Anaerobic fungi (phylum Neocallimastigomycota) are the most effective fiber degrading microorganisms in the gut of mammalian herbivores. This is due to their combined mechanical and enzymatic activity, and the associated ability to penetrate plant structural barriers (Orpin, 1975; Ho et al., 1988; Lee et al., 1999). These attributes are underpinned during the vegetative and motile stages of their life cycle, with a putative aerotolerant resistant stage more associated with survival outside the mammalian gut (Davies et al., 1993; McGranaghan et al., 1999). During the first 40 years following their recognition as fungi by Orpin in 1975, only six genera of anaerobic fungi were named within phylum Neocallimastigomycota (Anaeromyces, Caecomyces, Cyllamyces, Neocallimastix, Orpinomyces, and Piromyces). Within the last 2 years three new genera of anaerobic fungi have been described: Buwchfawromyces with its type species Buwchfawromyces eastonii (Callaghan et al., 2015; formerly known as SK2 clade, Koetschan et al., 2014), Oontomyces with its type species Oontomyces anksri (Dagar et al., 2015), and Pecoramyces (formerly known as Orpinomyces sp. C1A; Youssef et al., 2013) with its type species Pecoramyces ruminantium (Hanafy et al., 2017).

Anaerobic fungi have been largely described in the past on the basis of their morphological characteristics. However, due to factors such as convergent evolution and highly variable in vitro growth characteristics, morphological features alone are not sufficient to distinguish between certain genera and species. Within the nine characterized genera more than 20 species have been described, however, genetic analysis indicated that despite different generic names, some species are identical. On the other hand, description of new species supported by morphological and genetic analysis has recently enlarged the group of cultivated anaerobic fungi (Ariyawansa et al., 2015; Li G. J., et al., 2016).

Anaerobic fungi have been most extensively studied in ruminants, but in recent years an increasing amount of anaerobic fungal research has focused on their biotechnological exploitation (Prochazka et al., 2012; Gruninger et al., 2014; Haitjema et al., 2014; Gilmore et al., 2015; Peng et al., 2016; Solomon et al., 2016a; Dollhofer et al., 2017). This has been mainly in terms of their potent fiber degrading enzymes, due to challenges with cultivating anaerobic fungi in large scale continuous systems limiting their direct application (Gruninger et al., 2014; Solomon et al., 2016a). For the anaerobic fungal research community, this biotechnological interest has led to substantial investment that has enabled valuable advances in anaerobic fungal knowledge and resources, particularly from genomic and transcriptomic data in recent years.

In this review, we focus on the different types of molecular methods, including ‘omics approaches, that have been used to date in the study of anaerobic fungi, and we highlight the challenges that currently exist—many of which are fundamentally different from those encountered with the more routinely studied rumen bacteria and archaea.

The “Assembling the Fungal Tree of Life” (AFToL) project used a multi-gene approach to decipher, to high resolution, the low level evolutionary phylogenetic relationships between the fungal Kingdom (James et al., 2006). The six genes used were those encoding 18S ribosomal RNA (rRNA), 28S rRNA, 5.8S rRNA, Elongation Factor 1-alpha (EF1α), and two RNA polymerase II subunits (RPB1 and RPB2). However, it is important to recognise the difference between barcoding loci and other loci suitable for phylogenetic inference. Potential barcoding loci, particularly those useful for next generation sequencing (NGS) based diversity studies, are those found on the multicopy rrn (rRNA) operon (Figure 1). These loci have a high copy number (ca. 200) per genome meaning only small amounts of tissue or environmental sample are needed for efficient PCR amplification. Additionally, these loci are not protein coding, therefore having a relatively high mutation rate enabling good phylogenetic resolution (Hibbett et al., 2007). Consequently, for barcoding and environmental sequencing studies loci belonging to the rrn operon are most suitable, thus such research into anaerobic fungi over the last 25 years has focussed on numerous different regions within this operon (Figure 1, Table 1).

Contrasting with this, taxonomic loci tend to be single copy protein coding genes, including e.g., RPB1 and RPB2 (James et al., 2006) and EF-1α (Eckart et al., 2010). These genes code for critically important functional proteins. Hence, mutations across these genes are likely to cause a loss of fitness or death of the organism, and as a result these genes are highly conserved. Interestingly, it has been found that anaerobic fungi have two paralogous copies of EF-1α (Eckart et al., 2010), which is perhaps not that surprising considering the large amount of repetition that has been reported to occur in anaerobic fungal genomes (Haitjema et al., 2017). This phenomenon has also been reported to occur in other basal fungal taxa (James et al., 2006), and limits the value of this gene as a marker. Single copy protein encoding genes enable reliable higher level phylogenetic classification, but are not so useful in differentiating closely related fungi to the species level.

The small-subunit (SSU) rRNA gene is widely used as a barcode marker for bacteria, archaea and protists, and has also been looked at as a barcoding and quantification loci for the anaerobic fungi (Dore and Stahl, 1991; Brookman et al., 2000; Dollhofer et al., 2016). The much shorter 5.8S rRNA (185 bp) has also been used in qPCR based quantification methods for anaerobic fungi (Edwards et al., 2008). However, neither 18S nor 5.8S rRNA loci are variable enough to enable phylogenetic differentiation between all anaerobic fungal genera (Eckart et al., 2010; Dollhofer et al., 2016). Therefore, the internal transcribed spacer 1 (ITS1) region has instead been used most extensively for differentiating genera and species of anaerobic fungi, and has been widely applied to the study of anaerobic fungi in a range of mammalian herbivores (Table 1). In this section, the current state of the art regarding barcoding loci for anaerobic fungi is reviewed, and the reasons behind the recent move within the research community toward the use of the large subunit (LSU) 28S rRNA as a barcoding locus are highlighted.

Whilst genetic markers have value in enabling molecular tools to be developed and establishing a taxonomic structure, it is the genomes that provide a key foundation in terms of understanding anaerobic fungal biology. The first thorough analysis of the genic, intergenic and rRNA encoding regions of a variety of genomic segments of an anaerobic fungus was performed by Nicholson et al. (2005) using directed plasmid libraries. The study provided observations on rules governing intron boundaries, the codon biases observed with different types of genes, and the sequence of an anaerobic fungal promoter. However, anaerobic fungal genomes have proven notoriously difficult to sequence due to their high AT-content, repeat-richness, complex physiology and unknown ploidy. Since this initial study, five anaerobic fungal genomes are now published and publically available, which has been the direct result of improvements in long-read sequencing technologies to overcome the aforementioned issues: Piromyces sp. E2, P. ruminantium C1A (formerly Orpinomyces sp. C1A), Neocallimastix californiae, Piromyces finnis, and Anaeromyces robustus (Youssef et al., 2013; Haitjema et al., 2017). Despite this, however, the methods employed and/or developed in the process of generating these genomes have been found to be not always successful when applied to other anaerobic fungi within the research community. As a result, many recent efforts have failed either in terms of (i) generating high enough quality genomic DNA, (ii) assembling short read sequence data or (iii) annotating newly sequenced genomes. In this section, we review the practical considerations and current challenges faced when generating and analyzing anaerobic fungal genomes, as well as highlighting the valuable insights that have already been gained to date.

Whilst genome-based analysis gives insight into the fundamental biology of anaerobic fungi, gene expression brings our understanding a step closer to their actual activity and metabolism. In practical terms, it also overcomes the issue of eukaryotic genomes containing non-coding introns within their genes. Analysis of expressed anaerobic fungal genes, however, is not new, with the first work in this area being conducted more than 20 years ago, and having already generated fundamental biological insights on a limited subset of genes.

Reymond et al. (1992) determined an anaerobic fungal cDNA sequence, the phosphoenolpyruvate carboxykinase-encoding gene from Neocallimastix frontalis. Analysis of the gene's predicted protein structure revealed that the catalytic regions were highly conserved among anaerobic fungal and animal organisms, however, the yeast sequence showed no similarity to the N. frontalis sequence. Gilbert et al. (1992) isolated and characterized a xylanase cDNA from the rumen anaerobic fungus Neocallimastix patriciarum. Sequence analysis demonstrated significant homology between this enzyme and bacterial xylanases, which implied the horizontal transfer of genes between bacteria and anaerobic fungi in the rumen. Subsequently, this research group established a N. patriciarum cDNA library that was screened for xylanases (Xue et al., 1992), and they then modified the xylanase cDNA to obtain a high-level expression of the enzyme in E. coli (Xue et al., 1995).

Whilst other cDNA library based studies have also been conducted, most have focused on a limited number of hydrolytic enzymes (Supplementary Table 1). The first of the more high-throughput studies was conducted by Kwon et al. (2009), where they constructed an expressed sequence tag (EST) library of the rumen fungus N. frontalis. The functional genes from the library were analyzed to elucidate the carbohydrate metabolism pathways of this anaerobic fungus. With the development of next-generation sequencing technology, however, transcriptome based analysis has become the method of choice. Transcriptomics can be either the study of global gene expression as a function of different conditions (i.e., RNAseq) or the generation of a transcriptome for de novo assembly/annotation (often in combination with genome sequencing). In this section, we review the recent transcriptomic developments that have occurred with anaerobic fungi.

Proteomics bridges the gap between transcriptomics and metabolomics, and permits the large scale analysis of proteins. Proteomics can be classified as being either native or translated proteome analysis. The latter involves the translation of individual mRNAs or transcriptomes prior to the proteome analysis, and has been made possible due to the development of NGS techniques. Translated proteome analysis, however, cannot assess post-translational modifications such as glycosylation or methylation, which requires more targeted analyses.

Native proteome analysis involves three key steps: protein separation, sequence analysis and protein identification. The current standard for protein separation utilizes liquid chromatography (LC), where proteins are separated based on characteristics such as polarity and molecular weight (Lin et al., 2003), and then directly analyzed using mass spectrometry (MS). Due to technical advances in MS and the laboriousness of other methods, direct proteome analysis with LC-MS is now becoming the method of choice (Feist and Hummon, 2015). However, only two studies to date have utilized such methods for analyzing anaerobic fungi (Solomon et al., 2016a; Haitjema et al., 2017). Protein separation can also be performed using gel electrophoresis, either one or two dimensional gel electrophoresis (2D-GE). With 2D-GE, native proteins are separated based on isoelectric point (pI, the horizontal first dimension) and molecular weight (the vertical second dimension) (Gorg et al., 2004). The 2D-GE can provide information about individual proteins within a gel; however, due to the variations in protein separation among 2D gels it can be difficult to compare protein expression between biological samples. This has led to the development of two dimensional differential gel electrophoresis (2D-DIGE) (Gorg et al., 2004). With 2D-DIGE protein samples are pre-stained with different fluorescent dyes, and then run on the same gel in order to eliminate gel to gel variation (Figure 6). Sequencing of the gel separated proteins, however, requires additional steps to excise the individual proteins, and then sequence them either with Edman degradation (Aebersord et al., 1987) or mass spectrometry (MS) based methods after digestion (Graham et al., 2007).

Protein identification is performed by searching the acquired peptide mass results of each protein against a database, with the protein database being the key factor for success in identifying a protein. Within the 452 protein sequences currently available in the NCBI database (representing less than 0.3% of the deposited rumen bacterial protein sequences) only four filamentous anaerobic fungal genera are currently represented. Furthermore, the deposited proteins are also heavily biased toward CAZymes due to these being a primary area of research focus. Of the 452 protein sequences there are 109 hexose degrading enzymes and 46 pentose degrading enzyme, whilst only 102 of the deposited sequences are related with intra-cellular carbohydrate metabolism including glycolysis, hydrogenosome metabolism and pseudo-TCA cycle (Kwon et al., 2009).

Using publically available protein data, a virtual proteome map was constructed (Figure 7). Among the 203 protein sequences analyzed, 97 (including cellulase, xylanase, lichenase, and acetylxylan esterase) were predicted to be secretory proteins and the other 106 non-secretory. Of the 106 non-secretory intracellular proteins, 29 were related with hexose or pentose degrading enzymes. The average molecular size of the 203 deposited proteins was 50 kDa. The largest protein was endoglucanase 5A (CAB92326.1) from Piromyces equi (Eberhardt et al., 2000) with 192.9 kDa, and it contained four GH5s and CBM10s. This enzyme was the only reported anaerobic fungal protein bigger than 100 kDa, and was predicted to be secretory. In contrast, cyclophilin B and its precursor, also predicted to be secretory, were the smallest protein at 19.7 kDa (Chen et al., 1995). More recent proteomic analyses have shown that the size of proteins detected from anaerobic fungi ranges much more widely than previously determined (Haitjema et al., 2017). For instance, a 694.2 kDa protein was observed in the secretome using LC-MS methods, which was not detected using standard gel-based separations due to its large size.

In addition to more advanced proteomics techniques to reach a wider range of proteins, targeted proteomics are required to assess the post-translational modifications previously shown to decorate some of the anaerobic fungal proteins (Haitjema et al., 2017). These targeted methods most commonly consist of a preliminary capture step, where proteins bearing the desired modification are captured using a specific antibody or sugar-binding lectin in the case of glycosylation. The captured proteins are then analyzed using MS. Additionally, glycosylation patterns can be analyzed by treating proteins with a deglycosylation mix and then observing the released sugars with MS. Along with these sample preparation considerations, protein identification is also a challenge in anaerobic fungal proteomics studies. For example, Park (2011) reported that of the 800 protein spots that were selected from gels and analyzed using MS, only 180 protein sequences were identified due to database limitations. The current challenges facing the research community, therefore, are to improve sample preparation procedures as well as increase the amount of available annotated sequence data.

Metabolomics refers to the qualitative and quantitative analysis of the metabolites present inside (endometabolome) and outside (exometabolome) growing cells at a given time (Mashego et al., 2007). The metabolome of an organism is comprised of numerous different types of metabolites (e.g., carbohydrates, fatty acids, alcohols, organic acids, amino and non-amino acids, lipids, etc.) with each having a different origin and function (Villas-Bôas et al., 2005). As the metabolites define the phenotype of an organism, their analysis gives insights into both genetic and environmental changes. Furthermore, metabolomics helps to provide less biased information on genotype and phenotype (Abdelnur et al., 2014), as information contained in the genome, transcriptome or proteome does not always result in the phenotype. Metabolomics, therefore, is a complementary method to use in combination with other ‘omics techniques (Villas-Bôas et al., 2005; Mashego et al., 2007). Despite several technical advancements in sample processing and data analysis, however, no single technique can fully elucidate the metabolome. Therefore, a combination of different approaches is required, with mass spectrometry (MS) and nuclear magnetic resonance (NMR) currently being the most commonly used.

Three distinct approaches are used in metabolomics studies: targeted analysis, metabolite profiling and metabolic fingerprinting. Each of these approaches however, has its own advantages and disadvantages (Shulaev, 2006). Targeted analysis is the most developed quantitative approach, whereas metabolite fingerprinting and metabolite profiling are more global approaches that are only semi-quantitative. Numerous studies have been published on the targeted analysis of anaerobic fungal metabolite profiles of several genera in axenic culture (Table 4). In contrast, only one non-targeted metabolomics based study of anaerobic fungi has been published to date (Cheng et al., 2013).

The relative abundances of the metabolites produced by anaerobic fungi are likely to vary based on substrate or strain. The major fermentation products of all species of anaerobic fungi in axenic culture, however, are hydrogen, carbon dioxide, formate, acetate and lactate, with succinate, and ethanol normally produced in lesser quantities (Table 4). Propionate and butyrate, major fermentation products of rumen bacterial metabolism, are not produced by anaerobic fungi. Using an NMR-based metabolomic approach, Cheng et al. (2013) showed for the first time that α-ketoglutarate was also a major fermentation product of anaerobic fungi. Lactate, succinate, α-ketoglutarate, and ethanol are produced in the cytosol of anaerobic fungi, while hydrogen, carbon dioxide, and acetate are produced in the hydrogenosome (O'Fallon et al., 1991; Kwon et al., 2009; Cheng et al., 2013). Formate is the only product that is produced in both the cytosol as well as the hydrogenosome. In addition to major fermentation products, many fungi produce molecules termed secondary metabolites, which have putative functions in diverse roles such as defense, signaling, and stress response (Keller et al., 2005). Secondary metabolites are an important class of molecule to study, because they have been harvested from other clades of fungi for use as antimicrobial compounds, anti-tumor agents, and insecticides. They are often difficult to characterize, because they are typically produced at low levels or only under certain conditions. To date, there have been no studies characterizing the secondary metabolites produced by anaerobic fungi.

Numerous in vitro studies have shown that some of the typical end products of anaerobic fungal metabolism are often not accumulated when the fungi are grown in co-culture (Table 5). Hydrogenotrophic methanogens utilize hydrogen and formate as substrates, resulting in the production of methane (Theodorou et al., 1996). In contrast, co-culturing of anaerobic fungi with the aceticlastic methanogen Methanosarcina barkeri resulted in the accumulation of formate, and conversion of acetate into methane (Mountfort et al., 1982). Aceticlastic methanogens however represent a very minor part of the rumen archaeal community (Henderson et al., 2015). Numerous studies have confirmed increased production of carbon dioxide and acetate in the presence of hydrogenotrophic methanogens, while the production of lactate, succinate, α-ketoglutarate, and ethanol decreases (Table 5). Cheng et al. (2013) also showed that citrate was a major fermentation product of anaerobic fungi, when these were grown in co-culture with methanogens.

As well as influencing metabolism, the presence of methanogens also enhances the lignocellulolytic activities of anaerobic fungi (Bauchop and Mountfort, 1981; Nakashimada et al., 2000; Leis et al., 2014). This is mainly due to interspecies hydrogen transfer leading to methane production and efficient regeneration of oxidized nucleotides like NAD+, NADP+ (Dollhofer et al., 2015). This syntrophic association of anaerobic fungi and methanogens is well defined, with methanogens also attaching themselves to fungal rhizoids and sporangia (Jin et al., 2011; Leis et al., 2014; Wei et al., 2016a). In contrast to the numerous methanogen co-culture studies though, a more limited number of studies with rumen bacteria and protozoa have been performed (Table 5).

Rumen fibrolytic bacteria and anaerobic fungi compete for the same substrates. However, the ability of anaerobic fungi to physically disrupt the plant particles using their invasive rhizoids gives them an advantage over bacteria when utilizing lignocellulosic substrates. The invasive growth of the anaerobic fungal rhizoids can, however, benefit the rumen bacteria by increasing surface area and/or access to internally exposed areas of the plant. Several studies have been conducted on cocultures of anaerobic fungi with fibrolytic and non fibrolytic bacteria (Table 5). The bacterial cultures generally result in a reduction of the fibrolytic activity of anaerobic fungi (Williams et al., 1991; Roger et al., 1992; Bernalier et al., 1993). Most of the fungal and bacterial co-culture studies have also shown increased carbon dioxide and decreased hydrogen, formate, lactate, and ethanol production relative to the fungal monoculture, with varying acetate levels (Table 5).

Rumen protozoa are generally thought to be antagonistic toward anaerobic fungi. Certain protozoa have been reported to prey on anaerobic fungal zoospores (Gordon and Phillips, 1998; Newbold et al., 2015), produce fungal cell wall degrading enzymes (Morgavi et al., 1994b) and decrease the fibrolytic activity of anaerobic fungi (Widyastuti et al., 1995). In vivo, it has also been shown that removal of protozoa from the rumen increases the anaerobic fungal population. Similar to fungal co-cultures with methanogens and bacteria, culturing of anaerobic fungi with protozoa also results in a fermentation shift from lactate toward enhanced acetate production (Table 5).

Overall, the information regarding anaerobic fungal metabolites is primarily from targeted analysis, and not all of the major metabolites have been measured in studies to date. Particularly, α-ketoglutarate and citrate production in axenic and co-cultures of anaerobic fungi respectively, which have only recently been discovered to be major products (Cheng et al., 2013). As a consequence there is a limited ability to systematically compare fermentation profiles of different anaerobic fungal genera/species/strains, especially in co-culture studies and when different substrates and/or media have been used. The novel insights gained by Cheng et al. (2013) also highlight that the use of global metabolomics approaches are key to furthering our understanding of the fundamental biology of anaerobic fungal metabolism both in pure and mixed culture.

Anaerobic fungi are the most effective fiber degrading organism in the herbivore gut, with numerous studies confirming their value as a direct fed microbial for ruminants. Reported benefits include improved feed intake, feed digestibility, feed efficiency, daily weight gain and milk production (Lee et al., 2000; Dey et al., 2004; Paul et al., 2004; Tripathi et al., 2007; Saxena et al., 2010; Puniya et al., 2015). Furthermore, anaerobic fungi have a well-established syntrophic interaction with rumen methanogens. On this basis, anaerobic fungi should therefore be of great interest in rumen microbial studies considering current research is focused on increasing the sustainability of ruminant livestock production and decreasing its environmental footprint. Despite this, the characterization of anaerobic fungi in rumen microbial ecology studies is not routinely undertaken, even though suitable molecular tools are available.

Molecular tools for ecological studies of anaerobic fungi now provide better depth of characterization and taxonomic resolution than before, as it is apparent that several of the previously used ITS1 primers did not provide complete coverage of the Neocallimastigomycota phylum. In addition to ITS1 based profiling methods such as ARISA and DGGE (Table 1), tools and data files have also been generated to support the bioinformatics analysis of NGS amplicon based analysis of this region (see section Internal transcribed spacer region). Several quantitative PCR methods also exist based on the ITS1 region (Denman and Mcsweeney, 2006; Lwin et al., 2011; Kittelmann et al., 2012; Marano et al., 2012), as well as the more highly conserved 5.8S rRNA gene (Edwards et al., 2008). The 28S rRNA gene, however, is likely to become the barcode of choice for targeted anaerobic fungal studies in the future due to its benefits over ITS1 as a barcoding locus for this phylum, although the combined use of both barcodes may also have merit. The ITS1, however, will remain of value in the detection of potential novel anaerobic fungal ITS1 sequences in wider environmental surveys, where the diversity of all fungi is of interest (Schoch et al., 2012). A recently published gene method for quantifying fibrolytic activity of cultures using cDNA also offers a new avenue to expand our understanding of how different factors affect fiber degradation by anaerobic fungi (Dollhofer et al., 2016). The increasing amount of anaerobic fungal genomic information will also in the future enable further such functional based quantification assays to be developed. This is important, as quantitatively anaerobic fungi are often incorrectly considered to be a minor part of the rumen microbial community.

A large proportion of early rumen microbial studies focused primarily on the analysis of rumen fluid, where anaerobic fungi only occur transiently as zoospores. Furthermore, whilst rumen zoospore numbers are low compared to counts of bacteria and archaea; anaerobic fungi have been reported to represent up to 20% of the rumen microbial biomass (Rezaeian et al., 2004): a figure comparable to the 10–16% of rRNA transcript abundance in metatranscriptomic studies (Elekwachi et al., 2017). This is because much of the rumen anaerobic fungal biomass is intimately associated with dietary plant material, due to the nature of the invasive growth of the rhizomycelium during the vegetative growth stage of the anaerobic fungal life cycle. This is why it is crucial that ruminal solids are also sampled when conducting rumen microbial studies, not just the ruminal liquid, as otherwise the amount of anaerobic fungi will be greatly underestimated. This is important not just for anaerobic fungi, however, as other microbial taxa also differ between ruminal liquid and solid fractions (Henderson et al., 2013).

Ruminal sampling is always best performed via a ruminal cannula, as oral stomach tubing is capable of sampling only small feed particles which are normally highly degraded. A ruminal sample of at least 500 grams should be obtained, ideally from the pooling of samples collected at multiple different locations (i.e., the front and middle of the ventral sac and the cranial sac). If required, this sample can then be portioned into liquid and fiber fractions using the French-press method described by Kong et al. (2010) or by using cheesecloth. The sample preparation method used to isolate RNA or DNA from the collected ruminal samples, however, will also have a large effect on whether or not the sample extract is truly representative in terms of anaerobic fungal nucleic acid content.

Subsamples of rumen content that utilize only 100–200 mg quantities for isolation purposes and rely on chemical lysis or bead beating methods to isolate anaerobic fungal RNA/DNA are considered to be inadequate (Wang et al., 2011). With popular bead-beating based techniques the impact of the beads is cushioned by the fiber matrix, causing incomplete lysis of anaerobic fungal cells. If such methods are used, however, it is important to pre-process samples using mechanical grinding either in combination with freeze-drying or liquid N, as hand grinding is inadequate (Wang et al., 2011). Ground samples not only enable a better extraction of DNA from the microbes associated with the fiber matrix, but also allow a more representative ruminal sample to be extracted. This is important when only milligram quantities are used for extraction. Recently, Elekwachi et al. (2017) optimized a liquid N grinding/TRIzol extraction method which enabled effective lysis of anaerobic fungal cells, and resulted in a representative proportion of anaerobic fungi being detected in the sequence data subsequently generated. Application of this extraction method in future metagenomic and metatranscriptomics studies should, therefore, result in a more accurate understanding of the contribution of anaerobic fungi to the rumen microbiome.

Anaerobic fungi are often considered to play a limited role in the rumen when animals are fed diets supplemented with cereals. However, due to some anaerobic fungi having amylolytic activity, mixed effects of grain supplementation on anaerobic fungal numbers have been reported in the literature (Gordon and Phillips, 1998). Furthermore, fermentation of certain types of carbohydrates has also been shown to be strain specific (Trinci et al., 1994; Orpin and Joblin, 1997). Based on this information it is perhaps, therefore, not surprising that the type of anaerobic fungi enriched from rumen content samples was found to vary with the type of carbon source used (Griffith et al., 2009). Combined with the recent contrasting transcriptomic findings with respect to the influence of substrate type on the expression of CAZymes (section Gene expression analysis of axenic/monoxenic anaerobic fungal cultures), it is clear that there is an urgent need to rethink our understanding of anaerobic fungi as simply “fiber degraders.” There is now a clear evidence base supporting the concept of niche differentiation of anaerobic fungi in response to carbon source (Griffith et al., 2009), as well as the type of herbivorous host (Liggenstoffer et al., 2010).

As all the currently described Neocallimastigomycota belong to just one family, it is perhaps not entirely unexpected that there are increasing reports of anaerobic fungi in novel herbivorous hosts such as the iguana (Mackie et al., 2004; Liggenstoffer et al., 2010), sea urchin (Thorsen, 1999), and termite (Lee et al., 2015). More interestingly, however, there are increasing reports of anaerobic fungi being detected in non-gut environments using molecular and/or microscopy based techniques (Lockhart et al., 2006; Mohamed and Martiny, 2011; Ivarsson et al., 2016; Wurzbacher et al., 2016). Whilst some of these reports may be explainable by contamination of the environment by herbivorous hosts, it has recently been suggested that a novel lineage may exist within phylum Neocallimastigomycota (Picard, 2017). Until viable anaerobic fungal cultures are isolated from these non-gut environments, however, the ability of anaerobic fungi to exist in a free-living form will always be treated with skepticism. Perhaps to some extent this has a feeling of déjà vu, when 40 years ago the first reports of an anaerobic fungus challenged the dogma of the time.

To date the long term availability of anaerobic fungal cultures has been an issue, with several type strains being lost over the years. This is due to the inability of commercial culture collections to propagate anaerobic fungal cultures. Therefore, if a culture is deposited in a collection, it is only as good as the shelf-life of the corresponding cryovial. This is a problem that anaerobic fungal researchers are actively seeking to resolve through improving cryopreservation methods (Solomon et al., 2016b) and developing a centralized non-commercial repository of anaerobic fungal type strains and/or well characterized cultures within the anaerobic fungal research community.

Anaerobic fungal cultures will always be essential to our ability to understand the biology of anaerobic fungi, particularly as there is an increasing need to link ‘omics data with activity and physiology. This is even more important now that it is recognized that a large proportion of anaerobic fungal sequences cannot be functionally annotated, including highly expressed “core” genes. In order to help deliver improved models for future annotations, databases need to be (a) promptly updated with sequenced anaerobic fungal genomes as well as (b) strategies developed to characterize the function of the currently non-annotated “core” genes.

Anaerobic fungi are a central component of the rumen microbiome, and are well established in terms of their key role in ruminal fiber degradation. Application of PCR based approaches in the past decade has increased our understanding of their diversity, and highlighted that many novel taxa remain to be cultivated. Understanding of their ecological role in the rumen is, however, undergoing a paradigm shift in light of the increasing evidence base indicating that anaerobic fungi have undergone substantial niche differentiation. There is a need to move beyond the broad generalization of anaerobic fungi as fiber-degraders, and explore the fundamental differences that underpin their ability to exist in distinct ecological niches. Application of genomics, transcriptomics, proteomics and metabolomics to their study in pure/mixed cultures and environmental samples will be invaluable in this process. A more central key challenge however, is advancing our understanding of the biology of axenic cultures using the wide range of ‘omics approaches that have been successfully developed to date. A central problem for all analyses is the limited functional annotation of anaerobic fungal sequence data. There is, therefore, an urgent need to expand information held within publicly available reference databases. Once this challenge is overcome, along with improved rumen sample collection and extraction, the application of these techniques will be key in furthering our understanding of the ecological role and impact of anaerobic fungi.

JE initiated and co-ordinated the writing project. All authors contributed text to the manuscript. All co-authors contributed to the development of the manuscript as a whole by giving constructive feedback on the manuscript during its preparation. All authors gave approval of the manuscript for publication.