In this study, we compared the performance of PacBio HiFi and Illumina sequencing platforms and tested different methods of assembly for metagenomes of deadwood samples. The two platforms generated roughly similar amounts of data, but the PacBio HiFi sequencing produced much longer reads than Illumina. Hifiasm-meta produced the best assemblies for PacBio HiFi data, yielding twice the amount of data and longer contigs than Illumina megahit assemblies. Hybrid assembly was more computationally demanding, and did not improve the number and size of contigs. This is probably because hybrid assembly was originally developed to improve short-read assembly (Wick et al., 2017), as the sequencing error generated by long-read sequencing was high. However, the repetitive library which calls for consensus reads developed by the PacBio HiFi sequencing approach, has significantly improved nucleotide accuracy (Marx, 2023). As a result, short reads are no longer needed to improve the assembly of long reads if sequenced by PacBio HiFi.
PacBio HiFi metagenomic sequencing has demonstrated outstanding efficiency for reconstructing the genomes from deadwood microbial communities, including bacteria that could not be previously cultivated from these samples (Tláskal et al, 2021) or elsewhere. A total of 69 bacterial genomes were generated from 16 Gb of PacBio HiFi reads from 4 deadwood samples, including 67 newly reconstructed MAGs, 14 high-quality and 7 composed of a single contig. It outperformed Illumina assemblies by 6.3 times (11 MAGs), producing less fragmented, less contaminated and more complete genomes. In comparison to related studies, our findings overtake those of (Tláskal et al., 2021b), who assembled 58 MAGs from 25 short-read metagenomes of deadwood from the same forest. Our investigation extended to the eight most abundant deadwood bacterial phyla, revealing successful genome assembly for all phyla except Planctomycetota. Our results suggest that while genome size and relative abundance may not be the ultimate limiting factors for successful binning, the inherent complexity of Planctomycetota, characterized by high species richness and phylogenetic diversity (Fig. 2A), challenges the recovery of their genomes.
Illumina and PacBio HiFi sequencing generated a proportional number of eukaryotic sequences, but PacBio HiFi assemblies yielded more eukaryotic contigs than Illumina assemblies. This discrepancy is likely attributed to the longer sequencing lengths and high accuracy of PacBio HiFi sequencing (Uliano-Silva et al., 2023). Eukaryotic genomes, characterized by intricate features like repetitive regions, introns, and exons (Galagan et al., 2005), require greater metagenomic sequencing depth and sample size to be successfully assembled compared to prokaryotic genomes. For instance, (Saraiva et al., 2023) needed 6,000 terrestrial metagenomes to assemble 197 eukaryotic bins, whereas (Ma et al., 2023) achieved the assembly of 40,039 prokaryotic MAGs from 2,941 soil metagenomes.
Employing PacBio HiFi sequencing, we unveiled the important role of Myxococcota in deadwood decomposition. While Myxococcota have been previously assembled from various ecosystems, including aquatic, terrestrial, host-associated, and built environments (Nayfach et al., 2021), our study marks the first assembly of Myxococcota genomes within the deadwood ecosystem. Characterized by relatively large genomes (8 Mbp according to GTDB database r202), Myxococcota exhibit unique traits, encompassing motility (Nan et al., 2013), predation (Thiery and Kaimer, 2020), fruiting bodies formation (Muñoz-Dorado et al., 2016), and the ability to decompose cellulose (López-Mondéjar et al., 2022). Here, we successfully assembled two abundant Myxococcota (Polyangiaceae) genomes expressing high cellulosic activity, including endoglucanase, exoglucanase, and cellobiohydrolase – enzymes exclusively found in Myxococcota MAGs (Table S10). While Polyangiaceae also expressed CAZymes targeting bacterial biomass, the lack of chitinase prevented them from recycling fungal biomass. Conversely, chitinases were consistently present in Bacteroidota and detected in Verrucomicrobiota, Proteobacteria and Patescibacteria. Remarkably, the sole chitinase found in Patescibacteria 5_UBA9983_A demonstrated the second-highest level of expression (Table S10).
Patescibacteria had been identified in a wide range of ecosystems (Nayfach et al., 2021) including deadwood habitats (Choi et al., 2022; Tláskal et al., 2021b), yet their ecological role remains enigmatic. While their limited genome size, lacking crucial metabolic genes, suggests an obligate epibiotic lifestyle (Kuroda et al., 2022), their parasitic status remains subject to debate (Wang et al., 2023). In our study, UBA9983_A demonstrated a specialized opportunistic lifestyle, exclusively recycling fungal cell wall components (Table S10). Additionally, Patescibacteria of the order Sacharrimonadales showcased elevated mannase expression, an enzyme involved in hemicellulose degradation, highlighting their active contribution in wood decomposition processes. These findings collectively imply that Patescibacteria may not solely depend on host resources. Further analyses are needed to elucidate whether hemicellulose- and fungal-derived residues serve as energy sources for ATP generation or benefits to the host (e.g., commensalism or mutualism). Nevertheless, our results suggest a broader ecological relationship for Patescibacteria beyond parasitism.
Nitrogen concentration in deadwood is low (Tláskal et al., 2021a) and probably represents a limiting factor for efficient decomposition. As metabolic pathways involved in nitrogen assimilation were scarce in the studied samples (see Results), microbial biomass could represent an alternative source of nitrogen (López-Mondéjar et al., 2018). We found that CAZymes targeting bacterial biomass were more frequent than CAZymes targeting fungal biomass, but that peptidoglycanases were 2 times less expressed than chitinases (Table S11, Table S10). This result is surprising since fungal biomass contains less nitrogen than bacterial biomass (Paul and Frey, 2024), but these processes might simply be regulated by the availability of bacterial and eukaryotic biomass. While microbial biomass decomposition represents an interesting strategy for recycling nitrogen during decay, an external source of nitrogen might be required to initiate decomposition of fresh dead wood. The initial nitrogen input is likely provided by bacterial nitrogen fixation, as illustrated by nitrogen fixation rates being eight times higher in young deadwood compared to old deadwood (Tláskal et al., 2021a). The conversion of atmospheric N2 to NH3, being a highly energy-consuming process (Cherkasov et al., 2015), is only realized if no better suitable nitrogen sources are available (Burris and Roberts, 1993). The continuous expression of nitrogenase genes in later stages of decomposition (> 4 years) thus indicates that the recycling of microbial biomass does not completely meet the microbial nitrogen demand throughout the decomposition process.
Despite biological nitrogen fixation has been extensively investigated (e.g. Davies-Barnard and Friedlingstein, 2020; Dos Santos et al., 2012; Zehr and Capone, 2020), our study marks the first recovery of this function within the Steroidobacteraceae family. Beyond reporting a novel nitrogen-supplying bacteria in a nitrogen-limited environment, PacBio HiFi sequencing facilitated the investigation of genes involved in the conversion of N2 into ammonium. In the deadwood ecosystem, Proteobacteria catalyze the production of reduced ferredoxin/flavodoxin using the fix operon, particularly advantageous under oxygen-limited conditions (Alleman et al., 2022). In addition, Steroidobacteraceae employ the isc system for biosynthesizing [Fe-S] proteins, crucial under elevated oxygen conditions (Johnson et al., 2006). The coexistence of the fix operon and the isc system likely allows Steroidobacteraceae to maintain nitrogen fixation activity under oxygen concentration fluctuations. Alphaproteobacteria appear less sensitive to elevated oxygen, favoring the suf system over the isc system, which is known to be more beneficial for bacterial growth in the presence of hydrogen peroxide (Tokumoto, 2004). Although the proportional contribution of Steroidobacteraceae (Gammaproteobacteria) and Alphaproteobacteria to global deadwood nitrogen fixation requires further exploration, our study provides empirical evidence of their in situ activity.
PacBio HiFi sequencing unveiled the remarkable potential of deadwood microorganisms for the production of diverse secondary metabolites, an important trait that would be totally neglected if one would just rely on short read-based approaches. Our study identified over a thousand mostly novel BGCs, showcasing the extensive diversity of these BGCs in MAGs. This diversity suggests that deadwood bacteria display multiple interactions with other microorganisms in deadwood. Furthermore, our investigation highlighted bacterial groups in deadwood that hold promise for the production of novel bioactive compounds. Similar to observations in soil (Sharrar et al., 2020), the abundance of BGC types varied by taxonomy. In deadwood, we observed high numbers of BGCs in Myxococcota and Proteobacteria, as well as in other groups such as Verrucomicrobiota and Acidobacteriota, whose biosynthetic potential has only recently been reported (Crits-Christoph et al., 2018; Waschulin et al., 2022). Notably, we did not identify a shared BGC family across different bacterial taxa, indicative of a compound with broader importance in deadwood. In contrast, taxonomic conservation of BGC families in deadwood MAGs was observed. The presence of BGCs for flexirubins, pigments typical of Bacteroidota (Brinkmann et al., 2022), in several Chitinophagaceae MAGs constitutes convincing proof of validity of our metagenome analysis.
Long-read sequencing-based analysis of metagenomes, while powerful, still has limitations. More effort would be needed to appropriately describe the fungal component of the deadwood microbiome and this is partly true also for the Planctomycetes: although their sequences have been obtained by PacBio HiFi sequencing, the high species richness and phylogenetic diversity of this phylum challenged their genome recovery. Still, long-read sequencing-based metagenome analysis stands unparalleled indicating microbiome functions not recoverable using the conventional approaches – short-read metagenomics and culturing. It enabled the assembly of novel genomes of bacteria with key roles in deadwood decomposition: cellulose decomposition in Polyangiaceae, nitrogen fixation in Steroidobacteraceae. It also revealed significant contribution of Patescibacteria to wood decomposition processes, and identified the wealth of new BGCs with potential ecological and/or biotechnological significance.