The ISME Journal volume 17, pages 916-930 (2023)Citer cet article
4516 Accès
5 citations
22 Altmétrique
Détails des métriques
Alors que la décomposition de la matière organique par les bactéries joue un rôle majeur dans le cycle des nutriments dans les écosystèmes terrestres, l’importance des virus reste mal comprise. Ici, nous avons combiné la métagénomique et la métatranscriptomique avec l'échantillonnage temporel pour étudier l'importance des bactéries mésophiles et thermophiles et de leurs virus sur le cycle des nutriments lors du compostage hyperthermophile (HTC) à l'échelle industrielle. Nos résultats montrent que la dynamique et l'activité de la densité virus-bactéries sont étroitement couplées, où les virus spécifiques aux bactéries mésophiles et thermophiles suivent la densité de leurs hôtes, déclenchant la succession de la communauté microbienne via un contrôle descendant au cours du HTC. De plus, les virus spécifiques aux bactéries mésophiles codaient et exprimaient plusieurs gènes métaboliques auxiliaires (AMG) liés au cycle du carbone, ayant un impact sur le renouvellement des nutriments aux côtés des bactéries. Le renouvellement des nutriments était corrélé positivement avec le rapport virus-hôte, ce qui indique une relation positive entre le fonctionnement de l'écosystème, l'abondance virale et l'activité virale. Ces effets étaient principalement dus aux virus à ADN, car la plupart des virus à ARN détectés étaient associés aux eucaryotes et non au cycle des nutriments pendant la phase thermophile du compostage. Nos résultats suggèrent que les virus à ADN pourraient piloter le cycle des nutriments au cours du HTC en recyclant la biomasse bactérienne par lyse cellulaire et en exprimant des AMG clés. Les virus pourraient donc potentiellement être utilisés comme indicateurs du fonctionnement des écosystèmes microbiens pour optimiser la productivité des systèmes biotechnologiques et agricoles.
La décomposition de la matière organique est un processus écosystémique clé, ayant un impact sur le cycle des nutriments et la productivité dans les écosystèmes terrestres [1]. S’il est connu que les communautés bactériennes et fongiques jouent un rôle crucial dans le recyclage des nutriments via « la boucle microbienne » [1], le rôle des virus bactériens, c’est-à-dire des bactériophages (phages), est encore mal compris [2]. En tant qu'entité biologique la plus abondante sur Terre, les virus jouent un rôle essentiel dans la mortalité microbienne par la lyse cellulaire [3, 4], ayant un impact significatif sur le cycle des éléments via la libération de nutriments et affectant la composition, la diversité et la nécromasse microbienne de la communauté microbienne [5,6, 7,8,9,10]. Bien que l’importance des virus dans le cycle des nutriments dans les océans soit bien établie [11], nous commençons seulement à comprendre comment les virus modulent le renouvellement des nutriments et la minéralisation de la matière organique dans les sols [9, 10, 12]. Il a été démontré que les gènes métaboliques auxiliaires (AMG) codés par les virus associés aux glycosides hydrolases et au métabolisme du méthane contribuent au cycle du carbone dans les écosystèmes du sol (4, 6, 13). Par exemple, 14 AMG incluant 9 familles de glycosides hydrolases, comme l’endomannanase avec une activité fonctionnelle confirmée, indiquent que les virus ont la capacité potentielle de participer à une dégradation complexe du carbone [4]. En plus de piloter le cycle des nutriments via la lyse des cellules bactériennes dans les sols, il a récemment été démontré que les virus améliorent la survie de leurs bactéries hôtes sous un stress environnemental en codant pour des gènes métaboliques auxiliaires (AMG) qui améliorent la capacité métabolique des hôtes bactériens (14). Malgré ces progrès récents, nous avons encore une compréhension limitée de la manière dont les virus et les bactéries déterminent ensemble le cycle des nutriments et la décomposition de la matière organique dans les écosystèmes terrestres [15].
Ici, nous avons utilisé un compostage hyperthermophile (HTC) comme système modèle pour étudier le rôle des bactéries mésophiles et thermophiles et de leurs virus dans la décomposition de la matière organique. HTC est une technologie de traitement des déchets utilisée dans la dégradation de la fraction organique des déchets solides municipaux ou agricoles, atteignant des températures extrêmement élevées (jusqu'à 90 °C) sans échauffement exogène dû à l'activité de la communauté bactérienne thermophile [16,17,18]. Le HTC contient trois phases de température principales : hyperthermophile (>80 °C), thermophile (>50 °C) et phase de maturation (température ambiante). Au cours du processus, les substances polymères enrichies en carbone et en azote (lignocellulose, protéines, polysaccharides et lipides) sont dégradées pendant les phases thermophiles du compostage, tandis que les composés enrichis en humique se dégradent lentement pendant la phase de maturation. La composition des communautés microbiennes qui conduisent à la dégradation de la matière organique change dynamiquement en fonction de la température de compostage au cours du HTC [19], et les taxons thermophiles et résistants à la chaleur (Firmicutes ; Bacillus et Deinococcota ; Thermus) sont importants pour la décomposition de la matière organique au cours du HTC. phase thermophile [16]. Alors que les effets de la température, des matières premières et des propriétés physicochimiques du compostage ont été largement étudiés en relation avec l'assemblage de la communauté microbienne et la dégradation de la matière organique (20, 21), on sait très peu de choses sur le rôle des virus lors du HTC.
90 °C), thermophilic phase (from day 10 to 26: >55 °C) and maturation phase (from day 27 to 45: <45 °C). To cover changes during the whole composting process, eight samples from five compost piles were collected at different phases of hyperthermophilic composting on days 0 (D0), 4 (D4), 7 (D7), 9 (D9), 15 (D15), 21 (D21), 27 (D27), 33 (D33) and 45 (D45). To obtain well-distributed and homogenized samples, each pile was diagonally divided into five domains, and each domain was sampled from the same location at a depth of 40–50 cm at different sampling time points. Within each pile, five subsamples (5000 g each) per domain were collected, and then mixed into a single composite sample, which was further divided into two aliquots. One replicate aliquot was stored in liquid nitrogen for biological analyses and the other was kept at 4 °C for physicochemical analyses. An automatic temperature controller was used to determine temperature changes during the composting./p>30 and length >36 bases) [35]. All high-quality sequences were co-assembled using SPAdes v3.13.1 with the parameters “-k 33, 55, 77, 99, 111,127 --meta” [36]. We also assembled reads generated at each thermal phase of composting separately (composting phase-specific assemblies) using SPAdes with the same parameters. All assembled scaffolds longer than 2.0 kb were binned using metawrap [37] based on MetaBAT2 [38], MaxBin2 [39], and Concoct [40] with default parameters. Bins were further manually curated to obtain high-quality genomes using Bin_refinement module in Metawrap [37]. The completeness and contamination of genome bins were assessed using CheckM v1.0.13 [41], and metagenome-assembled genomes (MAGs) with more than 50% completeness and less than 10% contamination level were retained for further analyses. Bins from different samples were dereplicated to produce medium to high quality genomes using dRep v.2.3.2 [42] and assigned to taxonomic classifications based on the Genome Taxonomy Database (GTDB; release 03-RS86) using the GTDB-Tk toolkit (v.0.3.2) with the classify workflow [43]. To construct bacterial MAGs, genes were called using Prodigal with parameters “-p meta” [44] and annotated against the KEGG and Pfam databases using the Diamond tool [45]. The predicted proteins were screened for candidate CAZymes using hmmscan module from HMMER v3.2.1 and dbCAN database (cutoffs: coverage fraction: 0.40; e-value:1e-18) [46]. Genes encoding proteases and peptidases were identified using Diamond against the MEROPS database release 12.0 (cutoffs: e-value 1e-20 -accel 0.8). Ribosomal RNAs were predicted using RNAmmer v1.2 [47]. The optimal growth temperature (OGT) of MAGs was predicted by the machine learning method using the Tome v1.1 [48]. Thermophilic MAGs were defined as ones with OGT ≥ 50 °C, while MAGs were assigned as mesophilic when their OGT < 50 °C. To build phylogenetic MAG trees, the “classify” workflow in GTDB-Tk (v.0.3.2; default settings) was used to identify 120 bacterial marker genes, which were used for tree construction based on multiple sequence alignment. The resulting FASTA files containing multiple sequence alignments of the submitted genomes were used for maximum likelihood phylogenetic tree inference using FastTree v.2.1.10 with the default parameters [49]. Newick tree output files were visualized with iTOL v.5 [50]./p>1000 bp, composting phase-specific assemblies) were compared against the database containing all available viral RdRp gene sequences in NCBI/GenBank (37, 441 genes, downloaded on February 2023) and previous published studies [77, 78] using Diamond BLASTx (coverage ≥ 70%, E-value≤1e-10 and score ≥ 70). Sequences that had hits in the RdRp database with the RdRp core domain were considered as the potential RNA viruses [80]. This analysis identified 109 contigs with RdRp gene. These potential RNA virus contigs were clustered with CD-HIT using 95% average nucleotide identity across 85% alignment fraction, resulting in a total of 83 potential RNA viruses./p>5 kb) were obtained from the metatranscriptomic assemblies. After clustering (95% nucleotide similarity and over 85% coverage), a total of 41 dsDNA viral operational taxonomic units (dsvOTUs) were retained. By comparing the viruses’ contigs derived from transcriptomic data and metagenomic data, only 7 of 68 dsvOTUs could be assembled from the metagenomic data. This is not surprising as the DNA was removed during RNA library preparation and very few DNA sequences was retained in transcriptome./p>80 °C for 9 days (“hyperthermophilic phase”), after gradually declining to 55 °C (“thermophilic phase”) and ambient temperature by day 27 (“maturation phase”, Fig. 1a). The organic matter (OM) decomposition, carbon, and nitrogen turnover followed closely different phases of HTC (Fig. 1a). Compared to the initial composting raw materials, total carbon (TC, F3,23 = 33.6, p < 0.0001) and nitrogen (TN, F3,23 = 19.8, p < 0.0001) contents significantly decreased by 32% and 28% by the end of HTC, respectively (Fig. S1a). Similarly, the OM content that showed the highest degradation rate at the hyperthermophilic phase declined from 51.3% to 38.7% (F3,23 = 68.3, p < 0.0001), while the concentration of water-soluble carbon (WSC, F3,23 = 19.8, p < 0.0001) and water-soluble nitrogen (WSN, F3,23 = 26.4, p < 0.0001) increased during HCT, reaching peak concentrations at the hyperthermophilic phase (Fig. 1a). The degradation rate of OM correlated positively with temperature, WSC, and WSN (Fig. S1b), indicative of efficient nutrient cycling during HTC./p>5 kb) were obtained from the metagenomic assemblies. After clustering (95% nucleotide similarity and over 85% coverage), a total of 1297 viral operational taxonomic units (vOTUs) were retained (Table S1), which mainly belonged to double-stranded DNA viruses (97%) and were predicted to be mostly lytic (66.2%). The genome quality of vOTUs consisted of 0.7% of high-quality, 2.4% of medium-quality, and 85.6% low-quality vOTUs, while the quality of remaining 11.3% of vOTUs could not be determined. Overall, 78.6% of vOTUs were detected during non-thermophilic phases (D0 and D27), while 21.3% occurred during thermophilic phases (D4 and D15, Table S1). Only 7.7% of vOTUs could be clustered with taxonomically known viruses in RefSeq database (v216, released in February 2023), while only 35 vOTUs (2.6%) clustered with known viruses in IMG/VR (v3) database, suggesting that most of the composting viruses were novel. Primarily, they belonged to Dividoviricota (88%) and Uroviricota (2%) phyla and Mesyanzhinovviridae (27.2%), Herelleviridae (18.2%), Salasmaviridae (16.4%), Autographiviridae (5.4%), Vilmaviridae (5.4%) and Matshushitaviridae (3.6%) families (Table S1 and Fig. S4). Similar to bacteria, the richness (F3.8 = 4.7, p = 0.0359) and composition (R2 = 0.78, p < 0.001, PERMANOVA test) of viral communities followed different phases of HTC (Fig. 1d). While Vilmaviridae (37.8%) and Autographiviridae (14.5%) were dominant viruses in the composting raw material (D0), Vilmaviridae abundances significantly decreased to 1.5% by the maturation phase (D27; F3,8 = 9.7, p = 0.0047, Fig. S4). In contrast, relative abundance of Matshushitaviridae family under Dividoviricota phylum (consisting mainly of thermophile-associated Thermus phages) increased from 1.4% at D0 to 66.3% at D15 (F3,8 = 5.7, p = 0.0245, Fig. S4). As most of viruses could not be classified, viral abundances were also investigated based on their predicted host taxonomy (see Methods). The viral taxa abundances followed bacterial taxa abundances (Fig. 1d), and for example, the abundance of viruses infecting Deinococcota clearly increased with rising composting temperature by D15. Moreover, Matshushitaviridae viral abundances correlated positively with their Firmicute (R2 = 0.34, p = 0.028) and Deinococcota (R2 = 0.53, p = 0.0042, Fig. S5) host abundances. Overall, changes in viral community richness (R2 = 0.50, p = 0.0058) and composition (beta-dissimilarity, R2 = 0.71, p < 0.0001) correlated positively with changes in bacterial community richness and composition (Fig. 2a, b)./p> 50 °C) bacteria (upper panel) and 180 mesophilic and 47 thermophilic MAGs (lower panel) during HTC. b Box plots and heatmap representing changes in the transcriptional activity of viruses associated with mesophilic and thermophilic bacteria (upper panel) and individual vOTUs (lower panel) during HTC. Box plots and heatmaps representing changes in the transcriptional activity of mesophilic (OGT < 50 °C) and thermophilic (OGT > 50 °C) bacteria during HTC based on mean (upper panel) and individual (lower panel) MAGs (including 180 mesophilic and 47 thermophilic MAGs) in association to carbon (CAZyme) (c) and nitrogen metabolism genes (d). e Box plot and heatmap representing changes in the transcriptional activity of virus-associated carbon (CAZyme) metabolism genes linked with mesophilic MAGs (OGT < 50 °C). In all (a–e), the mean transcriptional activity (MAGs and vOTUs) shown in boxplots is based on transcript abundances (transcripts per million, TPM) normalized by MAG and vOTU abundances. Box plots encompass 25–75th percentiles, whiskers show the minimum and maximum values, and the midline shows the median (dots present the biologically independent samples, asterisks denote for significant differences (*p < 0.05, **p < 0.01. n.s, no significant differences). Heatmaps show the transcriptional activity (MAGs or vOTUs) based on non-normalized transcripts abundances (transcripts per million, TPM). In (c and d), selected CAZymes include GHs, GTs, PLs, CEs, CBMs, and AAs. Nitrogen metabolic pathways include assimilatory nitrate reduction, dissimilatory nitrate reduction, nitrification, and nitrogen fixation pathways. More detail about the functional genes included can be found in Supplementary Data 6 and 7, respectively./p>5 kb) were obtained from the metatranscriptomic assemblies. After clustering (95% nucleotide similarity and over 85% coverage), a total of 41 dsDNA viruses were retained. By comparing the dsDNA viruses contigs derived from transcriptomic data and metagenomic data, 89% viruses (61 of 68) could be assembled from both transcriptomic and metagenomic data, suggesting that very few dsDNA viruses exclusively exist in metatranscriptomic dataset. As a result, the RNA viral community abundance (based on beta-dissimilarity of abundance matrix) did not correlate with changes in composting properties (Mantel statistic r = 0.0173, p = 0.35), which suggests that they did not contribute to the nutrient cycling during composting./p>90 °C) and consistently surpassed bacterial abundances in terms of virus-host abundance ratio. Mesophilic and thermophilic bacteria and their viruses showed clear microbial community succession, where the initial phase of composting was dominated by mesophiles, which were subsequently replaced by thermophiles and subsequently by mesophiles towards the end of the HTC. Although similar compositional succession of bacterial and fungal communities have been observed in previous composting experiments [19, 90], this is the first evidence demonstrating that viruses can also drive ecological succession in microbial communities during HTC. These findings are also indicative of “Kill-the-Winner” hypothesis, where viruses target and regulate the most abundant group of host bacteria, reducing the dominance effects and evening out competition between different bacterial taxa [4, 89]. Such dynamics could explain the observed community shift between thermophilic and maturation phases of HTC, where thermophilic viruses likely drove down the abundances of thermophilic bacteria, giving rise to mesophilic bacteria and their phages. For example, Thermus and Planifilum bacterial genera play important role in heat production during HTC [17] and several lytic phages that infected Thermus thermophilus (T_bin.227) and Planifilum fulgidum (T_bin.201) were identified, including five potentially novel Thermus viruses that had genome sizes about 5 kbp similar to hyperthermophilic phage φOH3 isolated from Obama hot spring [91]. While 61% of detected phages were predicted to be lytic, it is possible that some of the correlations between bacterial and viral taxa were also driven by lysogenic phages or prophages because unfiltered DNA samples were used for metagenomics. As a result, our dataset likely underestimates phage diversity, and phage enrichment [92] should be used in future studies. Moreover, future work should also consider the potential role of RNA viruses for HTC, which we did not explore in detail as compost-associated bacteria are most often associated with DNA viruses [19]. Nevertheless, our results suggest that a small portion of thermophilic viruses played a key role in microbial activity during the thermophilic phase of HTC, indicating that compost ecosystem functioning was at least temporally driven by low-diversity microbial communities. Terrestrial phages could hence be important drivers of biogeochemical cycling in soil ecosystems via “viral shunt” akin to marine phages [11, 93]./p>
Français
English
Deutsch
Español
Italiano
Português
日本語
한국어
Русский