CAZyme domain architectures suggest fine-scale functional differentiation among anaerobic fungi and bacteria during lignocellulose conversion to volatile fatty acids

1 Anaerobic fermentation with microbial communities (microbiomes) is an emerging platform for 2 conversion of lignocellulosic biomass to biofuels and bioproducts . The process relies on diverse 3 anaerobic microbes that interact to deconstruct and convert lignocellulosic biomass into a range of 4 products, such as volatile fatty acids (VFAs), which can be achieved by arresting methanogenesis 5 during fermentation. However, defining the distinct functional roles played by various fungi and 6 bacteria during anaerobic biodegradation remains poorly understood. Here, we performed parallel 7 enrichment experiments from cow faeces, goat faeces, and anaerobic digester sludge, selecting for 8 fungal or bacterial dominated communities that convert sorghum biomass into VFAs. 9 Subsequently we reconstructed metabolic networks across the se enrichments based on recovered 10 bacterial metagenome-assembled genomes (MAGs) and fungal isolate genomes and profiled their 11 metabolic activity using metatranscriptomics to identify potential functional niches. Our findings 12 implicate diverse bacteria affiliated with the Bacteroidales and Lachnospiraceae in the direct 13 conversion of lignocellulosic biomass to propionate and butyrate, respectivel y, whereas 14 Neocallimastix-dominated fungal enrichment s converted lignocellulose to lactate, acetate and 15 formate. Analysis of carbohydrate -active enzymes (CAZymes) revealed fine-scale differences 16 between microbes that expressed unique multi-functional enzymes linking two or more CAZymes 17 together with distinct carbohydrate binding motifs , implicating lignocellulose structure as a key 18 driver of selection and niche differentiation. Most of these multi -functional enzymes localized 19 complementary degradation functions together , likely conferring synergistic degradation effects 20 within and between microbiome members . We anticipate that these findings will help inform 21 efforts to develop synthetic microbiomes with tailored functionality for low -cost conversion of 22 lignocellulosic biomass to fuels and bio-based chemicals. 23 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 3

1 Biomanufacturing of chemicals from renewable lignocellulosic biomass has the potential to offset 2 society's reliance on fossil fuels. One biomanufacturing strategy involves the use of anaerobic 3 microbial communities (microbiomes) to deconstruct and convert lignocellulosic biomass into 4 biogas (methane and carbon dioxide) via anaerobic digestion [1]. While biogas can be upgraded 5 to produce bioenergy (i.e., renewable natural gas or electricity), its market value is low, often 6 rendering anaerobic digestion not economically viable. Therefore, higher value chemicals, such as 7 volatile fatty acids (VFAs ; $1 -5/kg selling price ) represent an emerging target product for 8 anaerobic digestion that can be achieved by arresting methanogenesis during fermentation. These 9 products can be chemically upgraded to liquid biofuels [2] or used directly in consumer products, 10 antimicrobials or industrial materials [3]. 11 12 Several studies have demonstrated VFA production via arrested methanogenesis from different 13 organic feedstocks [4]. Conversion of lignocellulosic biomass has largely been achieved by 14 anaerobic fermentations using rumen contents as inocula, resulting in acetate, propionate, and 15 butyrate (C2-C4 VFAs) as the main fermentation products in the absence of an external electron 16 donor [5–7]. During fermentation, the microbiomes responsible for lignocellulose deconstruction 17 and conversion have been reported as containing diverse ruminal bacteria [6, 7], with no studies 18 reporting the presence of biomass degrading ruminal fungi , indicating that they are outcompeted 19 by bacteria . While the overall metabolic processes mediated by ruminal bacteria during 20 lignocellulose degradation to VFAs are generally understood, the specific metabolic networks and 21 interactions driving in situ microbiome metabolism remain largely unresolved. Moreover, the 22 contribution and specific metabolic activities expressed by ruminal fungi during lignocellulose 23 deconstruction are only starting to emerge [8]. 24 25 Here, we enriched five parallel anaerobic fungal and bacterial consortia derived from the cow and 26 goat rumen and an anaerobic digester on sorghum substrate in chemically defined media selecting 27 for VFA production. We quantified metabolite and gas production across the enrichments to 28 examine functional differences between bacteria l-dominated consortia and fung al-dominated 29 consortia, as well as inoculum source. Subsequently we reconstructed metabolic networks across 30 the enrichments based on recovered bacterial metagenome -assembled genomes (MAGs) and 31 fungal isolate genomes and profiled their gene expression using metatranscriptomics. Our results 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 4 identify specific organisms, carbohydrate -active enzymes (CAZymes), and metabolic pathways 1 active in the deconstruction and conversion of sorghum biomass to VFAs. These results were used 2 to infer key metabolic interactions responsible for producing specific VFA end products and 3 activities that compete with VFA production. Together, these findings offer molecular -level 4 insights on anaerobic microbiome function and will inform the rational design of microbiomes for 5 improved lignocellulose conversion to valuable chemicals. 6 7

8 Enrichment and cultivation of anaerobic bacterial and fungal cultures on sorghum biomass 9 Anaerobic consortia enriched for biomass-degrading bacteria and biomass-degrading fungi (herein 10 termed “enrichments”) were cultivated in anaerobic Hungate tubes on 1% (w/w) sorghum (2 mm 11 mill size; Idaho National Laboratory) as a sole substrate in 10 ml of chemically defined M2 media 12 [9] with a ~85% N2, 12% CO2, and 3% H2 headspace and resazurin (0.1% w/v) added as a visual 13 redox indicator. Cultures were initially inoculated inside an anaerobic chamber (AS-580, Anaerobe 14 Systems) with 1 mL of either cow (5 g wet feces in 25 mL M2 media) or goat (3 pellets ~2.5 g in 15 25 mL M2) fecal solutions, or with 0.5 mL of sludge from an anaerobic digester to a final volume 16 of 10 mL M2 media in the presence of either 5 mM chloramphenicol (CH, selecting for fungi) or 17 5 mM 2-bromoethanesulfonate (BES, selecting for bacteria) and passaged every 3 -4 days based 18 on 10% (v/v) transfers. Enrichment cultures were maintained at a pH of 7 -7.5 and a temperature 19 of 39 °C. This resulted in 3 bacterial enrichments from the cow, goat, and digester inocula 20 (cow+BES, goat+BES, digester+BES), and 2 fungal enrichments from the cow and goat inocula 21 (cow+CH, goat+CH). Cow and digester enrichments were passaged for 10 generations (~40 days), 22 while goat enrichments were passaged for 5 generations (~20 days). Analytical m easurements 23 were made on biological triplicate s from each of the 5 enrichments cultures daily over a 4-day 24 period. DNA/RNA samples were collected on Day 3 for metagenomic and metatranscriptomic 25 analysis (see methods below). 26 27 Measurement of fermentation products and total pressure accumulation 28 Total pressure accumulation in the headspace of Hungate tubes was monitored daily using a 29 pressure transducer method as described previously [10]. Quantification of liquid fermentation 30 products, including formate, acetate, propionate, butyrate, valerate, succinate, and lactate, was 31 performed on a 1260 Infinity high-performance liquid chromatography (HPLC) system equipped 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 5 with an Aminex HPX-87H column (part no. 1250140, Bio-Rad) with an inline 0.22 μm filter (Part 1 No. 50671551, Agilent) followed by a Micro -Guard Cation H guard column (Part No. 1250129, 2 Bio-Rad, Hercules, CA, USA) before the analytical column . Samples were prepared by 3 acidification to a concentration of 5 mM H 2SO4, followed by incubation at room temperature for 4 5 min, centrifugation at 21,000 x g for 5 min, and filtration through a 0.22 µm polyethersulfone 5 (PES) membrane filter into HPLC vials with 300 µl polypropylene inserts. Metabolites were 6 separated using a mobile phase of 5 mM sulfuric acid at 50°C with a flow rate of 0.6 ml/min, a run 7 length of 30 min , and an injection volume of 20 μl. Signals were measured using a variable 8 wavelength detector (λ = 210 nm) and s tandard curves for each compound were made at 5 9 concentrations bracketing the range expected in the samples. 10 11 Quantification of hydrogen (H2) and methane (CH4) gas in the headspace of Hungate tubes were 12 analyzed on a Thermo Fisher Scientific TRACE 1300 gas chromatograph (GC) equipped with a 13 TracePLOT™ TG-BOND Msieve 5 A (Part No. 26003 -6100, Thermo Fisher Scientific) and an 14 Instant Connect Pulsed Discharge Detector (PDD) (Part No. 19070014, Thermo Fisher Scientific). 15 100 µL of headspace gas was collected and subsequently purged three times in a 100 µL air -tight 16 syringe and needle. Then, 20 µL of headspace gas was collected and injected directly into the GC. 17 The oven temperature for each run was 30 ºC and the PDD temperature was 150 ºC. High -purity 18 helium (Part No. HE 5.0UH -55, Praxair, Danbury, CT, USA) was further purified with a heated 19 helium purifier (Part No. HP2, VICI) and used as the carrier gas with a flow rate of 0.5 mL/min. 20 The same flushing and analysis procedures were followed for methane and hydrogen standards 21 including 500 ppm H2, 2% H2, 5 % H2, 20% H2, 0.5 % CH4, 1% CH4, 5% CH4, 10% CH4, and 22 20 % CH4 with balance helium (Douglas Fluid & Integration Technology, Prosperity, SC), which 23 were run at each measurement timepoint to account for the PDD baseline that varied slightly each 24 day. 25 26 Extraction and sequencing of DNA and RNA 27 Biomass samples from bacterial enrichment cultures (cow, goat, digester) were collected for DNA 28 and RNA extraction after 72 hours (Day 3) following culture transfer. DNA was collected from 29 bacterial enrichment samples in triplicate, while RNA was collected from bacterial and fungal 30 enrichment samples in duplicate. Hungate tube contents were centrifuged at max speed for 5 min 31 at 4°C and biomass pellets were immediately flash frozen in liquid nitrogen. RNAlater was added 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 6 to samples intended for RNA extraction prior to centrifugation. DNA was extracted from two 1 replicates using the QIAGEN PowerSoil® DNA Isolation Kit (Qiagen, Hilden, Germany) and one 2 replicate using a CTAB protocol. This was done to improve differential coverage binning, where 3 only duplicate DNA samples from the PowerSoil® Kit extraction were used for assessing MAG 4 relative abundance. RNA was extracted from duplicate samples using the QIAGEN RNeasy® Mini 5 Kit with on-column DNAase 1 digestion (Qiagen, Hilden, Germany). DNA and RNA sequencing 6 was performed on the Illumina MiSeq System (v2, 600 cycles) using the Nextera XT DNA Library 7 Preparation Kit (Illumina, CA, USA) and the NEBNext® Ultra II Directional RNA Library Prep 8 Kit at the Biological NanoStructures Lab at UC Santa Barbara. For bacterial metatranscriptomes, 9 rRNA was depleted using the NEBNext® rRNA Depletion Kit (New England Biolabs, MA, USA). 10 For fungal metatranscriptomes, enrichment of mRNA and separation from rRNA was achieved 11 using the NEBNext® Poly(A) mRNA Magnetic Isolation Module. Strand-specific cDNA libraries 12 were prepared with Invitrogen SuperScript II Reverse Transcriptase (Thermo Fisher Scientific, 13 MA, USA). Nucleic acid quantity and quality were determined using a Qubit fluorometer (Thermo 14 Fisher Scientific, MA, USA) and an Agilent TapeStation system (Agilent, CA, USA), respectively. 15 16 Metagenomic assembly, binning, metabolic reconstruction 17 Raw paired-end DNA reads were quality-trimmed and filtered, adapter-trimmed, and contaminant-18 filtered using BBDuk v38.75 and read quality was assessed using FASTQC v0.11.8 . Trimmed 19 reads from the same sample (cow, goat, and digester bacterial enrichments) were co-assembled 20 using MetaSPAades v3.15.3 [11] and mapped to assembled contigs using BBMap v38.75 with 21 minid=0.95. Resulting contigs of at least 1000 bp were binned with MaxBin2 v2.2.7 [12], 22 MetaBAT2 v2.1512 [13], and CONCOCT v1.1.0 [14] and dereplicated using DAS-Tool v1.1.213 23 [15]. Metagenome-assembled genomes (MAGs) were taxonomically classified with GTDB -Tk 24 v2.1.1 (release 9514) [16]. Bacterial MAGs were manually assigned a gram stain based on the 25 taxonomic classification and literature references. Subsequently, gram stain specific prediction of 26 protein product localization for each open reading frame was performed using PSORTb v3.0 [17]. 27 Estimates of MAG completeness and contamination was performed using CheckM v1.1.2 [18]. 28 29 Bacterial MAGs were annotated using the NCBI Prokaryotic Genome Annotation Pipeline version 30 6.10 [40] . Metabolic pathways were annotated across MAGs using gapseq v1.4.0 using the 31 parameters “gapseq find”, “all pathways”, and “bacterial mode ” [41]. Fungal reference genomes 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 7 for the previously isolated and genomically sequenced Neocallimastix lanati [20], 1 Anaeromyces sp. S4 [21], Orpinomyces sp. strain C1A [22], Piromyces sp. E2 [21], 2 Piromyces sp.finn [21], Neocallimastix californiae G1 [21], and Neocallimastix frontalis var. 3 giraffae were retrieved from the Joint Genome Institute MycoCosm database. CAZymes and 4 CAZyme motifs in fungal genomes and bacterial MAGs we re predicted using dbcanLight v1.0.2 5 [23] using ‘cazyme mode’ and ‘sub mode’ options. If multiple distinct CAZyme motifs or substrate 6 targets were predicted for the same open reading frame predictions were concatenated into a single 7 feature. 8 9 Metatranscriptomic analysis 10 Complementary DNA (cDNA) reads were quality filtered as described above for DNA. 11 SortMeRNA [24] was used to remove rRNA sequences using multiple databases for RNA 12 sequences. The remaining non -rRNA reads were used for mapping. RNA reads from fungal 13 enrichment cultures were aligned to a combined fungal reference genome (concatenation of all 14 retrieved fungal reference genomes listed above ) using the bbwrap.sh command from BBMap 15 v38.86 with the usejni parameter set to true, ten threads and all other parameters set to their default 16 values. Bacterial RNA reads were aligned competitively against the total metagenomic assembly 17 using the parameter minid=0.95 and then analyzed per MAG. Aligned reads were counted using 18 the featureCounts command from Subread v2.0.3 using ten threads and all other parameters set to 19 their default values. 20 21 For fungal genomes and bacterial MAGs, read counts from each sequencing sample were 22 normalized for depth and gene length by calculating Transcripts Per Million (TPM) by Equation 23 1. 24 𝑇𝑃𝑀! = "!/$! ∑ ""/$"" × 10& Equation 1 25 26 where TPMi is the TMP for transcript i, Ni is the number of reads mapped to transcript i and Li is 27 the length of transcript i. 28 29 TPM values assigned to the same CAZyme motif, in the same genome and from the same sample 30 (and for bacterial MAGs, the same localization) were summed. The mean TPM values across 31 replicate samples were calculated and used for downstream analysis. For each source (digester, 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 8 cow or goat), bacterial MAGs in the upper quantile of CAZyme expression, and CAZyme features 1 predicted by PSORTb to be extracellular, were retained for differential expression analysis. 2 Similarly, the two fungal genomes with the highest CAZyme expression from either source were 3 retained. Differential expression analysis was performed separately on bacterial MAGs and fungal 4 genomes from the different sources using a Wilcoxon rank sum test (wilcoxoauc function from the 5 presto R-package, v1.0.0). Predicted CAZyme and substrate target features were mapped to the 6 read count data, generating a CAZyme expression matrix. 7 8

9 Anaerobic fungal and bacterial enrichments efficiently degrade lignocellulose to VFAs 10 To establish different fungal and bacterial consortia capable of degrading crude lignocellulose into 11 VFAs, we first cultivated parallel anaerobic enrichment cultures on 1% milled sorghum in 12 chemically defined M2 media (see Methods). Cultures were inoculated anaerobically with either 13 cow or goat fecal pellets, or with sludge from an anaerobic digester in the presence of either 14 chloramphenicol (selecting for fungi) or 2-bromoethanesulfonate (BES, selecting for bacteria) and 15 passaged every 3-4 days over multiple generations (Figure 1 A). In total, inoculation of sorghum 16 media with cow and goat feces produced stable fungal and bacterial enrichments from each source, 17 whereas inoculation with anaerobic digester sludge only produced a bacterial enrichment, 18 suggesting the source digester had low (or no) anaerobic fungal abundance. 19 20 All enrichments were capable of sorghum deconstruction based on large increases in pressure 21 accumulation during 4 -day Hungate tube (i.e., batch) fermentations (Figure 1B). Fermentation 22 products produced by the fungal enrichments were consistent with a bacterial -like mixed acid 23 fermentation pathway previously reported for anaerobic fungi [25]. Both the cow and goat fungal 24 enrichments produced high yields of formate, acetate, lactate, and hydrogen gas (Figure 1C). In 25 comparison, the three bacterial enrichments from the cow, goat, and digester inocula produced 26 acetate, propionate, and butyrate (C2-C4 VFAs) as main fermentation products, as well as a small 27 amount of hydrogen gas (Figure 1C). These results are consistent with previous studies that 28 quantified VFA production during lignocellulose degradation in anaerobic bacterial fermentations 29 [5–7]. 30 31 Genome-centric metatranscriptomics reveals key microbes active across microbiomes 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 9 To identify key microbes responsible for lignocellulose degradation and conversion across the 1 fungal and bacterial enrichments we recovered genomes of individual species and profiled their 2 gene expression using metatranscriptomics. 3 4 A total of 92 MAGs were recovered across the cow, goat, and digester bacterial enrichments 5 affiliated with diverse bacteria belonging to the phyla Actinomycetota, Bacillota, Bacteroidota, 6 Desulfobacterota, Fusobacteriota, Pseudomonadota, and Spirochaetota, and Synergistota. A 7 summary of the genome statistics and taxonomy of these MAGs can be found in Supplementary 8 Table 1. Mapping of DNA and RNA reads back to these genomes revealed that MAGs affiliated 9 with the Lachnosporaceae and Bacteroidetes dominated in abundance and activity across all 10 bacterial enrichments (Figure 2A). In particular, several MAGs seemed to be especially important 11 to microbiome function based on their relative DNA and RNA abundance, including 12 Lachnosporaceae-1C (22% DNA; 30% RNA), Prevotella-1C (12%DNA; 8% RNA), and 13 Anaeroplasma-1C (2% DNA; 7% RNA) from the cow enrichment, Ruminococcaceae-1G (3% 14 DNA; 17% RNA), Lachnosporaceae-4G (12% DNA; 16% RNA), Lachnosporaceae-5G (3% 15 DNA; 8% RNA), Lachnosporaceae-7G (9% DNA; 5%RNA), Bacteroidales-1G (3% DNA; 11% 16 RNA), and Desulfovibrio-2G (1% DNA; 5% RNA) from the goat enrichment, and Bacteroides-17 3D (25% DNA; 27% RNA), Bacteroides-1D (25% DNA; 7% RNA), and Clostridium-1D (9% 18 DNA, 33% RNA) from the digester enrichment (Figure 2A). Interestingly, several of the 19 Bacteroides MAGs across the enrichments (e.g., Bacteroides -1C, Bacteroides -3G, and 20 Bacteroides-2G) had very low relative RNA abundance compared to their relative DNA abundance 21 (Figure 2A). This likely reflects temporal differences in activity across the 4-day batch 22 fermentation as RNA was collected on Day 3 , suggesting these species were more active during 23 the early stages of lignocellulose breakdown. 24 25 To identify active species from the fungal enrichments, RNA reads were mapped back to genomes 26 available at the time from seven isolated anaerobic fungi , including Anaeromyces robustus 27 (Anasp1), Neocallimastix californiae G1 (Neosp1), Neocallimastix frontalis var. giraffae 28 (NeoGfMa1), Neocallimastix lanati (Neolan1), Orpinomyces sp. (Orpsp1), Piromyces sp. E2 29 (PirE2), and Piromyces finnis (Pirfi3). A large fraction of RNA reads mapped successfully to these 30 fungal references (>95%), confirming substantial transcriptional activity from anaerobic fungi in 31 both cow and goat enrichments (Figure 2B ; Supplementary Data 1 ). However, due to extensive 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 10 sequence conservation and gene family expansion among anaerobic fungal genomes, many reads 1 mapped ambiguously across closely related taxa, limiting confident assignment to specific species 2 or genomes. Consequently, downstream analyses focused on functional profiles, especially CAZy 3 family expression, rather than species-resolved transcript abundances. At a higher taxonomic level, 4 read mapping patterns were broadly consistent with transcriptional activity from Neocallimastix-5 dominated enrichments (Figure 2B) , nevertheless high sequence similarity among genomes 6 precluded robust species-level resolution. 7 8 Carbohydrate active enzymes (CAZyme) expression profiles indicate substrate specialization 9 and niche partitioning during lignocellulose degradation 10 Sorghum biomass is predicted to consist of crystalline and amorphous cellulose that interacts with 11 xylan containing frequent and irregular arabinosyl substitutions. The monosaccharide composition 12 of non -cellulosic sorghum components in mole percent is approximately 63% xylose, 17% 13 glucose, 12% arabinose, 4% galactose, 3% galacturonic acid, and 1% glucuronic acid (Gao et al., 14 2020). This lignocellulosic matrix represents the primary carbon and energy substrates available 15 to the enrichment cultures and was expected to drive microbiome structure and function [26]. To 16 link the observed lignocellulose degradation activities to potential enzyme functions we analyzed 17 the expression of carbohydrate active enzymes (CAZymes) across all bacterial and fungal samples. 18 This revealed a wide range of expressed CAZymes, targeting diverse glycosidic linkages present 19 in cellulose, hemicellulose, starch, and pectin, especially cellulase (GH9), cellulose 1,4 -β-20 cellobiosidase (GH48/GH6), and β-glucosidase (GH3) involved in cellulose degradation and endo-21 1,4-β-xylanase (GH10/GH11), α-N-arabinofuranosidase (GH43), xylan 1,4-β-xylosidase (GH43), 22 and α -L-arabinofuranosidase (GH51) involved in hemicellulose degradation ( Supplementary 23 Table 2) that have previously been implicated in lignocellulose breakdown in the cow rumen [27]. 24 25 Across the rumen (cow and goat) and digester enrichments, bacterial MAGs affiliated with 26 Clostridia (Lachnospiraceae, Ruminococcus) and Bacteroidia (Bacteroides, Macellibacteroides, 27 Prevotella) were the main contributors to CAZyme expression (Figure 3), consistent with their 28 high abundance and activity based on read mapping (Figure 2A). In the digester bacterial 29 enrichment, Clostridium-1D was implicated as a key lignocellulose degrader with versatile 30 hydrolytic activities based on express ion of extracellular CAZymes involved in cellulose 31 hydrolysis (GH9), xylan backbone cleavage (GH 10/GH11), arabinoxylan debranching 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 11 (GH43/GH51), and ester bond hydrolysis between ferulic acid and arabinose moieties in 1 arabinoxylan (CE1) (Figure 3, Supplementary Table 3). As all these enzymes were extracellular, 2 byproducts of degradation (xylo-oligosaccharides, sugar monomers, etc.) likely became available 3 to other microbiome members. 4 5 In the goat bacterial enrichment, lignocellulose degradation activities appeared more distributed . 6 Diverse Ruminococcus MAGs (Ruminococcus-1G, Ruminococcus-2G, Ruminococcus-3G) in the 7 goat enrichment dominated CAZyme expression, including a wide array of CAZymes involved in 8 cellulose and arabinoxylan hydrolysis (Figure 3, Supplementary Table 3). The three Ruminococcus 9 MAGs expressed CAZymes with similar glycolsyl hydrolase motifs, such as GH9/GH48 (glucan) 10 and GH10/GH11 (arabinoxylan), but differed in the carbohydrate binding domains. This suggests 11 that in the goat enrichment, Rumminococcus MAGs were broadly competing for arabinoxylan but 12 could coexist through niche specialization which arose from differing modes of access to the 13 lignocellulosic biomass. For example , although all three Ruminococcus MAGs expressed 14 cellulases from the GH9 family, Ruminococcus-1G and Ruminococcus-3G expressed GH9 linked 15 to a CBM4, while Ruminococcus-2G expressed GH9 linked to a CBM37 , indicating that these 16 MAGs bind to differing sites on the glucan fraction of sorghum. 17 18 Lignocellulose degradation activities were expresedsed across an even broader set of taxa in the 19 cow bacterial enrichment, including Lachnospiraceae-1C, Prevotella-1C, and B acteroidales-2C. 20 Lachnospiraceae-1C and Prevotella-1C appeared to be the primary degraders of arabinoxylan, 21 while Cryptobacteroides-2C and Anaeroplasma-1C targeted pectin and starch, respectively. 22 Lachnospiraceae-1C, which had the highest expression of total and extracellular CAZymes , was 23 highly enriched for extracellular GH25 (lysozyme) which targets bacterial cell walls. The 24 simultaneous expression of GH25, and preference for arabinoxylan targets, by Lachnospiracea e-25 1C and Prevotella-1C suggests that the two MAGs were broadly competing for arabinoxylan but 26 may coexist through niche specialization conferred through diverse CBMs and debranching 27 activities. For example, Lachnospiraceae-1C and Prevotella-1C both express CAZymes from the 28 GH10 family, but GH10 CAZymes expressed by Prevotella-1C are further linked to bindin g 29 domains (CBM4, CBM 48) and esterases (CE1), which are not associated with the GH10 30 CAZymes from Lachnospiraceae-1C. 31 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 12 Across bacterial enrichments , several MAGs displayed high expression of multifunctional 1 enzymes consisting of several glycosyl hydrolases , glycosyl hydrolases with carbohydrate 2 esterases, or glycosyl hydrolyses with carbohydrate binding modules (Figure 3, Supplementary 3 Table 4). For example, Clostridium-1D had high expression of a CAZyme encoding two endo-β-4 1,4-xylanases (GH10/GH11) with a carbohydrate esterase (CE1) that may improve hydrolysis of 5 feruloylated arabinoxylan or oligosaccharides, whereas Ruminococcus-1G expressed CAZymes 6 encoding Endo-β-1,4-xylanase (GH10) with a α-L-arabinofuranosidase (GH43) as well as a 7 glucuronoarabinoxylan endo-1,4-beta-xylanase (GH30) with a carbohydrate esterase (CE1) that 8 also likely target branched arabinoxylans (xylan with arabinosyl and feruloyl substitutions). These 9 multifunctional enzymes were MAG specific, especially when comparing gene expression across 10 abundant Lachnospiraceae MAGs recovered from the different enrichments that encoded widely 11 varying lignocellulose hydrolysis capabilities, with Clostridium-1D appearing to be the most 12 robust degrader. Take n together, these results highlight different degradation strategies between 13 bacterial enrichments sourced from diverse inocula (digester sludge, cow feces, goat feces) 14 reflective of niche partitioning and substrate specialization. 15 16 Anaerobic fungal enrichments from the cow and goat samples also had high expression of diverse 17 CAZymes. The repertoire of expressed CAZymes in the goat and cow enrichments were highly 18 similar and primarily targeted arabinoxylan and glucan fractions of the lignocellulosic biomass. 19 This suggested that, although different fungi dominated, the cow and goat fungal enrichments 20 performed similar biochemical functions shaped by sorghum as a substrate. Compared to bacterial 21 enrichments from the same source, the fungal enrichments consistently had a lower diversity of 22 CAZymes, especially glycosyl hydrolases (GHs) (Figure 4) . However, when only consider ing 23 bacterial transcripts predicted to be extracellular, the diversity of GHs expressed by bacteria or 24 fungi became more comparable . Comparing fungal CAZymes and the extracellular fraction of 25 bacterial CAZymes a cross cow and goat sources, the fungal enrichments expressed 38 GHs not 26 expressed by in the bacterial enrichments, including the exo-type cellobiohydrolase, GH6 (Figure 27 4; Supplementary Table 4). Conversely, the bacterial enrichments expressed 11 GHs not expressed 28 by the fungal enrichments, such GH51_2, an α -L-arabinofuranosidase involved in hemicellulose 29 degradation (Figure 4; Supplementary Table 4) . This may reflect different degradation strategies 30 between anaerobic fungi and bacteria that could be complementary, as both are well known to co-31 exist in the rumen. 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 13 1 Across all bacterial and fungal enrichments, genes encoding canonical lignin -depolymerizing 2 enzymes, including lignin peroxidases, manganese peroxidases, and β-etherases, were not 3 detected. Instead, transcripts encoding carbohydrate esterases (CE1) were detected across several 4 abundant bacterial MAGs and anaerobic fungal genomes (Figure 3, Supplementary Data 2 and 3), 5 consistent with the potential to hydrolyze feruloyl- and p-coumaroyl ester linkages associated with 6 sorghum arabinoxylan. Canonical pathways for the subsequent catabolism of released ferulic 7 and/or p-coumaric acids (e.g., feruloyl-CoA synthetase, vanillate O-demethylase, or benzoyl-CoA 8 degradation pathways) were not detected. However, phenylacetate-CoA ligase genes were across 9 multiple bacterial MAGs, and near-complete phenylacetate catabolic pathways were detected in a 10 small number of low -abundance MAGs (Enterobacteriaceae-1D, S higella-1C; Supplementary 11 Data 4). Together, these results indicate that lignin was not extensively depolymerized under the 12 enrichment conditions, but instead functionally circumvented through uncoupling of lignin -13 carbohydrate complexes. Importantly, this analysis does not preclude lignin modification via non-14 canonical or uncharacterized anaerobic mechanisms, as previously demonstrated by direct 15 structural analyses of lignin during anaerobic fungal growth [42]. 16 17 Metatranscriptomic analysis reveals key fermentation pathways and microbes involved in VFA 18 production from lignocellulosic sugars 19 To examine the metabolic networks of different fungal and bacterial consortia degrading 20 lignocellulose into VFAs we reconstructed anaerobic fermentation and respiration pathways across 21 the recovered genomes and analyzed their gene expression. MAGs were assigned to different 22 functional guilds based on gene expression of key enzymes and metabolic pathway responsible for 23 specific substrate utilization and expected fermentation/respiration products. This analysis 24 revealed five major guilds across the bacterial enrichments , including hydrolytic propionate-25 producing bacteria, hydrolytic butyrate-producing bacteria, hydrolytic alcohol-producing bacteria, 26 succinate/lactate producing bacteria, and sulfate-reducing bacteria. 27 28 The most abundant and active MAGs were hydrolytic bacteria involved in complex polysaccharide 29 fermentation to propionate or butyrate, together with acetate and H 2 formation. Propionate 30 production was dominated by MAGs affiliated with the order Bacteroidales, including Prevotella-31 1C and Prevotella-2C from the cow enrichment, Bacteroides-2G, Bacteroidales-1G, Bacteroides-32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 14 1G, and Prevotella-1G from the goat enrichment, and Bacteroides-3D, Bacteroides_1D, and 1 Macellibacteroides-1D from the digester enrichment (Figure 5; Supplementary Dataset 4). These 2 MAGs all expressed enzymes of the methylmalonyl-CoA pathway (Supplementary Dataset) , 3 including a sodium ion pumping methylmalonyl-CoA decarboxylase [28], that couples propionate 4 formation with the generation of an ion motive force that can be used for ATP synthesis. These 5 MAGs also had high expression of pathways for sugar utilization (glucose, xylose, and arabinose) 6 (Figure 5; Supplementary Dataset 1) consistent with their high expression of CAZymes involved 7 in liberating these sugars during cellulose and hemicellulose degradation (Figure 3). 8 9 Butyrate production across the enrichments was dominated by MAGs affiliated with the 10 Lachnospiraceae (Figure 5). All Lachnospiraceae MAGs had high gene expression of the reverse 11 beta-oxidation pathway, including an electron bifurcating acyl-CoA dehydrogenase complex [29]. 12 They also expressed the terminal enzymes phosphate butyryltransferase and butyrate kinase that 13 couples butyrate synthesis with substrate level phosphorylation, except for Clostridium-1D, which 14 instead expressed a butyryl-CoA:acetate CoA transferase [30]. While all Lachnospiraceae MAGs 15 expressed diverse CAZymes (Figure 3) and sugar utilization pathways (Figure 5), Clostridium-1D 16 from the digester enrichment especially stood out among the Lachnospiraceae MAGs as a 17 specialized cellulose and hemicellulose degrad er based on its high copy number and gene 18 expression of 1,4-beta-xylanases, α-L-arabinofuranosidases, and cellulases (Figure 3). 19 20 In addition to propionate and butyrate, s everal MAGs affiliated with the Ruminococcus and 21 Sphaerochaetaceae were predicted to be involved in complex polysaccharide fermentation to 22 ethanol, acetate, and H2 formation. Ruminococcus-1G and Ruminococcus-2G had high expression 23 of a bifunctional acetaldehyde -CoA/alcohol dehydrogenase and NAD-dependent electron -24 bifurcating [FeFe] -hydrogenase (HydABC) , together with CAZymes and substrate utilization 25 pathways involved in both cellulose and hemicellulose degradation (Figures 5; Supplementary 26 Dataset 1 ). This metabolism is consistent with previous batch fermentation results from other 27 Ruminococcus species, such as Ruminococcus albus isolated from the cow rumen [31, 32]. Diverse 28 Sphaerochaetaceae MAGs, including Treponema-1C, Sphaerochaetaceae-1C, Sphaerochaeta-1D, 29 and Sphaerochaetaceae-1C were also implicated in ethanol, acetate, formate, and H 2 formation 30 based on the high gene expression of alcohol dehydrogenase s (iron-containing alcohol 31 dehydrogenase and/or bifunctional acetaldehyde-CoA/alcohol dehydrogenase) and hydrogenases 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 15 (iron-only hydrogenases and/or ferredoxin hydrogenases) (Figure 5; Supplementary Data set 1). 1 Additionally, Sphaerochaetaceae-1G also had high expression of lactate dehydrogenase and 2 hydroxyacid dehydrogenase suggesting they may also produce lactate as an end-product. 3 4 Aside from fermentation pathways , all enrichments contained bacteria affiliated with 5 Desulfovibrio sp. predicted to be capable of anaerobic sulfate respiration to hydrogen sulfide 6 coupled to ethanol, formate, lactate, and H 2 oxidation. In particular, Desulfovibrio-1C, 7 Desulfovibrio-1G, Desulfovibrio-2G, and Nitratidesulfovibrio-1D had high expression of sulfate 8 adenylyltransferase, adenylylsulfate reductase , and dissimilatory sulfite reductase involved in 9 dissimilatory sulfate respiration together with high expression of hydrogenases, formate 10 dehydrogenase, lactate dehydrogenase, alcohol dehydrogenase and aldehyde dehydrogenase 11 (Figure 5; Supplementary Dataset 1), suggesting co-utilization of ethanol, formate, lactate, and H2 12 for energy generation. This is consistent with the availability of these substrates as fermentation 13 by-products (Figure 1). 14 15

16 The use of genome -centric metatranscriptomics allowed us to analyze the metabolism and gene 17 expression of 5 different anaerobic enrichment cultures degrading lignocellulosic biomass to 18 VFAs. Our results reveal that across all bacterial enrichments, MAGs affiliated with the order 19 Bacteroidales (Bacteroides and Prevotella) and family Lachnospiraceae were dominant 20 lignocellulose degraders that ferment ed biomass-derived sugars to propi onate and butyrate, 21 respectively, together with acetate and H 2 formation. Other MAGs affiliated with Ruminococcus 22 from the goat bacterial enrichment, and to a lesser extent the family Sphaerochaetaceae across all 23 bacterial enrichments, were also implicated in biomass breakdown to ethanol, lactate, formate, 24 acetate, and H 2 (Ruminococcus, Sphaerochaetaceae), consistent with the metabolism of closely 25 related representative isolates [33, 34]. The resulting fermentation products, especially ethanol, 26 formate, lactate, and H 2 were predicted to be substrates for Desulfovibrio MAGs that performed 27 dissimilatory sulfate reduction across all bacterial enrichments . In the fungal enrichments, 28 lignocellulose degradation was dominated by Neocallimastix sp. that employed a mixed acid 29 fermentation pathway, producing lactate, acetate, ethanol, H 2, and CO 2 as major end products , 30 consistent with the metabolism of representative isolates [20]. Taken together, these results 31 highlight that substrate (sorghum biomass) and chemical inhibitors (chloramphenicol and BES) 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 16 selected for microbiomes with similar overall functional guild structure, despite containing 1 different taxa and starting from disparate inocula. 2 3 Detailed analysis of CAZyme gene expression across the enrichments revealed common 4 biodegradation strategies between diverse bacteria and fungi while also highlighting potential 5 complementarities and fine -scale functional specialization . While all bacterial and fungal 6 enrichments expressed CAZymes necessary for cellulose and hemicellulose degradation, including 7 those involved in cellulose hydrolysis (GH9), arabinoxylan backbone cleavage (GH10/GH11), and 8 arabinoxylan debranching (GH43/GH51/CE1), fine-scale differences between microbes based on 9 multi-functional enzymes linking two or more CAZymes together with diverse CBMs were 10 widespread. Most of these multi-functional enzymes localized complementary degradation 11 functions together, such as xylan backbone cleavage with various debranching activities, likely 12 conferring synergistic degradation effects as shown by purified enzyme studies [35]. The 13 association of these enzymes with diverse CBMs also suggests targeted specialization towards 14 specific lignocellulose structures (e.g., substituted backbones or branched linkages), consistent 15 with recent observations among human gut microbiota [36]. These results implicit fine -scale 16 substrate specialization as a major driver of microbiome assembly, niche partitioning, and species 17 co-existence during anaerobic lignocellulose degradation. 18 19 Hydrolysis and subsequent fermentation of lignocellulosic substrates resulted in a range of 20 fermentation acids shaped by culture conditions. Indeed, it has been shown that the spectrum of 21 fermentation products is highly influenced by both substrate and environmental parameters, such 22 as pH and temperature. In our study , a neutral pH of 7 -7.5 and temperature of 39 °C was 23 maintained, selecting for bacterial enrichments that produced short-chain fatty acids (acetate, 24 propionate, butyrate) and H2 as major fermentation products, and fungal enrichments that produced 25 mixed acid fermentation products (lactate, acetate, formate) and H2. These acids appeared to be 26 directly produced from lignocellulose -derived sugars versus indirectly via lactate as an 27 intermediate, which can result via cross feeding between lactate -producing species (e.g., LAB, 28 Bifidobacteria) and lactate -utilizing species. This may have resulted from having neutral versus 29 mildly acid pH ( ~5-6) conditions, the latter of which has been shown to select for lactate cross-30 feeding. However, general principles on what controls the distribution of fermentation products 31 remains unclear and represents an important knowledge gap to advance anaerobic fermentation as 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 17 a platform for biorefining of specific volatile fatty acids, lactate, alcohols, and other bulk chemicals 1 [37]. 2 3 While further studies are still needed to comprehensively map the functional niches driving 4 anaerobic lignocellulose degradation to VFAs, our results provide important insights on fine-scale 5 substrate specialization and metabolic pathways structuring microbiome assembly and function . 6 We anticipate that these findings will help inform efforts to develop synthetic microbiomes with 7 tailored functionality for low-cost conversion of lignocellulosic biomass to fuels/chemicals [38]. 8 This will require isolating and assembling novel anaerobes , such as those identified here, into 9 defined consortia to eliminate competing metabolic pathways , as well as introducing novel 10 functionalities via metabolic engineering [39]. Such consortia would also serve to elucidate basic 11 principles governing microbiome assembly and function, resulting in robust anaerobic 12 fermentation platforms for sustainable biomanufacturing. 13 14 Conflicts of Interest 15 The authors declare no conflicts of interest. 16 17

18 This research was sponsored by the U.S. Department of Energy, Office of Science through grant 19 DE-SC0022142. The work conducted at the Joint BioEnergy Institute was supported by the U.S. 20 Department of Energy, Office of Science, Biological and Environmental Research Program, 21 through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the 22 U.S. Department of Energy. The United States Government retains and the publisher, by accepting 23 the article for publication, acknowledges that the United States Government retains a 24 nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form 25 of this manuscript, or allow others to do so, for United States Government purposes. Any 26 subjective views or opinions that might be expressed in the paper do not necessarily represent the 27 views of the U.S. Department of Energy or the United States Government. 28 29 Data Availability 30 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 18 Raw DNA and cDNA read data and annotated metagenome assembled genomes (MAGs) can be 1 found on the National Center for Biotechnology Information (NCBI) website under BioProject 2 accession no. PRJNA1379681. 3 4 Figure Legends 5 Figure 1 – Fungal and bacterial enrichment cultures converting lignocellulosic biomass to VFAs. 6 A) Enrichment of anaerobic fungal and bacterial consortia from the cow and goat rumen and an 7 anaerobic digester. B) Accumulated gas production (psig) after 4 days and C) VFA, lactate, and 8 H2 production by anaerobic fungal and bacterial enrichments sourced from the cow and goat rumen 9 and an anaerobic digester . VFA and lactate concentrations reported in mg/L, H 2 concentration 10 reported in nanomolar. H2 concentrations for fungal and bacterial enrichments shown on left and 11 right y-axes, respectively. 12 13 Figure 2 - Abundance and activity of bacterial MAGs and fungal isolate genomes recovered from 14 enrichments. Percentage of raw metagenomic (DNA - red) and metatranscriptomic (RNA - pink) 15 reads mapped to recovered MAGs (A) and existing fungal isolate genomes (B). Fungal isolate 16 genomes labels: Anasp1 - Anaeromyces robustus , Neosp - Neocallimastix californiae G1 , 17 NeoGfMa1 - Neocallimastix frontalis var. giraffae , Neolan1 - Neocallimastix lanati , Orpsp1 - 18 Orpinomyces sp., PirE2 - Piromyces sp. E2, Pirfi3 - Piromyces finnis. 19 20 Figure 3 - CAZyme expression across bacterial MAGs from (A) digester enrichment, (B) cow 21 enrichment, (C) goat enrichment. Left panel – RNA expression (transcript per million, TPM) of 22 annotated CAZymes based on predicted subcellular localization using PSORTb v3.0. Right panel 23 – log2 TPM expression across CAZyme motifs per MAG. Target substrate of CAZyme motif 24 (bottom row) predicted using dbCAN3. 25 26 Figure 4 – CAZyme expression across fungal reference genomes and comparison to bacterial 27 MAGs. (A) RNA expression (log 2 TPM) of CAZyme motifs across abundant fungal isolate 28 genomes. Target substrate of CAZyme motif (bottom row) predicted using dbCAN3. (B) Number 29 of motifs annotated per CAZyme family between bacterial MAGs and fungal isolate genomes. AA 30 - Auxiliary Activities, CBM – Carbohydrate Binding Motifs, CE - Carbohydrate Esterases, GH - 31 Glycoside Hydrolases, GT - GlycosylTransferases, PL - Polysaccharide Lyases. 32 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 19 1 Figure 5 - Expression of key metabolic pathways across MAGs . Rows indicate RNA expression 2 (log2[TPM]) of MAG open reading frames annotated as MetaCyc reactions across key metabolic 3 pathways via gapseq. Columns indicate bacterial MAGs from digester enrichment (left), cow 4 enrichment (middle), and goat enrichment (right). 5 6

7 1. Sawatdeenarunat C, Surendra KC, Takara D, Oechsner H, Khanal SK. Anaerobic 8 digestion of lignocellulosic biomass: challenges and opportunities. Bioresour Technol 9 2015; 178: 178–186. 10 2. Huq NA, Hafenstine GR, Huo X, Nguyen H, Tifft SM, Conklin DR, et al. Toward net-11 zero sustainable aviation fuel with wet waste-derived volatile fatty acids. Proceedings of 12 the National Academy of Sciences 2021; 118: e2023008118. 13 3. Agnihotri S, Yin D-M, Mahboubi A, Sapmaz T, Varjani S, Qiao W, et al. A Glimpse of 14 the World of Volatile Fatty Acids Production and Application: A review. Bioengineered 15 2022; 13: 1249–1275. 16 4. Stamatopoulou P, Malkowski J, Conrado L, Brown K, Scarborough M. Fermentation of 17 Organic Residues to Beneficial Chemicals: A Review of Medium-Chain Fatty Acid 18 Production. Processes 2020; 8. 19 5. Weimer PJ, Nerdahl M, Brandl DJ. Production of medium-chain volatile fatty acids by 20 mixed ruminal microorganisms is enhanced by ethanol in co-culture with Clostridium 21 kluyveri. Bioresour Technol 2015; 175: 97–101. 22 6. Agematu H, Takahashi T, Hamano Y. Continuous volatile fatty acid production from 23 lignocellulosic biomass by a novel rumen-mimetic bioprocess. J Biosci Bioeng 2017; 24 124: 528–533. 25 7. Fonoll X, Shrestha S, Khanal SK, Dosta J, Mata-Alvarez J, Raskin L. Understanding the 26 Anaerobic Digestibility of Lignocellulosic Substrates Using Rumen Content as a 27 Cosubstrate and an Inoculum. ACS ES&T Engineering 2021; 1: 424–435. 28 8. Hagen LH, Brooke CG, Shaw CA, Norbeck AD, Piao H, Arntzen MØ, et al. Proteome 29 specialization of anaerobic fungi during ruminal degradation of recalcitrant plant fiber. 30 ISME J 2021; 15: 421–434. 31 9. Teunissen MJ, den Camp HJM, Orpin CG, in ‘t Veld JHJ, Vogels GD. Comparison of 32 growth characteristics of anaerobic fungi isolated from ruminant and non-ruminant 33 herbivores during cultivation in a defined medium. Microbiology (N Y) 1991; 137: 1401–34 1408. 35 10. Haitjema CH, Solomon K V, Henske JK, Theodorou MK, O’Malley MA. Anaerobic gut 36 fungi: Advances in isolation, culture, and cellulolytic enzyme discovery for biofuel 37 production. Biotechnol Bioeng 2014; 111: 1471–1482. 38 11. Nurk S, Meleshko D, Korobeynikov A, Pevzner PA. metaSPAdes : a new versatile 39 metagenomic assembler. Genome Res 2017; 27: 824–834. 40 12. Wu Y-W, Tang Y-H, Tringe SG, Simmons B a, Singer SW. MaxBin: an automated 41 binning method to recover individual genomes from metagenomes using an expectation-42 maximization algorithm. Microbiome 2014; 2: 26. 43 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 20 13. Kang DD, Froula J, Egan R, Wang Z. MetaBAT, an efficient tool for accurately 1 reconstructing single genomes from complex microbial communities. PeerJ 2015; 3: 2 e1165. 3 14. Alneberg J, Bjarnason BS, de Bruijn I, Schirmer M, Quick J, Ijaz UZ, et al. Binning 4 metagenomic contigs by coverage and composition. Nat Methods 2014; 11: 1144–1146. 5 15. Sieber CMK, Probst AJ, Sharrar A, Thomas BC, Hess M, Tringe SG, et al. Recovery of 6 genomes from metagenomes via a dereplication, aggregation and scoring strategy. Nat 7 Microbiol 2018; 3: 836–843. 8 16. Chaumeil P-A, Mussig AJ, Hugenholtz P, Parks DH. GTDB-Tk: a toolkit to classify 9 genomes with the Genome Taxonomy Database. Bioinformatics 2020; 36: 1925–1927. 10 17. Yu NY, Wagner JR, Laird MR, Melli G, Rey S, Lo R, et al. PSORTb 3.0: improved 11 protein subcellular localization prediction with refined localization subcategories and 12 predictive capabilities for all prokaryotes. Bioinformatics 2010; 26: 1608–1615. 13 18. Parks DH, Imelfort M, Skennerton CT, Hugenholtz P, Tyson GW. CheckM: assessing the 14 quality of microbial genomes recovered from isolates, single cells, and metagenomes. 15 Genome Res 2015; 25: 1043–55. 16 19. Konwar KM, Hanson NW, Bhatia MP, Kim D, Wu SJ, Hahn AS, et al. MetaPathways 17 v2.5: Quantitative functional, taxonomic and usability improvements. Bioinformatics 18 2015; 31: 3345–3347. 19 20. Elmo W St., M. MJ, A. LP, E. LC, S. LT, Susanna S, et al. Experimentally Validated 20 Reconstruction and Analysis of a Genome-Scale Metabolic Model of an Anaerobic 21 Neocallimastigomycota Fungus. mSystems 2021; 6: e00002-21. 22 21. Haitjema CH, Gilmore SP, Henske JK, Solomon K V., De Groot R, Kuo A, et al. A parts 23 list for fungal cellulosomes revealed by comparative genomics. Nat Microbiol 2017; 2: 1–24 8. 25 22. H. YN, B. CM, G. SC, S. LA, A. PR, Z. NF, et al. The Genome of the Anaerobic Fungus 26 Orpinomyces sp. Strain C1A Reveals the Unique Evolutionary History of a Remarkable 27 Plant Biomass Degrader. Appl Environ Microbiol 2013; 79: 4620–4634. 28 23. Zheng J, Ge Q, Yan Y, Zhang X, Huang L, Yin Y. dbCAN3: automated carbohydrate-29 active enzyme and substrate annotation. Nucleic Acids Res 2023; 51: W115–W121. 30 24. Kopylova E, Noé L, Touzet H. SortMeRNA: Fast and accurate filtering of ribosomal 31 RNAs in metatranscriptomic data. Bioinformatics 2012; 28: 3211–3217. 32 25. Boxma B, Voncken F, Jannink S, Van Alen T, Akhmanova A, Van Weelden SWH, et al. 33 The anaerobic chytridiomycete fungus Piromyces sp. E2 produces ethanol via 34 pyruvate:formate lyase and an alcohol dehydrogenase E. Mol Microbiol 2004; 51: 1389–35 1399. 36 26. Lindemann SR. A piece of the pie: engineering microbiomes by exploiting division of 37 labor in complex polysaccharide consumption. Curr Opin Chem Eng 2020; 30: 96–102. 38 27. Comtet-Marre S, Parisot N, Lepercq P, Chaucheyras-Durand F, Mosoni P, Peyretaillade 39 E, et al. Metatranscriptomics Reveals the Active Bacterial and Eukaryotic Fibrolytic 40 Communities in the Rumen of Dairy Cow Fed a Mixed Diet . Frontiers in Microbiology . 41 2017. , 8 42 28. Hilpert W, Dimroth P. Conversion of the chemical energy of methylmalonyl-CoA 43 decarboxylation into a Na+gradient. Nature 1982; 296: 584–585. 44 29. Fuli L, Julia H, Henning S, Jin Z, Wolfgang B, K. TR. Coupled Ferredoxin and Crotonyl 45 Coenzyme A (CoA) Reduction with NADH Catalyzed by the Butyryl-CoA 46 Dehydrogenase/Etf Complex from Clostridium kluyveri. J Bacteriol 2008; 190: 843–850. 47 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 21 30. Louis P, Young P, Holtrop G, Flint HJ. Diversity of human colonic butyrate-producing 1 bacteria revealed by analysis of the butyryl-CoA:acetate CoA-transferase gene. Environ 2 Microbiol 2010; 12: 304–314. 3 31. Zheng Y, Kahnt J, Kwon IH, Mackie RI, Thauer RK. Hydrogen formation and its 4 regulation in Ruminococcus albus: involvement of an electron-bifurcating [FeFe]-5 hydrogenase, of a non-electron-bifurcating [FeFe]-hydrogenase, and of a putative 6 hydrogen-sensing [FeFe]-hydrogenase. J Bacteriol 2014; 196: 3840–3852. 7 32. Iannotti EL, Kafkewitz D, Wolin MJ, Bryant MP. Glucose fermentation products in 8 Ruminococcus albus grown in continuous culture with Vibrio succinogenes: changes 9 caused by interspecies transfer of H 2 . J Bacteriol 1973; 114: 1231–1240. 10 33. Paster BJ, Canale-Parola E. Physiological diversity of rumen spirochetes. Appl Environ 11 Microbiol 1982; 43: 686–693. 12 34. Paster BJ, Canale-Parola E. Treponema saccharophilum sp. nov., a large pectinolytic 13 spirochete from the bovine rumen. Appl Environ Microbiol 1985; 50: 212–219. 14 35. Sørensen HR, Jørgensen CT, Hansen CH, Jørgensen CI, Pedersen S, Meyer AS. A novel 15 GH43 α-l-arabinofuranosidase from Humicola insolens: mode of action and synergy with 16 GH51 α-l-arabinofuranosidases on wheat arabinoxylan. Appl Microbiol Biotechnol 2006; 17 73: 850–861. 18 36. Yao T, Deemer DG, Chen M-H, Reuhs BL, Hamaker BR, Lindemann SR. Differences in 19 fine arabinoxylan structures govern microbial selection and competition among human gut 20 microbiota. Carbohydr Polym 2023; 316: 121039. 21 37. Regueira A, Rombouts JuliusL, Wahl SA, Mauricio-Iglesias M, Lema JM, Kleerebezem 22 R. Resource allocation explains lactic acid production in mixed-culture anaerobic 23 fermentations. Biotechnol Bioeng 2021; 118: 745–758. 24 38. Lynd LR, Beckham GT, Guss AM, Jayakody LN, Karp EM, Maranas C, et al. Toward 25 low-cost biological and hybrid biological/catalytic conversion of cellulosic biomass to 26 fuels. Energy Environ Sci 2022; 15: 938–990. 27 39. Lawson CE. Retooling Microbiome Engineering for a Sustainable Future. mSystems 2021; 28 6: 10.1128/msystems.00925-21. 29 40. Caspi R, Billington R, Keseler IM, Kothari A, Krummenacker M, Midford PE, Ong WK, 30 Paley S, Subhraveti P, Karp PD. The MetaCyc database of metabolic pathways and 31 enzymes: a 2019 update. Nucleic Acids Res 2020; 48: D445–D453 32 41. Zimmermann J, Kaleta C, Waschina S. gapseq: informed prediction of bacterial metabolic 33 pathways and reconstruction of accurate metabolic models. Genome Biol 2021; 22: 81. 34 42. Lankiewicz, T.S., Choudhary, H., Gao, Y. et al. Lignin deconstruction by anaerobic fungi. 35 Nat Microbiol 2023 8, 596–610. https://doi.org/10.1038/s41564-023-01336-8 36 37 38 39 40 41 42 43 44 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 22 1 Figure 1. Fungal and bacterial enrichment cultures converting lignocellulosic biomass to VFAs. 2 3 4 5 6 7 0 200 400 600 800 0 1 2 3 4 Butyrate 0 4000 8000 12000 0 1 2 3 4 H2 0 500 1000 1500 0 1 2 3 4 Days Lactate 0 300 600 900 0 1 2 3 4 Conc. (mg/L) Formate 300 600 900 1200 0 1 2 3 4 Acetate 0 200 400 600 0 1 2 3 4 Propionate 0 400 800 1200 nM Sample Cow+BES Digester+BES Goat+BES Cow+Ch Goat+Ch control 0 3 6 9 12 Control Cow +BES Goat +BES Digester +BES Cow +Ch Goat +Ch Pressure Accumulation (psig) Fungal Enrichments Bacterial Enrichments Cow Goat No fungi from digester Cow Goat Digester A B C After 4 Days .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 23 1 Figure 2. Abundance and activity of bacterial MAGs and fungal isolate genomes recovered from 2 enrichments 3 Goat Fungi Cow Fungi 0 20 40 60 80 0 20 40 60 Neolan1 NeoGfMa1 Neospfc1 Neosp1 Orpspfg1 Orpsp1 PirE2 Anasp1 Pirfi3 Neolan1 NeoGfMa1 Neospfc1 Neosp1 Orpspfg1 Orpsp1 PirE2 Anasp1 Pirfi3 % Reads Mapped DNA RNA Digester Bacteria Goat Bacteria Cow Bacteria 0 10 20 30 0 5 10 15 20 0 10 20 30 40 Desulfovibrio_1_D Thomasclavelia_1_D Clostridium_2_D Nitratidesulfovibrio_1_D Fusobacterium_1_D Streptococcus_1_D Escherichia_1_D Lacrimispora_1_D Eggerthella_1_D Phascolarctobacterium_1_D Enterococcus_1_D Oscillibacter_1_D Kineothrix_1_D Aminobacterium_1_D Bacteroides_2_D Enterocloster_1_D Dethiosulfovibrio_1_D Sphaerochaeta_1_D Avimicrobium_1_D Scatacola_1_D Macellibacteroides_1_D Bacteroides_1_D Bacteroides_3_D Clostridium_1_D Escherichia_1_G Lachnospiraceae_3_G Streptococcus_1_G Bacilli_1_G Phascolarctobacterium_1_G Fimenecus_1_G Cryptobacteroides_2_G Ventricola_1_G Bacteroides_1_G Bacteroidaceae_1_G Cryptobacteroides_1_G Slackia_1_G Enteromonas_1_G Bacteroides_3_G Bacteroides_2_G Ruminococcaceae_4_G Sphaerochaetaceae_1_G Desulfovibrio_1_G Ruminococcaceae_3_G Anaeroplasma_1_G Pyramidobacter_1_G Lachnospiraceae_1_G Lachnospiraceae_2_G Prevotella_1_G Lachnospiraceae_6_G Ruminococcaceae_2_G Lachnospiraceae_7_G Desulfovibrio_2_G Lachnospiraceae_5_G Bacteroidales_1_G Lachnospiraceae_4_G Ruminococcaceae_1_G Succinivibrio_1_C Enterococcus_1_C Avilachnospira_1_C Escherichia_1_C Phocaeicola_1_C Christensenellales_1_C Eubacterium_1_C Synergistes_1_C Lachnospiraceae_5_C Streptococcus_1_C Cryptobacteroides_1_C Bianqueaceae_1_C Bacteroidaceae_1_C Faecalicoccus_1_C Butyricicoccaceae_1_C Bacteroides_1_C Bacteroidales_1_C Butyrivibrio_1_C Oscillospiraceae_1_C Prevotella_2_C Lachnospiraceae_6_C Enterocloster_1_C Bacilli_1_C Sphaerochaetaceae_1_C Lachnospiraceae_2_C Cryptobacteroides_2_C Desulfovibrio_1_C Lachnospiraceae_4_C Lachnospiraceae_3_C Treponema_1_C Prevotella_1_C Anaeroplasma_1_C Lachnospiraceae_1_C .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 24 1 Figure 3. CAZyme expression across bacterial MAGs from (A) digester enrichment, (B) cow 2 enrichment, (C) goat enrichment. 3 4 5 6 7 8 9 10 11 12 13 Sphaerochaeta_1_D Kineothrix_1_D Avimicrobium_1_D Bacteroides_1_D Bacteroides_3_D Clostridium_1_D 0 10 20 30 TPM x 103 MAG Extracellular Cell wall OM CM Periplasmic Cytoplasmic Unknown 2.5 5.0 7.5 10.0 12.5 log2(TPM + 1) CBM13;CBM2;GH11CBM13;CBM2;GH10 CBM3;GH9 CBM13;CBM2;CBM6;GH43_16 CBM13;CBM2;CE1;GH10 CBM36;GH11CBM3;GH48 CBM6;CE1;GH10;GH11 CBM2;GH5_4 CBM36;CE4;GH11 CBM48;CE1 CE1 GH5_2GH18GT51 CE8 GH16_3;GH43_24 GH28GH16_3 CBM6;GH43_2;GH8 CE12;CE8GH30_3GH25GH51_2 CBM48;CBM50;CBM83;GH13_41 CBM4;GH16_3 GH5_46 CAZyme Motif Target arabinoxylan chitin galactan glucan mucin pectin peptidoglycan starch Lachnospiraceae_4_C Lachnospiraceae_2_C Lachnospiraceae_3_C Treponema_1_C Prevotella_1_C Anaeroplasma_1_C Cryptobacteroides_2_C Lachnospiraceae_1_C 0 2 4 TPM x 103 MAG Extracellular Cell wall OM CM Periplasmic Cytoplasmic Unknown 2 4 6 8 log2(TPM + 1) GH25CE4GH10 CBM34;GH13_39 GH5_37 CBM61;GH53 GH43_26 GH28 CE12;CE8;PL1 CE8;GH95GH16_3 CE8 GH106;GH28 CBM56 CBM26;GH13_32CBM26;GH13_28 GH13_28GH51_2 CBM4;GH10 CBM48;CE1;GH10 GH5_38 PL1 CBM6;GH43_29 GH13_46 CBM50;GH73 GH5_46 CBM4;GH16_3 GH13_4CE12;CE8 PL1_6PL9_1GH5_2GH136 CBM5;GH18 GT51GH9GH73 CAZyme Motif Target arabinoxylan chitin galactan glucan mucin pectin peptidoglycan starch Prevotella_1_G Lachnospiraceae_7_G Ruminococcaceae_3_G Lachnospiraceae_5_G Bacteroidales_1_G Ruminococcaceae_2_G Lachnospiraceae_4_G Ruminococcaceae_1_G 0 5 10 TPM x 103 MAG Extracellular Cell wall OM CM Periplasmic Cytoplasmic Unknown 3 6 9 log2(TPM + 1) CBM4;GH9 GH48 CBM22;GH10;GH11 GH9 CBM3;GH9CBM79;GH9 GH25 CBM22;CBM6;GH10;GH43_16 CBM22;CE1 CBM22;CE1;GH30_8 GH28GH5_39 CE1CE8 CBM37;GH9 CBM22;CBM37;GH11 CBM37 CBM3;CBM37;GH9CBM22;CBM37;GH10 CBM22;GH30_8 CBM37;CBM77;PL1 CBM37;PL11 CBM22;CE4;GH11 CBM13;CBM35;CE12 CBM50GH13_36CE4;GH9 CBM50;GH23 GH73GH5_2 CBM22;GH10 GH11CBM3 GH10;GH11CBM35;GH97CBM22;GH11 GH10 CBM30;GH9 GH146CBM6CBM4PL9PL1_2 CE12;CE8 PL1PL4_4 CAZyme Motif Target arabinoxylan chitin galactan glucan mucin pectin peptidoglycan starch .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 25 1 Figure 4. CAZyme expression across fungal reference genomes and comparison to bacterial 2 MAGs. 3 4 5 6 7 8 9 10 11 12 13 NeoGfMa1 cow Neolan1 goat 0 20 40 60 TPM x 103 Genome 10.5 11.0 11.5 12.0 12.5 13.0 log2(TPM + 1) CBM10;GH6 CE4 GH11 GH43_1 GH48 CE1 GH1 GH10 GH9 GH45 CE1;GH11CE1;CBM10 CAZyme Motif Target arabinoxylan glucan other 0 50 100 150 200 AA CBM CE GH GT PL CAZyme Family Number of Motifs 0 50 100 150 AA CBM CE GH GT PL CAZyme Family Number of Motifs 0 20 40 60 AA CBM CE GH GT PL CAZyme Family Number of Motifs Kingdom Bacteria Fungi 0 20 40 60 AA CBM CE GH GT PL CAZyme Family Number of Motifs Kingdom Bacteria Fungi Cow Cow Goat Goat All ORFs Extracellular ORFs A B .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 26 1 Figure 5. Expression of key metabolic pathways across MAGs . Rows indicate RNA expression 2 (log2[TPM]) of MAG open reading frames annotated as MetaCyc reactions across key metabolic 3 pathways via gapseq. 4 5 6 7 8 9 10 11 12 13 14 15 Digester Cow Goat Xylose_utilization L−Arabinose_utilization Pentose_phosphate_pathway Glycolysis_and_Gluconeogenesis Pyruvate_metabolism_II:_acetyl−CoA, _acetogenesis_from_pyruvate Pyruvate:ferredoxin_oxidoreductase Fermentations:_Mixed_acid Ethanol Butanol pyruvate fermentation to (S)−lactate pyruvate fermentation to (R)−lactate Acetyl−CoA_fermentation_to_Butyrate lactate fermentation to propanoate dissimilatory sulfate reduction IMF Formate hydrogen oxidation/production Clost ridium_1_D Bacteroides_3_DBacteroides_1_DAvimicrobium_1_DKineoth rix_1_D Sphaerochaeta_1_D Nit ratidesul fovib rio_1_D Dethiosul fovib rio_1_D Desul fovib rio_1_D Lachnospi raceae_1_C Cryptobacteroides_2_CAnaeroplasma_1_C Prevotella_1_C Treponema_1_C Lachnospi raceae_3_C Lachnospi raceae_2_C Lachnospi raceae_4_C Desul fovib rio_1_C Ruminococcaceae_1_GLachnospi raceae_4_G Ruminococcaceae_2_G Bacteroidales_1_G Lachnospi raceae_5_G Ruminococcaceae_3_GLachnospi raceae_7_G Prevotella_1_G Desul fovib rio_2_G Desul fovib rio_1_G XYLISOM−RXN XYLULOKIN−RXN RIBULPEPIM−RXN RXN0−5116 ARABISOM−RXN RIBULP3EPIM−RXN RIB5PISOM−RXN TRANSALDOL−RXN 1TRANSKETO−RXN PHOSPHOKETOLASE−RXN GLUCOKIN−RXN PGLUCISOM−RXN 6PFRUCTPHOS−RXN PHOSACETYLTRANS−RXN ACETATEKIN−RXN PYRUFLAVREDUCT−RXN PYRUVFORMLY−RXN ACETALD−DEHYDROG−RXN ALCOHOL−DEHYDROG−RXN RXN−161 BUTANAL−DEHYDROGENASE−RXN L−LACTATE−DEHYDROGENASE−RXN DLACTDEHYDROGNAD−RXN ACETYL−COA−ACETYLTRANSFER−RXN RXN−11662 RXN−11667 RXN−16834 PHOSPHATE−BUTYRYLTRANSFERASE−RXN BUTYRATE−KINASE−RXN R11−RXN RXN0−268 METHYLMALONYL−COA−MUT−RXN METHYLMALONYL−COA−EPIM−RXN RXN−19738 ADENYLYLSULFATE−REDUCTASE−RXN RXN−17803 RXN−17804 SULFATE−ADENYLYLTRANS−RXN NA−TRANSLOCATING−RNF 1.2.1.2−RXN 1.12.1.3−RXN HYDROGEN−DEHYDROGENASE−RXN RXN−12215 FHLMULTI−RXN HYDROG−RXN MAG Reaction 2.5 5.0 7.5 10.0 12.5 log2(TPM + 1) .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 27 SUPPLEMENTARY INFORMATION 1 2 CAZyme architectures suggest fine-scale functional differentiation among anaerobic fungi 3 and bacteria during lignocellulose conversion to volatile fatty acids 4 5 Christopher E. Lawson1,2,3,†, Joel P. Howard3†, Thomas S. Lankiewicz1,4, James B. Brown2, 6 Steven W. Singer1,2, Héctor García Martín1,2,5,6, Michelle A. O’Malley1,4 7 8 1Joint BioEnergy Institute, Emeryville, CA, USA 9 2Biological Systems and Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA 10 3Department of Chemical Engineering & Applied Chemistry, University of Toronto, Toronto, ON, Canada 11 4Department of Chemical Engineering, University of California, Santa Barbara, Santa Barbra, CA, USA 12 5DOE Agile BioFoundry, Emeryville, CA, USA 13 6BCAM, Basque Center for Applied Mathematics, Bilbao, Spain 14 †These authors contributed equally to this work 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint 28 Supplementary Figures 1 2 3 Supplementary Figure 1 – Top CAZyme motifs expressed across bacterial and fungal 4 enrichments. CAZyme motifs were predicted using dbcanLight v1.0.2 . BES – 2-5 bromoethansulfonate (bacterial enrichments), Ch-Chloramphenicol (fungal enrichments). 6 7 Supplementary Datasets 8 Supplementary Data 1 – MAG Statistics. MAG statistics, including MAG size, GC content, 9 completeness and redundancy, and DNA /mRNA read mapping, from cow (A), goat (B), and 10 digester (C) bacterial enrichments. 11 12 Supplementary Data 2 – Bacterial MAG CAZyme expression table. Annotation and expression 13 of CAZymes across all bacterial enrichment MAGs. 14 15 Supplementary Data 3 – Fungal CAZyme expression table . Annotation and expression of 16 CAZymes across fungal reference genomes. 17 18 Supplementary Data 4 – MAG pathway summary. Predicted MetaCyc reactions and metabolic 19 pathways across abundant MAGs recovered from cow, goat, and digester enrichments. 20 21 Cow+Ch 0 5 10 15 CBM10 GT8 GT35 CE15;CBM10 PL3_2 CBM6;GH43_16 CE6 GH45_1 CE1;GH11 CBM29 GH45 GH9 GH10 GH1 GH48 GH43_1 CE1 GH11 CE4 CBM10;GH6 log2(tpm + 1) Goat+Ch 0 5 10 15 CBM6;GH43_16 CBM13;CBM6;GH43_16 GH3;GH6 CBM29 GH45 CBM10;GH11 GH45_1 CE6 CE15 GH9 CE1 GH1 GH10 CE1;CBM10 CE1;GH11 CE4 GH11 GH43_1 GH48 CBM10;GH6 log2(tpm + 1) Digester+BES 0 5 10 15 CBM13;CBM2;CE4 GH5_37 CBM3;CBM86;GH10 CBM22;CBM3;GH10 GH73 CBM4;GH9 GH16_21 CBM13;CBM2;CBM6;GH43_29 CBM13;CBM2;PL11 GH9 CBM36;CE4;GH11 CBM2;GH5_4 CBM6;CE1;GH10;GH11 CBM3;GH48 CBM36;GH11 CBM13;CBM2;CE1;GH10 CBM13;CBM2;CBM6;GH43_16 CBM3;GH9 CBM13;CBM2;GH10 CBM13;CBM2;GH11 log2(tpm + 1) CAZyme Motif Cow+BES 0 5 10 15 GH5_38 CBM61;GH53 CE12;CE8;PL1 CE1 GH16_3 GH5_37 GH9 CBM48;CE1;GH10 CBM4;GH10 CBM34;GH13_39 CE8 GH10 GH51_2 GH13_28 CBM4 GH28 CE4 CBM26;GH13_28 CBM26;GH13_32 GH25 log2(tpm + 1) Goat+BES 0 5 10 15 GH5_1 CBM50 GH28 GH16_21 CBM22;CE1;GH30_8 CBM3;CBM37;GH9 CBM37 CBM22;CE1 GH11 CBM22;CBM6;GH10;GH43_16 CBM22;CBM37;GH11 CBM22;GH10 CBM79;GH9 GH25 CBM3;GH9 GH9 CBM37;GH9 CBM22;GH10;GH11 GH48 CBM4;GH9 log2(tpm + 1) .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 10, 2026. ; https://doi.org/10.64898/2026.02.09.704939doi: bioRxiv preprint