF enrichments, faecal material was dispersed in 6 ml medium C and subsequently diluted 20-fold into 6 ml medium C, with 1% w/v 0.5 mm wheat straw particles. Cultures were supplemented with antibiotics to repress bacterial growth as follows: 50 µg ml−1 chloramphenicol (for giraffe samples) or 166 units ml−1 penicillin and 166 µg ml−1 streptomycin (for nyala samples). After 4 days, when floating plant material and gas pressure accumulation indicated possible fungal activity, aliquots from the enrichment cultures were used to inoculate roll tubes consisting of 2 ml medium C with 5 g l−1 glucose and 15 g l−1 agar, supplemented with the corresponding antibiotics, in 15 ml Hungate tubes.
After incubation for 3 days at 39°C, formation of fungal colonies was confirmed via microscopy. Then, 20 isolated fungal colonies were picked under anaerobic atmosphere (80% N, 10% H2, 10% CO2) in a Whitley A25 workstation (Don Whitley Scientific) and transferred to tubes with 3 ml medium C with 1% w/v wheat straw and 50 µg ml−1 chloramphenicol. Obtained cultures were maintained by transfer of 0.3 ml to a new tube of culture medium every 3–4 days.
Additional enrichments for nyala samples were performed after 8 months of sample storage at −20°C, as described above, except replacing chloramphenicol with 166 units ml−1 penicillin and 166 µg ml−1 streptomycin during culture maintenance. For additional enrichments from zebra samples, a panel of antibiotics and vitamins was trialled, including combinations of 50 µg ml−1 chloramphenicol, or 166 units ml−1 penicillin and 166 µg ml−1 streptomycin, 50 µg ml−1 kanamycin, as well as the chloramphenicol or ampicillin/streptomycin together with 100 µl vitamin solution 2. Vitamin solution 2, described in (Teunissen et al.
1991), consisted of 5 mg l−1 thiamin.HCl, 5 mg l−1 riboflavin, 5 mg l−1 calcium D-pantothenate, 5 mg l−1 nicotinic acid, 2 mg l−1 folic acid, 1 mg l−1 cyanocobalamin, 1 mg l−1 biotin, 10 mg l−1 pyridoxin.HCl, and 5 mg l−1 p-aminobenzoic acid.
Growth profiles for carbon source use
The ability of fungal isolates to use different sugars as carbon source was screened in medium C containing either 5 g l−1 glucose, fructose, or galacturonic acid as the sole carbon source. Sugars were added to the autoclaved media as filter-sterilized stock solution and 50 µg ml−1 chloramphenicol was supplied to repress bacterial growth. Fungal isolates were initially cultured on medium C with wheat straw as carbon source for 3 days to achieve an exponentially growing culture, after which a 10% v/v inoculum was transferred to initiate cultures for profiling of carbon source use.
Growth was assessed after 4 days of cultivation, via visual assessment of fungal biomass formation, and measurements of fermentation gas pressure. For selected isolates, additional rounds of purification were performed to achieve axenic cultures, after which C-source use profiles were confirmed via cultivation in the media described above.
DNA extraction
Fungal isolates were grown in 6 or 50 ml cultures with glucose as carbon source for 3–4 days, after which biomass was harvested via centrifugation (7 min at 3893 g), frozen and lyophilized. DNA was extracted using a CTAB extraction of pulverized biomass followed by chloroform extraction and isopropanol precipitation, as described (Griffith et al. 2009) but replacing the hot ethanol wash step with ambient ethanol.
Obtained DNA was used to perform PCR-RFLP or ITS2-based amplicon sequencing.
For DNA from stool samples sent for AGF-specific amplicon sequencing with LSU-EnvS primers, DNA was extracted with the Qiagen DNeasy Plant Pro Kit with the following modification: cells in stool samples (∼100 mg wet weight) in CD1 lysis buffer were disrupted for 3 × 1 min in a Qiagen Tissuelyser LT at 50 Hz, followed by a 10 min incubation at 65°C, before continuing purification as described by the extraction kit manufacturer.
PCR-RFLP
PCR-RFLP using the ITS region was performed generally as described by (Fliegerová et al. 2006). In short, PCR was performed with primers ITS1 and ITS4 using Promega GoTaq Mastermix, whereafter product formation was confirmed via agarose gel electrophoresis.
PCR products were purified from 20 µl PCR mixture using the Wizard® SV Gel and PCR Clean-Up System (Promega) as directed by the manufacturers. PCR products were eluted in water and ∼0.5 µg DNA was digested with 5 U of DraI or HinfI at 37°C for 1 h, after which the resulting DNA fragments were visualized on a 2.5% agarose gel using Gelred nucleic acid stain and Thermo Scientific GeneRuler 100 bp ladder. PCR-RFLP using the LSU region was performed similarly, but using the primers NL1 and NL4, and digestion with Hpy188I or AluI, followed by visualization on a 3% agarose gel.
To generate an ITS PCR-RFLP database of predicted PCR-RFLP fragments for each of the so far identified species of AGF, the AF_Full_Regiondatabase version 2.0 was obtained from the Anaerobic Fungi Network (www.anaerobicfungi.org). Initial sequence alignments were performed in Clustal Omega (Sievers et al. 2011).
Using R 4.4.2 with the R-packages DECIPHER version 3.2.0 (Wright 2024) and Biostrings version 2.74.1 (Pagès et al. 2025), the sequences were trimmed to the ITS1 and ITS4 primer borders and complete sequences were separated from incomplete sequences (lacking the ∼50 bp at the ITS1 border). Following in silico digestion with DraI and HinfI restriction enzymes, fragment sizes were tabulated.
For selected sequences, results were visualized on an in silico 2.5% agarose gel using SnapGene 8.2.1. (Fig. 3).
For Fig. 3a, DraI digestion, the following sequences were used: JQ782543, JQ782547, MT085722, Aoud18_14 398, MT085694, MW907587, JQ782554, MZ044643, MT085707, MW899532, MT085733, JQ782551, JQ782553, MT085711, MT085666, MT0856676, MT085680, MT085712, MT085689, Horse_10 178, AxDoe_98 435, ZebrA149764, MT085698, MT085728, MT085727, MT085708, MT085709, MT085738, OQ382931, and OQ382915. For Fig.
3b sequences were as for DraI digestion except, to display the most representative profiles, JQ782547, MZ044643, and MT085709 were omitted, MT085707 was replaced by MT085705, MT085733 by MG605690, JQ782551 and JQ782553 by MT085734, and ZebrA146613, MG605690 were added.
For development of an LSU-based PCR-RFLP, the AF_LSU_database version 2.0 was obtained from the Anaerobic Fungi Network (www.anaerobicfungi.org) and sequences were trimmed to NL1 and NL4 primer sequence borders. Sequences lacking the NL4 primer sequence were separated from complete sequences and in silico digestion with a range of restriction enzymes was tested for ability to generate fragments that allowed discrimination between AGF genera and/or species. For selected sequences, results were tabulated and visualized on an in silico 3% agarose gel using SnapGene 8.2.1.
The following sequences are used for visualization in Fig. 5: for AluI: JQ782543, MT085722, MW019480, MT085694, MW907586, JQ782554, MT085707, MW899532, MG605686, MT085733, MT085711, MT085668, MT085675, MT085679, Piromyces_spUH3_1_LSU_scaffold317_9682_10 448, KP205570, Zebra_149 764, MT085698, MT085701, MT085728, MT085709, MT085738, OP253962, and OP253957. For Hpy188I, to display the most representative profiles, MW907587 replaced MW907586, MT085666 replaced MT085668 and Horse_101 214_Piromyces was added.
Predicted LSU PCR-RFLP profiles were subsequently experimentally validated as described above using DNA obtained from a panel of AGF strains available in the laboratory: N. frontalis CoB3 (Shen et al. 2026), N. cameroonii N9 and N13, Caecomyces communis SHB (Shen et al. 2026), Capellomyces forminis Cap2A (Hanafy et al.
2023), Piromyces edwardsiae isolate SHC (Shen et al. 2026), and Feramyces sp. isolates G3 and G5.
Amplicon sequencing
DNA of isolates and faecal samples was sent to Novogene for amplicon sequencing of the ITS2 rRNA gene with primers ITS3-2024F (5’-GCATCGATGAAGAACGCAGC-‘3) and ITS4-2409R (5’- TCCTCCGCTTATTGATATGC-‘3), or for 28S LSU rRNA gene region with primers AGF-LSU-EnvS-F (5’-GCGTTTRRCACCASTGTTGTT-‘3) and AGF-LSU-EnvS-R (5’-GTCAACATCCTAAGYGTAGGTA -‘3) (Young et al. 2022) that produce a ∼370 bp fragment, using amplification conditions as described in (Meili et al. 2023).
Negative controls did not yield any amplification products and were omitted from further analysis. Illumina 250 bp paired end sequencing was used via NovaSeq 6000 to obtain a minimum of 30k raw reads per sample. Sequences were analysed using Mothur v.1.48.5 (Schloss et al.
2009) following a protocol based on (Meili et al. 2023) and (Kozich et al. 2013).
Demultiplexed reads were assembled to contigs while removing primer and barcode sequences. Quality trimming was performed to include only sequences between 300 and 320 bp in length, with homopolymer stretches of ≤8 bases, that did not contain ambiguous bases and aligned to Neocallimastigomycota. Preclustering of the 19.8–64k reads per sample was performed allowing 3 nucleotide differences and chimeras were removed via the chinera.vsearch command using dereplication.
Sequences were classified using the aforementioned AF_LSU_database (v2), and OTUs were generated using a cutoff of 0.03 and classified using the same database. Metadata, taxonomy and OTU tables were imported into the R-package phyloseq (McMurdie and Holmes 2013) for analysis and data visualization. Alpha-diversity (Observed, Shannon, and Simpson Indices) was calculated from raw counts and tested for significance using a Kruskal–Wallis chi-squared test followed by Dunn’s Pairwise Comparison with Benjamin–Hochberg correction for multiple testing.
Taxa were pruned to include only those with mean >0.5 reads, resulting in 169 OTUs across 9 samples. Data were converted to relative abundances, beta-diversity calculated via weighted UniFrac ordination and PCoA analysis, and tested for significance using PERMANOVA via adonis2 in the vegan package (Oksanen et al. 2026).
Results
Faecal samples of zoo-housed zebra, nyala and giraffe were used for amplicon-based profiling of resident AGF communities (Fig. 1a), and also as source for isolation of AGF strains (Fig 1b, c), which we characterized using a combination of three methods (Fig. 1d–f).
Zebra and nyala, but not giraffe harbour simple AGF communities
To assess AGF diversity in the host mycobiome, we performed amplicon sequencing using AGF specific primers AGF-LSU-EnvS (Young et al. 2022). After filtering 169 OTUs were obtained over the different samples.
Taxonomy (Fig. 2a, b) indicated that the AGF community from zebra consisted almost exclusively of (99.8%–99.9% of reads) Khoyollomyces. The AGF community from nyala also had limited α-diversity and mainly consisted of Neocallimastix cameroonii and an uncultured proposed genus NY08 that is closely associated (Meili et al.
2023) with Piromyces. The giraffe faeces harboured an AGF community with higher α-diversity, in terms of Observed, Shannon as well as Simpson indexes (Fig. 2c), with 3 distinct dominant genera (Fig.
2a). Besides Neocallimastix frontalis, and Feramyces austinii, the giraffe samples were also abundant in Piromyces-associated NY08 (100% identity of respective OTU sequences). In addition, up to 4.8% of Neocallimastigomycota reads in giraffe sequences could not be classified to genus level.
Notably, while the proposed genus NY08 previously was only identified to have <5% abundance over a range of animal samples (Meili et al. 2023), it represents 6%–44% of reads in the giraffe samples and 46%–94% of reads in the nyala samples, suggesting that these samples may be a good source to isolate this so far uncultured fungus from. Based on the PCoA analysis the three animals harboured significantly different communities (P = 0.007), with PC1 and PC2 explaining 32.9% and 24.7% of the variation observed, respectively.
Distinct clustering was observed by host animal (Fig. 2d); however, the samples did not seem to cluster by gut type.
Obtaining fungal isolates from faecal samples
Stool samples were used to inoculate AGF enrichment cultures in complex medium C with wheat straw as carbon source, a complex fibrous plant material that can be degraded by broad range of AGF (Fig. 1b). The obtained enrichment cultures were used to inoculate solid medium in roll tubes (Fig.
1c). Roll tubes inoculated from enrichments originating from zebra samples did not initially yield any colonies, despite multiple attempts to repeat enrichments and roll tube cultivation, employing different combinations of antibiotics.
For roll tubes inoculated from enrichments from nyala and giraffe samples, 20 colonies were subsequently picked per enrichment, of which in total 80% were successfully taken through all (27 isolates) or a subset (5 isolates) of the characterization steps, with the remainder lost from cultivation. Isolates were characterized, via a combination of profiling of carbon source use, morphological assessment and molecular biology (PCR-RFLP and sequencing) (Fig. 1d–f), to assess diversity of cultured isolates and to prioritize isolates for identification and further purification to the axenic cultures required for full characterization and functional studies.
Efficacy of the three characterization techniques, in terms of reliability and resource use, were also assessed.
Profiling of carbon source use of isolates from nyala
Growth on plant-derived simple sugars fructose, glucose and galacturonic acid was assessed, whereby glucose and fructose are expected to be metabolized by all or most AGF genera and species (Trinci et al. 1994). The plant cell wall polysaccharide pectin, or its monomer galacturonic acid, is not commonly metabolized by AGF, with exception of Feramyces austinii (Hanafy et al.
2018), and some Neocallimastix isolates (Kopečný and Hodrová 1995). We recorded fungal growth via fermentation gas levels, a standard in the field used as a quantitative proxy for growth of AGF (Theodorou et al. 1995, Wilken et al.
2020) as well as visual observations of fungal biomass. However, considerable discrepancies between both (Fig. S1) with false positives, possibly due to bacterial contamination of initial fungal isolates, as well as some false negatives, indicated that gas pressure measurements alone are ineffective when assessing growth of non-axenic fungal isolates.
Visual observation of fungal biomass (Fig. 3a) demonstrated that after 4 days all isolates used glucose and fructose to support their growth, and most used galacturonic acid, except isolates N4, N5, N6, and N13. However, once purified to axenic cultures, representative isolates N2, N6, N9, and N13, grew well on glucose and fructose over 7 days (Fig.
S2a) generating 10.7–18.1 mg dry weight of biomass per culture (n = 3), whereas on galacturonic acid the accumulated gas pressure was negligible and biomass only 2.0–5.4 mg per culture. Overall, this suggests that growth of all these isolates on galacturonic acid was slow or limited, and initial observed differences during growth on galacturonic acid were likely reflections of variations of growth rather than differences in innate capability.
Morphological assessment of isolates from nyala
Microscopy identified isolates in liquid culture all had filamentous rhizoids and polyflagellated zoospores (Fig. 3b). The macromorphology of mycelium was similar in all cultures, whereby in initial growth stages biomass was visible in liquid media as dense mycelial balls that later merged in a sheer network of mycelium (Fig.
3c). On solid medium, small (2–3 mm) beige colonies formed that reached their maximum diameter in 2 days, with a dense centre that became dark brown over time. Following the key for AGF genera identification (Hanafy et al.
2022), morphology indicates that Ghazallomyces, Aestipascuomyces, Neocallimastix may have been obtained.
DNA-based profiling and identification of isolates from nyala as N. cameroonii
As morphological assessment was as expected insufficient for unambiguous identification, we combined this with a molecular approach. We initially aimed to use ITS-based PCR-RFLP to assess genetic diversity between isolates, as PCR of the ITS region followed by digestion with DraI and HinfI enzymes was successful in experimentally demonstrating discrimination between at least 2 AGF genera in RFLP (Fliegerová et al. 2006, Griffith et al.
2009). To assess if this approach was able to discriminate between the larger diversity of currently established genera and to enhance information derived from PCR-RFLP profiles, we integrated the PCR-RFLP with sequence information for ITS rRNA gene regions. Using the sequence database for the AGF genera described so far [AF_Full_Region database version 2.0 available from the anaerobic Fungi Network (www.anaerobicfungi.org)], sequence information corresponding to the regions amplified by ITS1 and ITS4 primers was extracted.
While the ITS4 primer motif was identified in all database sequences, only 63 out of 197 sequences contained the ITS1 primer motif with remaining sequences having a 65–73 bp truncation at the 5’- end that created an uncertainty in the length of the first predicted RFLP fragment. In silico digestion with DraI or HinfI generated RFLP profile predictions for each cultured genus (Table S1), of which representative profiles were visualized (Fig. S3) to aid correlation with experimentally obtained RFLP profiles.
Results of the in silico analysis indicate that most genera with cultured representatives can be distinguished using the predicted RFLP patterns. However, for some genera, multiple RLFP patterns are predicted that likely reflect the previously reported diversity of ITS sequences within genera (Edwards et al. 2017).
Experimentally, amplification of the ITS region of our nyala-derived isolates with primers ITS1 and ITS4 resulted in a single PCR product. For all isolates, DraI digestion generated 3 main bands (259, 148, and 103 bp) and a faint additional band around 195 bp (Fig. 3d) that, based on intensity and cumulative size of all fragments, can only have arisen from partial digestion.
HinfI digestion resulted in profile with 2 fragments, 392 and 348 bp in size (Fig. 3e). Repeat measurements of bands of known size indicate an error margin of 20–30 bp for absolute fragments sizes should be taken into consideration.
When comparing reference patterns to the experimentally obtained profiles for nyala isolates (Fig. 2e–f), this indicated they correspond to Neocallimastix cameroonii. Specifically, experimental HinfI digestion patterns matched predictions for N. cameroonii, Caecomyces, Capellomyces, or Liebetanzomyces, while the DraI digestion results, taken uncertainty in the size of the first fragment into account, could match N. cameroonii, Ghazellomyces, and Cyllamyces.
PCR-RFLP identification as N. camerooni is in line with the morphological characteristics identified for these isolates. Identification was further confirmed by ITS2 amplicon sequencing; sequences making up >99% of reads for isolates N2, N6, N9, and N13 had 96.1%–99.7% identity to N. cameroonii clone 1 and 6 (Genbank MW175298, MW175302.1). Taken together, all fungal isolates obtained here from nyala consisted of N. cameroonii.
Profiling of cultures from giraffe confirm isolation of Feramyces austinii
As our isolation efforts from the nyala stool sample yielded only one AGF species, despite initially picking 20 colonies, we repeated the analysis for a faecal sample from giraffe. Results of visual observation of fungal biomass in cultures on glucose, fructose, and galacturonic acid (Fig. 4a), suggested that most isolates used all three carbon sources to support growth, whereas isolates G5 and G8 had not grown on galacturonic acid after 4 days.
However, as observed for nyala isolates, verification with selected axenic isolates G2, G3, G5 over 7 days indicated that these strains eventually grew on all substrates (Fig. S2b), although with less biomass on galacturonic acid (3.4–5.9 mg dry weight per culture) compared to glucose and fructose cultures (12.7–17.9 mg per culture). Morphology of observed mycelium was similar for all isolates, with monocentric thalli, filamentous rhizoids, and polyflagellated zoospores (Fig.
4b). In liquid culture, thick, dispersed growth on glucose and fructose was observed (Fig. 4c), with more sheer growth on galacturonic acid.
On solid medium light beige colonies with a dense, irregular shaped centre were formed that grew to over 5 mm diameter. These morphological characteristics correspond to those described (Hanafy et al. 2022) for genera Ghazallomyces, Aestipascuomyces, and Feramyces.
PCR-RFLP resulted in a profile conserved among all isolates from giraffe but distinct from the nyala isolates. DraI digestion generated 3 distinguishable bands (Fig. 4d), determined to be 239, 191, and 166 bp, as consistent only with patterns predicted for Feramyces.
HinfI digestion resulted in two visible fragments, measured as 390 and 363 bp (Fig. 4e), consistent with fragments predicted (Table S1) for Feramyces, N. cameroonii, Ghazallomyces, Anaeromyces, or Capellomyces. Together the PCR-RFLP indicated that all isolates likely belong to the genus Feramyces.
This is consistent with our morphological observations and with the reported characteristics of F. austinii (Hanafy et al. 2018), the only known species in this genus. ITS2 amplicon sequencing resulted in >99% of reads from representative isolates G2, G3, and G5 having 98.4%–100% identity to F. austinii strain DF1 and type strain F3a, thus confirming this identification.
Improved PCR-RFLP strategy for identification for AGF using LSU sequences
ITS sequences have been reported to be variable and diverse in length between AGF genera and even demonstrate high diversity within species (Edwards et al. 2017). The ITS-based PCR-RFLP strategy described above has been useful for discrimination between fungal isolates but, as highlighted by Fig.
S3 and Table S1, multiple common RFLP patterns were obtained for 6 out of 21 analysed genera and species for each of the restriction enzymes, with 4 patterns for Piromyces. This complicated tentative identification of isolates based on PCR-RFLP. As drawbacks of the ITS-based phylogenetic marker have resulted in a shift in the community towards using LSU-based identification (Elshahed et al.
2022, Young et al. 2022), and LSU-based PCR-RFLP enabled discrimination between Orpinomyces joyonii and O. intercalaris (Dagar et al. 2011), and between Orpinomyces and Anaeromyces isolates (Fliegerová et al.
2006), we explored if LSU-based PCR-RFLP could be used to aid fast identification of a wide range of AGF isolates.
We performed in silico digestion of LSU sequences with a range of restriction endonucleases, assessing distinct fragment length polymorphisms between genus and species representatives. Initially a minimal set of LSU sequences representing each genus was assembled and trimmed to contain only the region amplified by NL1 and NL4 primers. Recognition sites were assessed for a selection of restriction enzymes available from commercial providers, including enzymes recognizing abundant motives, as well as those cleaving once or twice in representative LSU sequences, omitting enzymes dependent on methylation.
In silico digestion profiles (Table S2), were ranked on number of different fragment patterns produced, taking into account possible poor visibility of small (<60 bp) DNA fragments on gel. As no single enzyme digestion could discriminate between al genera, pairs of highly discriminatory enzymes were assessed for production of complementary fragment patterns that together discriminate between fungal genera and species. Conservation of selected enzyme sites was then confirmed across sequences of the complete LSU sequence database.
Digestion with the enzymes AluI (Fig. 5a) or Hpy188I (Fig. 5b) resulted in patterns of fragment lengths (Table S3) which, when combined, would enable differentiation between all genera.
A single conserved or highly dominant pattern was found for all genera besides Piromyces. At the species level, N. frontalis and N. cameroonii can be easily distinguished on distinct RFLP, as can O. joyonii and O. intercalaris, while the RFLP for each of the two species represented in genera Anaeromyces and Capellomyces were identical or highly similar. The number of common patterns obtained for Piromyces was reduced compared to the ITS-based strategy but could not directly be linked to known species.
This may reflect the phylogenetic complexity of the genus (Hanafy et al. 2023).
We experimentally verified the LSU-based PCR-RFLP strategy, testing the profiling of a range of AGF species: the Feramyces G3 and G5 and N. cameroonii N9 and N13 isolates described in this study, as well as N. frontalis CoB3 (Shen et al. 2026), Caecomyces communis SHB (Shen et al. 2026), Capellomyces forminis Cap2A (Hanafy et al.
2023), and Piromyces edwardsiae isolate SHC (Shen et al. 2026). The DNA fragment profiles generated after digestion with AluI (Fig.
5c) and Hpy188I (Fig. 5d) enabled straightforward discrimination between these species, and, when compared to predicted fragment sizes (Table S3), led to unambiguous identification of the species based on PCR-RFLP profile alone in almost all cases. The identification of the Cap2A profile solely from fragment sizes determined from the RFLP gel would be uncertain as, due to the 20–30 bp error margin, possible matches include both Piromyces or Capellomyces.
However, as visible on the gel (Fig 5a, b), including a reference sample for each of the strains allows clear distinction between the two genera. Overall, this highlights that PCR-RFLP based on NL1 and NL4 amplification and digest with AluI and Hpy188I allows excellent discrimination of AGF genera cultured so far.
Supplementing enrichment media with vitamins enabled Khoyollomyces growth
As Khoyollomyces dominated AGF in zebra faecal samples (Fig. 2), but initial AGF enrichment cultures were unsuccessful, we reviewed reported growth conditions for Khoyollomyces (Hanafy, Lanjekar et al. 2020, Jones et al.
2024, Schulz et al. 2025). A major difference with the medium used here was inclusion of vitamin and trace element solutions.
We therefore hypothesized that the enrichment medium was lacking vitamins or trace elements required for Khoyollomyces growth. Indeed, addition of vitamin solution, but not trace element solution, enabled successful enrichments. The subsequently obtained isolate Z3 produced small colonies on solid medium and grain-like biomass morphology in liquid cultures of medium C (Fig.
6a). Microscopy revealed filamentous mycelia with thalli with often multiple sporangia (Fig. 6b), and monoflagellated zoospores (Fig.
6c), all consistent with characteristics of Khoyollomyces ramosus. The isolate’s PCR-RFLP pattern (Fig 6d) was consistent with that predicted for Khoyollomyces (Table S3, Fig. 5a, b), and amplicon sequencing using AGF-envS primers confirmed this identification, with 100% of reads having 99.7%–100% identity to Khoyollomyces ramosus sequences.
Notably, vitamin supplementation was required only during establishment of the K. ramosus culture, the isolate has been in stable cultivation in medium C for ∼4 months without additional vitamin supplementation.
Isolation of chloramphenicol-sensitive NY08 representative from nyala
As nyala samples were rich in so far uncultured Piromyces-associated NY08, and considering 25% of nyala-derived isolates were previously lost when adding chloramphenicol, we repeated enrichment and isolation avoiding chloramphenicol. All 5 obtained isolates had identical PCR-RFLP profiles that were consistent with those predicted and observed for Piromyces (Fig. 7a, b), and Sanger sequencing of isolate N2.1 and N3.1 ITS and LSU regions confirmed NY08 representatives were obtained.
Intra-isolate sequence variation was 0–0.38% across 10 LSU sequences and 0% across 3 ITS sequences. LSU sequences were 99.6%–99.9% identical to sequences forming the basis for initially proposing NY08 (e.g. OP253938), indicating these belong to the same species, which we will refer to here as Piromyces sp.
N2.1 while awaiting reclassification based on ongoing phylogenetic analysis of the Piromycetaceae. For both isolates of the species obtained here, small brown colonies formed on solid medium, in liquid medium mature mycelium adhered tightly to culture tubes (Fig. 7c).
The filamentous hyphae were thin and wrinkly (Fig. 7d), and zoospores (8.0 ± 0.9 µm diameter, n = 30) were monoflagellated, consistent with what was previously described (Hanafy et al. 2022) for Piromyces species.
Discussion
Here we described the isolation and basic characterization of AGF representing four different genera relatively abundant in faecal samples of their host, including the so far uncultured Piromyces sp. N2.1 [previously proposed as novel genus NY08 (Meili et al. 2023)].
We demonstrated how an LSU-based PCR-RFLP strategy that we developed and validated here can be employed to quickly and efficiently profile and identify fungal isolates. We highlight how enrichment and isolation conditions, including use of antibiotics, may be critical to isolation success.
The AGF community in the faecal samples of zebra was dominated by Khoyollomyces, which is line with the reported strong association of the genus Khoyollomyces with horses and zebras (Hanafy, Johnson et al. 2020, Meili et al. 2023) and reported isolation of this genus from horse and zebra (Hanafy, Lanjekar et al.
2020, Stabel et al. 2021). Feramyces was abundant in giraffe faecal samples, similar as described (Liggenstoffer et al.
2010, Meili et al. 2023) but isolation of this genus has only been reported from wild Barbary sheep and fallow deer (Hanafy et al. 2018).
In contrast, isolation of Neocallimastix, which here was relatively low in abundance in the giraffe AGF community (Meili et al. 2023), has been reported (isolate GF-Ma3-1, genome available on mycocosm, isolate G341 (Stabel et al. 2021)).
We report here to our knowledge for the first time, the composition of the AGF community of the nyala, a spiral-horned antelope related to the kudu, bushbuck, and eland. Results indicated low AGF diversity with two dominant species, and a high relative abundance of an uncultured fungus. We were able to capitalize on this via successful isolation of both species, which included the NY08 representative Piromyces sp.
N2.1. Notably the phylogeny of Piromyces and the wider Piromycetaceae family is complex (Hanafy et al. 2023) and isolation of NY08 followed by obtaining genomic or transcriptomic data would aid required revision of the family’s phylogeny and clarify the phylogenetic placement of NY08.
In our workflow, we obtained fungal isolates of only one species per enrichment, whereby simultaneously performed replicate enrichments yielded similar results. Others (Hanafy, Lanjekar et al. 2020, Peng et al.
2021, Stabel et al. 2021) have reported similar outcomes. As also suggested by (Stabel et al.
2021), it seems probable that obtained isolates would be the dominant species of fungus in the faecal sample, or well adapted to the enrichment conditions. Indeed (Hanafy, Johnson et al. 2020), who isolated one to three genera per faecal sample in a large isolation effort, reported a positive correlation between frequency of isolation of monocentric taxa (such as Feramyces and Neocallimastix) and abundance in a sample and a negative correlation of frequency of isolation with community evenness.
AGF community profiling in a study including samples from 5 animal species (Joshi et al. 2022) suggested that enrichment cultures are dominated (>90% of relative abundance) by one or two genera even if up to three to four genera are each present at >10% relative abundance in the corresponding faecal sample. Notably, genera dominating enrichments did not always correspond with the most abundant genera in the faecal sample (Joshi et al.
2022), underlining how species that are well adapted or competitive under the enrichment conditions were obtained. Similarly, in our results, successful enrichment and isolation of Khoyollomyces Z3 required addition of vitamin solution, and for cultivation of Piromyces sp. N2.1 the omission of chloramphenicol, routinely used for AGF enrichments, from the cultivation medium was required.
Combined, these findings suggest that enrichment cultures may be often dominated by one or few AGF genera, and isolating 15–20 colonies per enrichment, as performed here, would be more than sufficient to enable isolation of the expected one to three enriched fungal genera. It is likely that stratifying enrichment conditions (carbon source, media composition, and antibiotics employed) or inoculum treatment (storage time, method) may be effective when aiming to obtain distinct genera from a sample.
We tested a workflow for the identification of fungal isolates that integrated the assessment of morphology, carbon source usage as well as molecular profiling of genetic diversity. Observations of morphology (filamentous or bulbous) and zoospore flagellation were informative in assessing diversity of fungal isolates, but as reviewed (Hanafy et al. 2022) were, in isolation non-conclusive in confirming identity.
The profiling of carbon source usage proved resource intensive as well as error prone and did not bring benefit here, it is best reserved for profiling selected axenic fungal isolates. Integration of morphological analysis with molecular tools has enabled identification and description of a range of recent genera (Hanafy et al. 2017, 2022).
However, in this study cloning and sequencing of ITS or LSU regions of all isolates was cost prohibitive and time consuming, prompting us to develop a PCR-RFLP strategy that proved efficient and effective in profiling individual isolates, and provided unambiguous results even in the hands of inexperienced students. With the LSU-based method described here we were able to process ∼20 isolates (6 ml cultures each) in 2 days for ∼£90. Thus, PCR-RFLP enables economical and fast assessment of a large number of fungal cultures, to prioritize isolates for further analysis such as sequencing of the D1-D2 LSU rRNA gene.
In developing the PCR-RFLP method, we capitalized on the availability of rRNA gene sequences for distinct AGF genera and species as available in a database via the AGF network website. Initially, we used in silico analysis to generate a table of fragment profiles that can be expected when ITS-based RFLP is executed as described by (Fliegerová et al. 2006).
However, the resulting multiple dominant RFLP profiles per genus made identification onerous. We aimed to improve the PCR-RFLP method by using the D1–D2 region of the 28S LSU rRNA gene sequence as marker. As AGF have considerable divergence in ITS sequence length and sequence composition, even within one species, the ITS region is not considered a good phylogenetic marker region for AGF (Edwards et al.
2017, Hanafy, Johnson et al. 2020, Hanafy et al. 2023), and alternative approaches using the D1–D2 LSU region have been developed (Edwards et al.
2017, Hanafy, Johnson et al. 2020, Young et al. 2022).
Aligned with this, and encouraged by previous reported PCR-RFLP-based distinction of Orpinomyces joyonii and Orpinomyces intercalaris (Dagar et al. 2011) and further experiment discrimination between 6 AGF genera using this region (Dagar et al. 2014), we explored use of LSU-based PCR-RFLP.
Our in silico prediction combined with experimental validation on axenic cultures demonstrated that profiling with a combination of Hpy118I and AluI provided discriminatory patterns across the board for most reported cultured genera. Further testing on mixed AGF communities from enrichments suggested that this method may be employed to perform initial screening of AGF genera enriched under different conditions. Such an initial screening could be performed within the timeframe of one fungal cultivation cycle (2–4 days) and could complement initial sequencing and selection of faecal samples, which is reported to be highly beneficial in directing isolation efforts for obtaining novel AGF species (Hanafy, Johnson et al.
2020). With the ongoing expansion of the LSU database, more full-length DNA sequences are expected to become available, which will further increase the accuracy of the fragment size predictions.
Supplementary Material