Decomposition of Fomes fomentarius fruiting bodies – transition of healthy living fungus into a decayed bacteria-rich habitat is primarily driven by Arthropoda

Abstract Fomes fomentarius is a widespread, wood-rotting fungus of temperate, broadleaved forests. Although the fruiting bodies of F. fomentarius persist for multiple years, little is known about its associated microbiome or how these recalcitrant structures are ultimately decomposed. Here we used metagenomics and metatranscriptomics to analyse the microbial community associated with healthy living and decomposing F. fomentarius fruiting bodies to assess the functional potential of the fruiting body-associated microbiome and to determine the main players involved in fruiting body decomposition. F. fomentarius sequences in the metagenomes were replaced by bacterial sequences as the fruiting body decomposed. Most CAZymes expressed in decomposing fruiting bodies targeted components of the fungal cell wall with almost all chitin-targeting sequences, plus a high proportion of beta-glucan-targeting sequences, belonging to Arthropoda. We suggest that decomposing fruiting bodies of F. fomentarius represent a habitat rich in bacteria, while its decomposition is primarily driven by Arthropoda. Decomposing fruiting bodies thus represent a specific habitat supporting both microorganisms and microfauna.


Introduction
Fomes fomentarius is a basidiomycete from the family Polyporaceae and a common deadwood-decomposing fungus throughout the Northern Hemisphere forests, preferentially colonizing Eur opean beec h ( Fagus sylv atica ) and silv er birc h ( Betula pendula ) w ood (Schw arze 1994 , Müller et al. 2007, Kacprzyk et al. 2014 ).It is an effective white rot decomposer and an important contributor to forest carbon and nitrogen cycling.As a white rot fungus, F. fomentarius is capable of digesting all the major components of deadwood (lignin, cellulose and hemicellulose) resulting in more complete decomposition of the deadwood and preparing the deadwood for organisms which are unable to digest lignin (Fukasa wa 2021 ).F omes fomentarius is both a frequent parasite of living trees and an early coloniser of deadwood.As the latter, it has a stimulatory effect on deadwood colonization by many other wood-decay fungi, influencing the de v elopment of the subsequent micr obial comm unity (Heilmann-Clausen and Bod d y 2005 ).
While the main role of F. fomentarius is deadwood decomposition, the fruiting bodies also provide other ecosystem services.Living spor ocar ps may serv e as a habitat for other taxa, are known to be associated with deadwood-inhabiting insects (Thunes 1994, Økland 2002, Friess et al. 2019, Seibold et al. 2019 ) and co-occur with bark beetles.Bark beetles may hav e de v astating effects on forests (Štursová et al. 2014 ) and are potentially all mycophagous at some point in their life cycle (Harrington 2005 ).Fungal fruiting bodies have been demonstrated to host specific bacterial communities that differ among fungal species (Pent et al. 2017, Liu et al. 2018, Gohar et al. 2020, P ent et al. 2020 ).Ho w e v er, no studies hav e specificall y addr essed the micr obiome of wood-r otting fungi, including the fruiting bodies of F. fomentarius which can live for up to 25 years (Thunes 1994 ).
After death, fungi continue to influence their associated communities .T hunes and Willassen ( 1997 ) identified the living/dead state of fungal fruiting bodies as the major factor determining the associated beetle community.While, in soils, decomposing fungal biomass r epr esents hotspots of micr obial acti vity and di versity, due to a r elativ el y high nitrogen content in an otherwise nitrogenlimited envir onment (Br abcov á et al. 2016 ).The rate of mycelia decomposition is related to the carbon: nitrogen ratio.Saprotrophic fungi with hard fruiting bodies, similar to wood, and a high carbon: nitr ogen r atio ar e decomposed slowl y (Koide and Malcolm 2009, Fernandez and Koide 2014, Br abcov á et al. 2018 ), providing a longer duration when they can serve as a habitat or growth substrate for other organisms.
We aimed to compare the symbiotic and decomposer community of natur all y-occurring healthy living and rotten fruiting bodies of F. fomentarius using metagenomic and metatranscriptomic analyses to evaluate the role of different organisms during the time when living fruiting bodies produce fungal spores as well as during their decomposition.This allows us not only to identify the micr obial comm unity that co-exists with F. fomentarius and their functional traits but also to determine the transcriptionally active fraction of the community.We hypothesised that the fruiting bodies would serve as a growth substrate for mycophagous fungi and bacteria, as was pr e viousl y observ ed when dead fungal mycelia were incubated in soil or litter (Br abcov á et al. 2016, Br abcov á et al. 2018 ).Due to the high abundance of F. fomentarius in Northern Hemispher e for ests, this work will shed light on the dynamics of one of the major drivers of deadwood decomposition and carbon cycling in this environment.

Sampling procedure
Ten fruiting bodies of F. fomentarius were sampled in the Žofínský Pr ales National Natur e Reserv e in the Czech Republic (48 • 39 57 N, 14 • 42 24 E) in September 2017.Fruiting bodies without signs of infection or damage were putatively considered healthy living while fruiting bodies with changed colour (darker and with neither a visible white line of growth nor fresh hymenia), visible infection, physical damage and no sign of fresh growth or spore production wer e putativ el y consider ed as dead and r otten.One putativ el y healthy living and one putativ el y r otten fruiting body was sampled from each of five selected decomposing tree trunks of Fagus sylv atica , the tr ee species dominant in this ecosystem (Baldrian et al. 2016 ).The mean size of the sampled fruiting bodies was 15-25 cm in diameter.The trees were chosen when both healthy living and rotten fruiting bodies wer e pr esent on the same deadwood object with pr efer ence giv en to tr ees that wer e part of the longer term monitoring experiment.All trees were of a similar diameter (50-80 cm), decomposition stage (15-30 years of decomposition) and came from the same locality (50-300 m between deadwood).The fruiting body was cut from the tree using a sterile knife and all surface parts potentially associated with organisms that sedimented on the surfaces wer e r emov ed.The entir e inner part was shredded into fine chips, using a hand saw, and immediately flashfrozen in liquid nitrogen before being transported to the lab and stored at −80 • C.

Nucleic acid extraction and sequencing
Prior to DN A/RN A extraction, the sample w as homogenized in liquid nitrogen by mortar and pestle.Total RN A w as extracted, in triplicate, from 200 mg of sample material using the NucleoSpin RNA Plant kit (Mac her ey-Na gel) according to manufactur er's pr otocol after mixing with 900 μl of the RA1 buffer and shaking on FastPrep-24 (MP Biomedicals) at 6.5 m.s −1 twice for 20 s.Triplicates were pooled and treated with OneStep PCR Inhibitor Removal kit (Zymo Research).The DNA was r emov ed using Turbo DNA-free kit (Invitrogen) and the efficiency of DNA r emov al was confirmed by PCR with the bacterial 515F (5 -GTGCCAGCM GCCGCGGTAA-3 ) and 806R (5 -GGA CTA CHV GGGTWTCTAAT-3 ) primers (Caporaso et al. 2012 ).Bacterial rRN A w as depleted with the MICROBEexpress kit (Ambion).Eukary otic, c ytoplasmic and mitochondrial rRN A depletion as w ell as sequencing libr ary pr epar ation wer e done using the Trueseq Stranded Total RNA Gold kit (Illumina) following the manufacturer's instructions .T he RNA quality was assessed using a 2100 Bioanalyzer (Ag ilent Technolog ies) after both DNAse treatment and library preparation.The libraries were sequenced on an Illumina HiSeq 2500 (2 × 250 bases) at Brigham Young University Sequencing Centre, USA.
Total genomic DN A w as extr acted fr om 200 mg of homogenized frozen material described above using the NucleoSpin Soil Kit (Mac her ey-Na gel, German y) following the manufactur er's instructions.Briefly, cells were lysed using SL1 lysis buffer with Enhancer SX added prior to lysis .T he samples were homogenized using a FastPrep-24 (MP Biomedicals, Santa Anna, United States) at 5 m.s −1 for 2 × 30 s. Two extractions per sample were performed and pooled.The metagenomic sequencing library was prepared using TrueSeq Nano DNA libr ary pr epar ation kit (Illumina) according to the manufactur er's pr otocol and sequenced in-house on the Illumina MiSeq (2 × 250 bp).
A putative taxonomic identity was assigned to each gene by selecting the highest bit score after a BLAST sear ch.Sear ches w ere performed against a database consisting of (1) an in-house assembled genome and transcriptome of a pure F. fomentarius cultur e isolated fr om a fruitbody collected in the study area (Supplementary File 1), (2) 2307 fungal genomes downloaded from the Joint Genome Institute's Mycocosm fungal genomics portal ( https: //mycocosm.jgi.doe.gov/, Grigoriev et al. 2014 ) and (3) the NCBI non-r edundant pr oteins database (downloaded 07/10/2022).Due to the challenges of taxonomic identification in metagenomes, we ar e onl y r eporting r esults at the famil y le v el or higher.
To obtain functional predictions, genes were annotated with KEGG Orthologs (KOs) by comparing the amino acid residues with the KOfam database (downloaded on October 2022, Aramaki et al. 2020 ) using HMM profiles and hmmsearch (HMMER v3.3.2, Ed d y 2011 ).Matc hes wer e onl y consider ed for scor es higher than the pr edefined thr esholds and an e-v alue lo w er than 1e −5 .Genes encoding the Carbohydr ate-Activ e Enzymes (CAZymes) were annotated using the dbCAN HMM database V6 (Huang et al. 2018 ).The default settings were used for all tools.
The complete analysis script is available on Github ( https: // github.com/jasonbosch/ Decomposing-Fomes-fomentariusfruiting-bodies-r epr esent-a-habitat-primarily-driven-by-Arthropoda ).In brief, the assembled metagenomes and metatr anscriptomes wer e imported and the number of r eads matc hing each gene were counted using Rsubread's featureCounts function using the default settings.To facilitate comparison between different samples, all gene counts, in both the metagenomes and metatr anscriptomes, wer e conv erted to tr anscripts per kilobase million (TPM).
In order to confirm the healthy living/rotten status of the samples, we determined the Euclidean distance between each sample .T hese samples were then plotted on a heatmap and analysed through Principal Components Analysis (PCA) with the function pr comp().Furthermore, samples w ere clustered via hierachical clustering, based on the Euclidean distance, with hclust() using the method Ward.D2 which showed the best clustering structure.The clusters were bootstrapped with shipunov::Bclust() using 1000 iterations.Based on the combined results of the metagenomic and metatr ansciptomic anal yses, the putativ e field classification was either retained or the sample was reassigned as either healthy living or rotten.
The community analysis and CAZyme expression was performed by a ggr egating the genes/transcripts by their attributes at differ ent taxonomic le v els.Metabolic potential was tested by converting KEGG modules into a logical expression and assessing the pr e viousl y assigned KOs of genes with a TPM > 0.

Classification of fruiting bodies
Fruiting body decomposition is a continuous process and visual assessment of the fruiting body a ppear ance may not be fully reliable to identify the stage of decomposition of the fruiting body.Mor eov er, as material was sampled from inside the fruiting body, it was impossible to determine whether the outer a ppear ance of the fruiting body corresponded to the status of its internal tissue while in the field.Supported by the metagenomic profile analysis ( Fig. S1 ), we confirmed the classification of all putative healthy living samples (live fruiting bodies) as well as of the putative rotten samples H03, H16 and H20.Ho w e v er, the meta genomic analysis sho w ed putativ e r otten samples H08 and H22 to cluster with the healthy living samples ( Fig. S1 ).Sample H19 sho w ed an intermediate metagenomic profile, but, due to the metatranscriptomic profile (the presence of F. fomentarius mRN A), w as classified as healthy living.Sample H22 was excluded from further metatr anscriptomic anal ysis due to the low number of reads obtained ( < 3500).

Community composition of fruiting bodies
In the metagenome, we observed that ∼81% of the untransformed reads and ∼76% of the reads by TPM matched to the F. fomentarius genome in the samples from healthy living fruiting bodies (hereafter healthy living samples; Fig. 1 ).Less than 0.1% of reads matched to F. fomentarius in the rotten samples, indicating an absence of F. fomentarius DNA.Sample H19 showed a high abundance of bacterial sequences and a low abundance of F. fomentarius , indicating that it is in a transitional stage between healthy living and r otten.The r otten meta genome primaril y consisted of r eads assigned to bacteria ( ∼93% untransformed reads and ∼86% by TPM) and displayed higher taxon diversity than that of the healthy living samples; best hits of predicted genes in the metagenomes belonged to a median of 1151 families in rotten samples and 917 in the healthy living samples.
The majority of the rotten metagenomic community could be assigned to Proteobacteria with a m uc h smaller contribution from Actinobacteria and Bacteroidetes ( Fig. S2 ).In these samples the most-abundant families were Burkholderiaceae (10.7%),Microbacteriaceae (9.7%) and Sphingomonadaceae (8.2%).The abundance of Ascomycota was notably higher in the rotten samples, up to almost 6% in sample H20 ( Fig. S3 ).In healthy living samples, the dominant family of Ascomycota was Magnaporthaceae but their relative share was low.Only sample H03 contained reads of one particularly dominant family: the Hypocreaceae.Samples H20 and H16 both had complex complements of Ascomycota with Hy alosc yphaceae, undefined Helotiales, Cephalothecaceae and Helotiaceae making up the largest share of the fungal reads.
T he metatranscriptome displa yed an o v er all similar composition to the metagenome; ∼67% of the untr ansformed r eads and ∼51% of the reads by TPM matched to F. fomentarius in the healthy living samples (Fig. 1 ) .In contrast to the metagenome, the metatranscriptome of the sample H19 closely matched that of the other healthy living samples.We also observe that most transcription in the rotten samples belonged to transcripts which could not be assigned to any particular taxon, indicating that the presence of various bacteria does not imply that they are all transcriptionall y activ e .T he origin of most r otten tr anscripts r emained unknown e v en at the phylum le v el ( Fig. S2 ).

Functional potential of the bacterial community
In order to understand the metabolic potential in the bacterial comm unity, we compar ed the KEGG Orthologs identified in the metagenome to those required for the complete metabolic pathways for methane, nitrogen and sulfur metabolism and photosynthesis as defined in KEGG modules (Kanehisa 2000 ).Due to the difficulties of matc hing meta genomic data to eukaryotic genes, we limited the analysis to bacterial genes only.Excluding nitrogen fixation, our analysis revealed that at least complete pathways could be reconstructed from the metagenomic data for formaldehyde assimilation, F420 biosynthesis, methanofuran biosynthesis, anoxygenic photosystem II and assimilatory sulfate reduction ( Table S1 ).In se v er al cases, complete pathways could be assigned to highly-abundant bacterial families ( > 1% of the TPM contributed by bacteria).
Regarding nitrogen metabolism, both the complete assimilatory and dissimilatory nitrate reduction pathways were present in both the healthy living and rotten fruiting bodies ( Table S1 ).Dissimilatory nitr ate r eduction was potentiall y performed by Burkholderiaceae , Yersiniaceae , Comamonadaceae , Rhodanobacter aceae and Br adyrhizobiaceae.While the necessary genes to perform denitrification were detected in rotten fruiting bodies, this pathw ay w as not complete in any of the highly-abundant bacterial families .T he pathwa y for nitrogen fixation was not complete in either the healthy living or rotten fruiting body meta genome.A searc h for the individual nitrogenase subunits, nifH (K02588), nifD (K02586) and nifK (K02591), r e v ealed no matches in the metatranscriptome and nifD transcripts in only four metagenomes (H03, H07, H20 and H21), all with a best hit to Xanthobacter aceae (Pr oteobacteria).

CAZyme expression
Carbohydr ate-Activ e Enzymes (CAZymes) have a known role both in decomposition of biopolymers and are important for internal restructuring of F. fomentarius fruiting bodies (Bowman and Free 2006 ).Looking at the metagenomes, it can be observed that a greater number of C AZymes , by TPM, are present in rotten samples; this could reflect more compact bacterial genomes where CAZymes r epr esent a lar ger percenta ge on the genome (Fig. 2 A).Ho w e v er, the metatr anscriptomics sho w ed that the CAZymes of Fomes wer e highl y expr essed in its healthy living fruiting bodies.This was true e v en in the case of sample H19, which contained a lar ge shar e of bacterial sequences (Fig. 2 B), suggesting that F. fomentarius was mor e metabolicall y activ e than the bacteria.Based on the metagenomes, F. fomentarius sho w ed a greater proportion of CAZymes targeting beta-glucans and cellulose compared to non-F.fomentarius CAZymes but lo w er proportions of CAZymes targeting hemicellulose br anc hes and peptidogl ycan (Fig. 2 ).Howe v er, the differ ences in the metatr anscriptome wer e mor e dr amatic.CAZyme expression by F. fomentarius broadly matched the gene complement while non-F.fomentarius expression in the rotten samples primarily targeted either beta-glucans or glycoconjugates (gl ycopr oteins, gl ycolipids or pr oteogl ycans).The pr oportion of CAZymes with an unknown target in rotten samples was also reduced, both in comparison to healthy living samples but also in comparison to the rotten metagenome.Unsur prisingl y, almost all CAZymes expressed in the healthy living sample belonged to F. fomentarius; many of them likely being involved in internal cell-wall remodelling processes but the transcription of genes required for deadwood decomposition was observed as well.CAZymes expressed in the rotten samples originated from a variety of phyla with the largest contribution from Arthr opoda, Actinobacteria, Evosea and Pr oteobacteria (Fig. 3 ).We found that the majority (by TPM) of CAZymes expressed in all but one of the rotten samples, including all CAZymes targeting chitin and the majority (16/29 genes) that wer e pr edicted to target beta-glucans (Fig. 3 , Table S2 ), were produced by Arthropoda.The Arthropoda chitinases were predicted to have the closest similarity to genes of three different families; Tenebrionidae (13 chitinases), Cer ambycidae (1 c hitinase) and Pyr ogl yphidae (1 c hitinase).Additionall y, one tr anscript belonging to Micr omonosporaceae (Actinobacteria) was predicted to target both cellulose and chitin but this was neither widespread nor highly expressed.

Healthy Living Rotten
Beta-glucanases were also produced by other arthropods (the Oppiidae and Acaridae mites and Chrysomelidae beetles), Actinobacteria (the families Micr omonospor aceae and Stre ptom ycetaceae) and Ascomycota (the families Cephalothecaceae and

Healthy Living
Rotten Metatranscriptome CAZymes (B)   Chaetosphaeriaceae). Ho w e v er, except for Micr omonospor aceae, c hitinases wer e not detected and, except for Opiidae, they were not present in all the rotten samples.
The remaining taxa produced their own complements of C AZymes , often targeting specific substrates ( Table S2 ).Acidobacteria produced a pectinase while Evosea, as well as se v er al families of Pr oteobacteria, pr oduced gl ycoconjugate-degr ading enzymes.Chitinopha gaceae (Bacter oidetes) and Myxococcales (Pr oteobacteria) produced CAZymes that targeted mannans and the mannan backbone respectively .Finally , Verrucomicrobia and Kic kxellaceae pr oduced CAZymes with cellulose as a putative target.

Discussion
We aimed to compare the microbial communities found on healthy living and decomposing fruiting bodies of F. fomentarius .Using metagenomics and metatranscriptomics , we ha ve analysed the genetic potential and taxonomic composition of both the complete and active fraction of the community.Special consideration was given to bacterial metabolic pathwa ys , due to the ease of assembling bacterial transcripts, and CAZyme expression.

Community composition of fruiting bodies
The three most abundant bacterial families in the rotten samples of F. fomentarius fruiting bodies were Burkholderiaceae (10.7%),Microbacteriaceae (9.7%) and Sphingomonadaceae (8.2%, Fig. 1 ).Members of the Burkholderia genus (Burkholderiaceae) are known to hav e inter actions with m ultiple hosts, act as a soil sa pr otrophs and producers of anti-fungal compounds (Sto y anova et al. 2007 ).Intriguingl y, Bur kholderia gladioli is a symbiont associated with se v er al Tenebrionidae beetles where it functions to protect their eggs from fungal colonisation (Kaltenpoth and Flórez 2020 ).The majority of CAZymes targeting chitin in the metatranscriptome had a best-hit match to Tenebrionidae sequences.Micr obacteriaceae ar e likel y pr esent due to a r ole in deadwood decomposition.The genus Curtobacterium, from the family Microbacteriaceae, was identified as a globally-distributed degrader of organic matter (Chase et al. 2016 ) and Microbacteriaceae, together with Burkholderiaceae, were observed at high abundance in buried Norway Spruce wood blocks (Valette et al. 2023 ).Furthermore, Burkholderiaceae and Microbacteriaceae were both associated with the fungal genera Penicillium and Trichoderma , which may imply they have other funtions than mer el y deadwood decomposition (Valette et al. 2023 ).Sphingomonadaceae has previously been found, at relatively low levels, in the mycangia of the w ood-burro wing ambrosia beetle Platypus c ylindrus (Nones et al. 2021 ).Additionall y, all thr ee of these bacterial families wer e found at high abundance in either the adult or larval form of the black tinder fungus beetle Bolitophagus reticulatus (Tenebrionidae) collected from F. fomentarius fruiting bodies (Kaczmarczyk-Ziemba et al. 2019 ).
Given the long lifespan of F. fomentarius fruiting bodies, one would expect them to have strong anti-microbial protections and few bacteria present in the metagenome.Setting aside sample H19 due to its transitional state, bacteria comprised 0.03-11.6% of the TPM metagenome of healthy living F. fomentarius fruiting bodies.In the six metagenomes where F. fomentarius was dominant, onl y Micr obacteriaceae (1.2%) was present at greater than 1% relative abundance .T he next most abundant bacterial families were Yersiniaceae (0.3%) and Burkholderiaceae (0.3%).Both Microbacteriaceae and Burkholderiaceae were also present in the rotten F. fomentarius samples.
Regarding the Ascomycota identified on the F. fomentarius fruiting bodies, ther e ar e low le v els of Ma gna porthaceae in all the healthy living samples.While many Magnaporthaceae are known as plant pathogens, they also include sa pr otr ophic fungi, suc h as the genera Plagiosphaera (Song et al. 2019 ) and Muraeriata (Huhndorf et al. 2008 ).Ho w e v er, the exact taxonomic position of these genera is in question (Feng et al. 2021 ).In the rotten samples, the most abundant families belong to three orders: Hypocreales, Helotiales and Sordariales.Hypocreaceae (Hypocreales) are known to act as sa pr otr ophic (Mihál et al. 2007, Kepler et al. 2017 ) mycopar asites (K epler et al. 2017 ) and hav e been observ ed on fruiting bodies of other pol ypor es (Mihál et al. 2007 ).The families Hy alosc yphaceae and Helotiaceae, as well as some taxa undefined at the family level, all belong to the sa pr otr ophic Helotiales order.Helotiales have previously been observed at high relative abundance in deadwood (Br abcov á et al. 2022 ) and increasing in abundance in the soil after tree dieback due to bark beetle infestation (Štursová et al. 2014 ).The Helotiales were also the second most abundant order in the healthy living samples .T he family Cephalothecaceae was at its highest abundance in sample H19 and was a member of the Sordariales order which are known to inhabit soil, wood and dung (Zhang et al. 2006 ).Together, this suggests that some of the detected taxa are either mycoparasitic, notabl y Hypocr eaceae, or incidentall y detected due to their close proximity as fellow deadwood-degrading fungi.

Functional potential of the bacterial community
We wer e particularl y inter ested in nitr ogen metabolism as the deadwood environment is nitrogen-scarce in comparison to the fungal biomass and nitrogen content has been shown to be an indicator of the rate of mycelial decomposition (Koide and Malcolm 2009, Fernandez and Koide 2014, Br abcov á et al. 2018 ).As in deadwood bacterial communities (Tláskal et al. 2021 ), both assimilatory and dissimilatory nitrate reduction pathways were present ( Table S1 ).Ho w e v er, we found no e vidence for potential nitr ogen fixation in the fruiting body metagenomes or metatranscriptomes in either the healthy living or rotten fruiting bodies .T his was intriguing as deadwood bacteria are best positioned to colonise the dead fruiting body and include nitrogen fixing bacteria that associate with wood-decay fungi (Hoppe et al. 2014, Bellenger et al. 2020, Gómez-Brandón et al. 2020, Tláskal et al. 2021 ).
The difference in nitrogen fixation capabilities between the bacteria inhabiting the deadwood and those which colonise the decaying F. fomentarius could hav e m ultiple causes and it's not yet possible to distinguish between them.Nitrogen fixation is inhibited by oxygen, and, while deadwood is anaer obic (Cov ey et al. 2016 ), the oxygen le v els in F. fomentarius fruiting bodies may select a gainst nitr ogen fixing bacteria.Pr e vious studies hav e noted nitrogen-fixing bacteria in other fruiting bodies (Gohar et al. 2020, Pent et al. 2020, Ren et al. 2022 ) and there is no reason to suspect that F. fomentarius r epr esents a dr amaticall y mor e aer obic environment.It should be noted that Gohar et al. ( 2020 ) did not examine an y pol ypor es and their longest-liv ed fruiting bodies onl y survived a month before decay .F .fomentarius fruiting bodies may surviv e for se v er al years and this could result in very different selectiv e pr essur es than experienced by shorter-lived species.Howe v er, the pr e viousl y-mentioned studies identified nitr ogen fixing bacteria in fruiting bodies by taxonomy and not by identifying nitrogen fixation genes.It's possible that the fruiting body communities do not require nitrogen fixing capabilities even though closel y-r elated taxa are capable of performing this ecological role.
The simplest explanation is that the bacterial community is deriv ed fr om the deadwood bacterial comm unity but under goes environmental filtering, in this case against nitrogen-fixing bacteria.The bacterial community of fruiting bodies of soil fungi show a large overlap with the soil (Pent et al. 2017, Liu et al. 2018, Pent et al. 2020 ) or deadwood (Ren et al. 2022 ) bacterial communities, with selection determined by the chemistry of the fruiting body (Pent et al. 2020 ).The nitrogen content of fungal fruiting bodies is m uc h higher than that of deadwood (Baldrian et al. 2016, Br abcov á et al. 2018 ), which might mean that there is no need for energyintensiv e, nitr ogen fixation on fruiting bodies.Alternativ el y, man y beetles are known to live in F. fomentarius fruiting bodies (Thunes 1994, Økland 2002, Friess et al. 2019 ) and are known to transfer fungi to new environments (Eskalen et al. 2013, Seibold et al. 2019 ).Under this scenario, the lack of nitrogen fixation is not due to environmental filtering but because the bacterial community is derived from the beetle inhabitants of the deadwood.This hypothesis may be supported by the fact that the most abundant bacterial families all occur in insect hosts, including those collected from F. fomentarius fruiting bodies (Kaczmarczyk-Ziemba et al. 2019, Kaltenpoth and Flórez 2020, Nones et al. 2021 ) and is congruent with r esearc h sho wing that w ood-boring beetles introduce beetle-associated fungi to deadwood (Skelton et al. 2019 ).Future r esearc h should attempt to distinguish between these two possibilities.
We identified the pathway for methanofuran biosynthesis, necessary for methanogenesis, in the bacterial metagenomes of both the healthy living and rotten fruiting bodies ( Table S1 ).Ho w e v er, despite F. fomentarius fruiting bodies being identified as a source of methane emissions (Mukhin andVoronin 2008 , 2009 ), no complete methanogenesis pathw ays w ere identified.This result held e v en when analysing all KOs simultaneously, regardless of taxonomy or condition.This suggests that the methanogenesis pathway is truly absent as it can not be completed by any combination of genes sequenced in the metagenome.

CAZyme expression
All the substrates targeted by CAZymes in the rotten metatranscriptomes are major components of fungal cell walls.In general, gl ycopr oteins can comprise up to 50% of the mass of fungal cell w alls (Bo wman and F ree 2006 ) while beta-glucans and chitin contribute between 20%-69% and 5%-6% of the fruiting body mass r espectiv el y (Kolundži ć et al. 2016, Sari et al. 2017, Kalitukha and Sari 2019, Pylkkänen et al. 2023 ).CAZymes targeting mannans wer e onl y detected in sample H16 (Fig. 3 ) but this too is a known component of fungal cell walls, although accounting for less than 1% of the dry mass (Kalitukha and Sari 2019 ).This is in line with expectations that the rotten fruiting body community will feed on the fungal structures that remain after the death of F. fomentarius .
The majority of CAZymes exclusiv el y tar geting c hitin wer e pr oduced by arthr opods, specificall y the orders Coleopter a (beetles) and Sarcoptiformes (mites; Fig. 3 & Table S2 ).Se v er al members of both Coleoptera (Sinha 1966, Filipiak and Weiner 2014, McFarlane et al. 2021 ) and Sarcoptiformes (Schneider and Maraun 2005, Sc hneider et al. 2005, Kouk ol et al. 2009, Naegele et al. 2013 ) are known to feed on fungi.Howe v er, it is difficult to confidently assign taxonomic identity to Arthropoda metagenomic sequences due to the lack of available genomes, especially for Coleoptera wher e fe wer than 0.01% of species hav e genomes av ailable (Fer on and Waterhouse 2022 ).
Chitin is not only a component of fungal cell walls but also a major building block of the Arthropoda exoskeleton.As many Arthropod species are known to inhabit, or associate with, F. fomentarius fruiting bodies (Thunes 1994, Økland 2002, Friess et al. 2019 ), it is possible that the Arthropoda CAZymes targeting chitin are intended for restructuring the exoskeleton and not digesting fungal cell walls.We belie v e this is unlikely as many arthropods are known to feed on fungi, CAZymes targeting beta-glucans were also best matches to Arthropoda sequences and no other taxa expr essed CAZymes tar geting c hitin.The amount of c hitin in the fruiting bodies was not quantified and it is possible that the F. fomentarius chitin had already been degraded.T his , again, underlines the importance of having a timeline of fruiting body decomposition.

Taxa involved in F. fomentarius fruiting body decomposition
Our metatr anscriptome anal ysis suffer ed fr om the difficulties with rRNA r emov al (median rRNA content of 75%) and, as a r esult, a r elativ el y shallo w sequencing depth w as ac hie v ed for mRNA and may have resulted in some taxa being omitted.While the high abundance of Arthropoda sequences implies that they are the dominant fungal decomposers, members of at least two microbial families may also play a role in F. fomentarius decomposition.The first is Micr omonospor aceae (Actinobacteria) whic h secreted CAZymes that targeted both beta glucans and cellulose/c hitin.The best-matc h genus was Actinoplanes and Actinoplanes missouriensis has been shown to be attracted to fungal spor es (Ar or a 1986 ) and to hav e anti-fungal ca pabilities due to the pr oduction of c hitinases (El-Tar abil y 2003 ).Micr omonospor aceae tr anscripts wer e not pr esent in all samples and likel y plays onl y a minor role or is present at only a certain stage of the decomposition process .T he second family is Physaraceae (slime mould from the phylum Evosea).Although Physaraceae only expressed gl ycoconjugate-degr ading enzymes, this is the most highly expressed CAZyme by genus ( Table S2 ) and might suggest that Physar aceae decomposes onl y the gl ycoconjugates in F. fomentarius cell walls .Furthermore , several Physaraceae members have been pr e viousl y r e ported to be m ycophagous (Ho w ar d and Currie 1932a ,b ), with Physarum polycephalum and Physarum tenerum plasmodia specifically being identified as feeding on F. fomentarius mycelia (Ho w ar d and Currie 1932b ).
Pr e vious r esults identified specific micr obial comm unities which established on dead fungal material (Brabcová et al. 2016 ) and we expected to find similar microbial, or functional, communities responsible for decomposing the dead fruiting bodies.Contrary to our expectations, almost all CAZyme transcripts targeting chitin came from Arthropoda.The community analysis in Brabcová et al. ( 2016 ) was conducted using 16S and ITS sequencing which would not have detected the involvement of insects even if they were the dominant decomposers .T his suggests that future studies of fungal decomposition should make use of metatranscriptomics to identify the active community members with fewer taxonomic biases.
Alternativ el y, ther e may be qualitative differences between fungal decomposition in the soil as compared to fruiting bodies on deadwood.There is a rich community of bacteria and fungi in the soils which may rapidly colonise dead fungal material (Brabcová et al. 2016, López-Mondéjar et al. 2018, 2020 ).In contrast, the fruiting bodies of F. fomentarius ar e gener all y not in contact with the soil but with the deadwood itself-whose comm unity, perha ps, does not have the same capacity for decomposition-and the microbescarce air.In this situation, it's possible that insects will find and feed on the fruiting bodies before microbes have an opportunity to colonise the dead fruiting body.T here ma y be ecological value in comparing the decomposition of fungal material dependent on whether it has contact with the soil or not and when insect access is restricted or not.Potentially, decomposition of fruiting bodies with soil contact is dominated by microbes while aerial fruiting bodies are predominantly decomposed by Arthropoda.

Limitations
We acknowledge that this work has some limitations, such as c hr onology, sample size and taxonomic identification.While the fruiting bodies were sampled as healthy living and highly decomposed and sorted into a binary classification of either healthy living or rotten, in reality they exhibited a gradient of decreasing F. fomentarius abundance and an increase in the abundance of bacteria and Ascomycota when going from healthy living to rotten samples of F. fomentarius fruiting bodies.Given the limited data a vailable , we can not speculate on the timeline of fruiting body decomposition nor whether it is made up of distinct sta ges.Suc h questions must be answered with further work on more extensive set of fungal fruiting bodies.Despite the small sample size, the differences between healthy living and rotten samples were clear (e.g.F. fomentarius in the healthy living samples accounts for ∼80% of the metagenomic reads compared to < 1% in the rotten samples.)and are consistent within classes.There is no reason to think that larger sample sizes would hav e c hanged our br oad conclusions.We present our results here at the family level due to the difficulties of confidently assigning a taxonomy to metagenomic sequences, particularly for Arthropoda.This is a problem for all work r el ying on sequence databases and is beyond our contr ol.Mor e sensitiv e tec hniques will be necessary to follow up on this work but giving the current technological and database limitations, our results serve as a useful guide for future research.

Conclusion
During the transition from live to rotten fruiting bodies, the microbiome of F. fomentarius fruiting bodies changes to become primarily bacteria-dominated.Despite being physically most closely associated with the adjacent deadwood, the bacterial community of F. fomentarius fruiting bodies did not appear to possess the ability to fix nitr ogen.We r ecorded a greater abundance of CAZymes targeting beta-glucans and cellulose from the F. fomentarius -derived DN A while bacteria-derived DN A had a greater abundance of CAZymes targeting hemicellulose and peptidoglycan.Ho w ever, while metatranscriptomics sho w ed that F. fomentarius CAZyme expr ession br oadl y matc hed its genomic potential, CAZyme tr anscripts expressed in rotten fruiting bodies were enriched for enzymes targeting components of the fungal cell wall, namely beta-glucans, glycoconjugates and chitin.The majority of these CAZymes best matched to Arthropoda sequences, including almost all CAZymes which target chitin.Given prior knowledge of mycophagous beetles and the fact that the Arthropoda CAZymes also target beta-glucans, it appears that Arthropoda are the likely primary decomposers of F. fomentarius fruiting bodies .

Ac kno wledgements
We would like to thank Rubén López-Mondéjar for providing information on CAZyme functionality and targets and Lenka Mic hal čík ová for providing a photogr a ph of the rotten F. fomentarius .We acknowledge support from Talking microbes -understanding micr obial inter actions within One Health fr ame work (CZ.02.01.01/00/22_008/0004597).

Figure 1 .
Figure1.Composition and activity of the microbiome within F. fomentarius fruiting bodies.Famil y-le v el taxonomy assigned to the predicted genes in the (A) metagenomic and (B) metatranscriptomic sequencing of the fruiting bodies .T he abundance of each gene was scaled using transcripts per kilobase million (TPM).Samples are ordered left-to-right in decreasing F. fomentarius TPM in the metagenome .T he P olyporaceae family, to which F. fomentarius belongs, is listed at the top.

Figur e 2 .
Figur e 2. C AZymes in the healthy living and rotten fruiting bodies of F. fomentarius .Each sample is represented by two bars, the narrow, left bar shows whether the identified gene has a best-hit match to F. fomentarius or Non-F.fomentarius genomes while the thicker, right bar shows the CAZyme target substrate .C AZyme target substrates are shown as stacked barplots of (A & B) TPM and (C & D) TPM proportions for both the (A & C) metagenome and (B & D) metatranscriptome.Samples are ordered left-to-right in decreasing F. fomentarius TPM in the metagenome.Any CAZyme targets which accounted for less than 1% TPM were grouped together as "Below 1%."

Figure 3 .
Figure3.Expression of CAZymes targeting particular substrates in rotten F. fomentarius fruiting bodies by phylum.Each bar is coloured to show which phyla express transcripts targeting a particular biopolymer.Any taxa or targets which add up to less than 1% of the total TPM are grouped together as "Below 1%."