Mutual cross-feeding drives marine biofilm assembly on various carbon sources

doi:10.21203/rs.3.rs-3209408/v1

Download PDF

Article

Mutual cross-feeding drives marine biofilm assembly on various carbon sources

https://doi.org/10.21203/rs.3.rs-3209408/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

A major factor affecting the assembly of microbial community is environmental carbon source. It is still unclear, however, to which extent the community structure is determined by the type of carbon source, especially for marine microbiota with high diversity. Here, this research question has been systematically addressed by enrichment culture of a marine biofilm community with 69 different carbon sources, under both aerobic and anaerobic conditions, followed by analysis of 3.2 Tb of metagenomic datasets. The finding revealed that the taxonomic composition of the enrichment cultured communities is not primarily determined by carbon source. Analysis of 535 high-quality metagenome-assembled genomes revealed strong microbial coexistence across different carbon sources. Moreover, co-culture experiments with isolated strains suggested extensive microbial cooperation, which expands the range of available carbon sources. Furthermore, co-culture metabolomics and transcriptomics indicated the presence of an extracellular amino acid pool that facilitate cross-feeding, which is probably regulated by complementary gene expression. Altogether, cross-feeding based on the metabolism of essential elements (e.g., amino acids) lays the foundation of microbial cooperation, diminishing the influence of carbon source on community assembly.

Biological sciences/Microbiology/Biofilms

Biological sciences/Microbiology/Microbial communities/Microbiome

Microbiota play fundamental roles in mediating biological processes in a variety of ecosystems, spanning from soil to marine environments. Environmental factors, such as the available carbon sources, affect the assembly of microbiota^1,2,3. For example, culturing of different soil microbiota in presence of leucine, arginine, glucose, and cellobiose as sole carbon sources led to only 34% similarity in the average taxonomic composition among all cultured microbiota¹. This suggested that different carbon sources support the assembly of different community structures. Another study³ monitored the assembly of soil- and plant-derived microbiota in presence of three single-carbon sources, including glucose, citrate, and leucine. The findings suggested that the community structure is majorly determined by the type of carbon source. However, internal factors, imposed by microbial interactions as well as the underlying mechanisms (e,g., competition or cooperative coexistence) can be an alternative factors shaping the microbial community^4,5. Using an ultrahigh-throughput platform, interactions among 20 soil bacteria were investigated to show common positive interactions among culturable bacteria⁶. In marine environments, dissolved organic carbon is essential in maintaining the growth and existence of heterotrophic microorganisms as well as the stability of whole marine biosphere⁷. Marine microorganisms can consume a variety of dissolved carbon compounds, which may have significant influence on the structure and composition of microbiota⁸. However, since most of the existing studies are confined to a few carbon sources, it is still unclear to which extent external carbon source affect the assembly of microbiota, especially the microbiota in the marine environment.

Biofilms are colonies of a single bacterial strain or communities comprised of different bacteria. Biofilms are formed by aggregation of microbial cells after attachment to biotic or abiotic surfaces^9–12. It has long been known that the interactions between the microorganisms within a biofilm community are complex. These interactions are mediated by sharing of products/nutrients and signal transductions. For example, some bacteria secrete digestive enzymes to degrade polymers into soluble products that can be captured by other neighboring bacterial cells¹³. Besides, fatty acyl homoserine lactone, whose production regulates the physiology and activity of biofilms, can influence the behavior of various biofilm-forming bacteria¹⁴. Recent studies on global ocean biofilms suggested that these microbial communities possess very high and novel microbial diversity^15,16. Moreover, a functional “core” has been observed across the biofilms collected from different ocean provinces, pointing to the existence of a universal principle that governs the assembly of marine biofilm communities¹⁵. Furthermore, marine biofilm-derived bacteria can respond quickly to the availability of carbon source¹⁷ and show versatility in energy production¹⁸_, suggesting that environmental carbon source may also exert significant influence on the prevalence of the biofilm communities. All these features highlight the ecological importance of the marine biofilm community assembly, and motivate a surge of interest in exploring the underlying mechanisms.

In the present study, the influence of carbon sources on microbial community assembly has been investigated through enrichment cultivation of marine biofilm microbiota using 69 single carbon sources. These carbon sources are typical dissolved organic matters that are abundant in marine waters, and can be classified into four categories including alcohols, amino acids, organic acids (exclude amino acids), and saccharides. Subsequently, comprehensive and comparative analyses of community structure and microbial functional traits were performed based on metagenomics to discover the taxonomic composition as well as the interactions between microorganisms. Experiments based on isolated strains were conducted to further explain the strength of microbial interactions and its role in determining the community assembly.

The taxonomic composition of the enriched biofilms was not based on the type of carbon source

The outline of study has been shown in Supplementary Fig. 1. The seed biofilm was collected from the surface of a rock immersed in subtidal zone. The selected carbon sources for enrichment culture included 17 alcohols, 23 amino acids, 15 organic acids (exclude amino acids), and 14 saccharides. Details of carbon sources has been provided in Supplementary Table 1. Since the utilization of carbon sources is most likely linked to the availability of oxygen¹⁹, the enrichment culture experiments were performed under both aerobic and anaerobic conditions. Thus, total 138 cultures were set up (69 carbon source × 2 oxygen conditions). After obtaining stable cell densities in the culture, these communities were subjected to metagenomic sequencing to generate in total 3.2 Tb Illumina datasets (22.8 ± 0.8 Gb per metagenome; Supplementary Table 2). Samples cultured in presence of same carbon source category were pooled together for Nanopore sequencing, generating 37.4 Gb datasets (12.5 ± 0.6 Gb per metagenomes; Supplementary Table 3). Classification of the miTags (183,028 ± 46,301 per metagenome) extracted

from the Illumina-sequenced metagenomes at 97% similarity revealed 9,497,699 operational taxonomic units (OTUs). These OTUs were further classified into 75 phyla (class level for Proteobacteria), 434 families, and 2,027 genera. Gammaproteobacteria was found to be the most prevalent phylum with a maximum percentage of 99.62%, followed by Alphaproteobacteria (86.31%) and Bacteroidetes (29.05%) (Fig. 1a). At the family level, Oceanospirillaceae (87.61%), Vibrionaceae (82.44%), Kiloniellaceae (67.89%), Rhodobacteraceae (67.42%, members of this family have been recently reclassified into Roseobacteraceae²⁰), Pseudomonadaceae (64.44%), and Aeromonadaceae (62.17%) were the most abundant taxa (Supplementary Fig. 2). At the genus level, Marinomonas, Pseudolateronomas, Marinobacterium, Microbacterium, Roseibium, and Ruegeria were prevalent taxa (Supplementary Fig. 3). Overall, no distinct taxonomic composition boundaries were observed between communities assembled on different categories of carbon source. However, huge variations were observed among the communities assembled on carbon sources of same category. Thus, it was speculated that the taxonomic composition of the enriched microbiota was not determined by the type of carbon source.

To confirm this speculation, similarities among the communities were observed through principal co-ordinates analysis (PCoA), which was conducted based on Bray-Curtis distances of the phylum- (Fig. 1b), family- (Supplementary Fig. 4a), and genus-level compositions (Supplementary Fig. 4b). In order to exclude the inference of oxygen conditions, the communities obtained under aerobic and anaerobic conditions were analyzed separately at each taxonomic level. A majority of the samples were overlapped, and aerobic and anaerobic samples were not well separated in the PCoA plots (Fig. 1b; Supplementary Fig. 4a, b). This suggested that availability of oxygen had no role in determining the community structure. To certify the findings of above analyses based on 16S miTags, single-copy marker gene sequences were identified for taxonomic profiling, followed by PCoA at the genus level. Again, most of the biofilm communities were overlapped (Supplementary Fig. 4c), which confirmed that the overall community structure of the enriched biofilms is not based on carbon source.

Extensive microbial coexistence across the enriched biofilm communities

To get genome-level and functional explanations for the dissociation between carbon source category and composition of microbial communities, the metagenomes were assembled and genome binning was performed. For each carbon source, aerobic and anaerobic communities sequenced by the Illumina platform were cross assembled and the contigs were subjected to differential coverage binning. The key metagenome-assembled genomes (MAGs) were reassembled with the Nanopore datasets for improvement of genome completeness. To this end, 535 non-redundant (average nucleotide identity < 99%) MAGs were obtained with high quality (completeness > 90% and potential contamination < 5%). The information of metagenomic contigs and MAGs are given in Supplementary Table 4 and Supplementary Table 5, respectively. These MAGs were distributed in 13 phyla (class level for Proteobacteria), 68 families, and 156 genera (Supplementary Table 5). Whole genome-based phylogenetic analysis revealed diverse phylogeny, and at the phylum level, these MAGs mainly belonged to Gammaproteobacteria, Alphaproteobacteria, Bacteroidota, and Spirochaetota phyla, amongst others (Supplementary Fig. 5).

Afterwards, the relative abundance of the 535 MAGs was calculated in the 138 metagenomes and the single-genome level distribution patterns were observed. In the resulting heatmap, many of the MAGs showed similar distribution patterns (Supplementary Fig. 6), implying that these MAGs may coexist across the enrichment cultured biofilms. To test this notion, Spearman correlation analysis was applied to construct a correlation matrix between vectors of MAG abundances across all the metagenomes. As a result, 383 MAGs showed coefficient of 0.9 or higher (average 0.95), indicating strong and widely distributed correlations (Fig. 2a). Four major clusters formed by MAGs from different microbial genera were observed (indicated by boxes in Fig. 2a). These MAGs were mostly affiliated to Alphaproteobacteria (e.g., Roseibium), Gammaproteobacteria (e.g., Pseudoalteromonas), Actinobacteriota (e.g., Glutamicibacter), and Bacteroidota (e.g., Polaribacter_A) phyla (Supplementary Table 6). Subsequently, a network was constructed using a threshold of coefficient above 0.9 and P-value lower than 0.05, to show the connections between the 383 MAGs. These MAGs formed a network with 1557 edges. The largest cluster consisted of MAGs belonging to Alphaproteobacteria, Gammaproteobacteria, and Bacteroidia (Fig. 2b). The average degree, network diameter, graph density, modularity, average clustering coefficient, and average path length were 8.13, 7.00, 0.02, 0.79, 0.86, and 2.40 respectively. According to the threshold used in a previous study²¹, these values here indicated that the MAGs were tightly connected to form a network with high density.

To understand the potential functions that were responsible for the observed microbial correlation, functional genomic comparison of selected MAG groups was performed. Two groups were created. The first group contained MAGs with highest connection numbers (edge number ≥ 14) and no significant abundance variations (P-value > 0.05) between carbon source categories. The second group comprised of MAGs with lowest connection numbers (edge number = 1) and significant (P-value < 0.05) abundance variation between at least one pair of carbon source categories. To this end, 33 MAGs in the first group were compared with 24 MAGs in the second group. Searching against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database revealed total 5630 KEGGs for the 57 MAGs. After statistical analysis using t-test and Hypergeo-test, 329 KEGGs showed significant variation (P-value < 0.05 in both tests) in terms of the average number of genes per MAG between the two groups. KEGGs corresponding to amino and peptide metabolism were highly prevalent in the MAGs with strong connections (Fig. 2c). For example, the peptide/nickel transport system substrate-binding protein (K02035) and the peptide/nickel transport system permease protein (K02033) had more than eight average copies per MAG in the strongly-connected MAGs, whereas only three average copies of these KEGGs were observed in the MAGs with no connection (Fig. 2c).

Positive interactions among biofilm-derived culturable strains

Above analyses revealed extensive microbial coexistence across biofilms on various carbon sources, which implied strong and comprehensive interactions among microorganisms. To confirm this notion using experiments, strains were isolated from the enriched biofilms to conduct pairwise co-culture. In total, 107 non-redundant (16S rRNA gene identity < 99.9%) strains were identified, which were mostly affiliated to Alphaproteobacteria and Gammaproteobacteria. At the genus level, Vibrio was the most prevalent taxa, followed by Microbacterium, Albirhodobacter, Glutamicibacter, Aurantimonas, Tritonibacter, Celeribacter, Roseibium, and Pseudoalteromonas (Supplementary Fig. 7). Six strains including Pseudoalteromonas shioyasakiensis C717, Roseibium aggregatum S1616, Microbacterium maritypicum S1611, Salipiger pacificus T6124, Leisingera aquaemixtae M597 (this strain was also isolated from the seed biofilm), and Sulfitobacter indolifex W002 (also isolated from the seed biofilm) were monocultured and co-cultured pair by pair in presence of 21 single carbon sources, including four alcohols, nine amino acids, four organic acids, and four saccharides. These strains were selected because MAGs belonging to the corresponding genera showed strong correlations in the MAG correlation analysis (Supplementary Table 6). Notably, growth of these bacteria in single carbon sources was largely promoted by co-culturing (Fig. 3a). In many cases, the monocultured bacteria could not grow well (OD₆₀₀ of cell density < 0.1). On the other hand, co-cultured consortia displayed apparently distinct patterns, with substantial cell densities for a broad range of carbon sources (Fig. 3a). For example, when cultured individually, S1616 or M597 could not utilize the saccharides. However, when cultured together, these two strains showed substantial cell densities in D-maltose, L-arabinose, L-rhamnose, D-arabinitol, and D-mannitol (Fig. 3a). Similarly, co-culturing of these two strains in D-gluconic acid yielded much higher cell densities than monoculture (Fig. 3a). Subsequently, time series of cell density dynamics were observed in the consortium of S1616 and M597 co-cultured in D-gluconic acid, which resulted in a cell density of OD₆₀₀ 0.37 after three days (Fig. 3b). To confirm the stable coexistence of these two strains, 16S rRNA gene sequencing (data information is given in Supplementary Table 7) was conducted with two-day intervals after seeding the strains in the medium containing D-gluconic acid. The relative abundance of S1616 was higher than that of M597 and increased within the first six days. Afterwards, the abundance of S1616 decreased on the 8th day and was stabilized at approximately 50% (Fig. 3c). The consumption of D-gluconic acid by the co-cultured and monocultured consortia was also examined through biochemical assays, which showed an increased carbon consumption rate during co-culturing (Supplementary Fig. 8). These results evidenced positive interactions among the biofilm-derived strains.

Metabolic and molecular bases for cross-feeding between two representative strains

Next, the functional basis for the positive microbial interactions were explored through the co-culture experiments. Given the overexpression of genes for peptide and amino acid metabolism in the strongly-correlated MAGs, we hypothesized that, amino acid utilization might be the limiting factor for individual bacterial growth, whereas cross-feeding based on amino acid metabolism may expand the range of available carbon sources during co-culture. To test this hypothesis, we designed an experiment comprising the following steps: 1) S1616 and M597 were co-cultured in the presence of D-gluconic acid until obtaining the stationary phase and the supernatant was extracted; 2) the supernatant was amended with D-gluconic acid and the mixture was used to culture the two strains individually; 3) the supernatants of the two monocultures the co-culture supernatant obtained in the first step were subjected to quantitative detection of individual amino acids (Fig. 4a). Monoculture supplemented with the co-culture supernatant revealed significantly higher cell density for these two strains, especially for M597, which could not grow in monoculture (Fig. 4b). Quantitative detection of individual amino acids in the supernatants revealed significant changes in concentration of certain amino acids before and after addition in the monoculture (Fig. 4c). For example, the glutamate concentration in the co-culture supernatant increased after monoculture of S1616, whereas it decreased after M597 monoculture (Fig. 4c). Other amino acids, including alanine, citrulline, phenylalanine, threonine, and tyrosine, also displayed changes in concentrations, and in most cases, these amino acids were consumed or excreted by different strains (Fig. 4c).

To explore gene-level mechanisms of the metabolic cross-feeding, genomics and transcriptomics were conducted on S1616 and M597. Complete genome sequencing was conducted, followed by gene prediction and KEGG annotation. Genome information, including the number of ribosomal RNA (rRNA), transfer RNA (tRNA), open reading frames (ORFs), and KEGG-annotated ORFs, are given in Supplementary Table 8. Transcriptomic sequencing was performed after co-culturing these two strains in D-gluconic acid. The related data have been presented in Supplementary Table 9. Subsequently, gene transcription profiles of the two strains were visualized individually based on the Reads Per Kilobase Million (RPKM). Nearly all KEGG-annotated genes (total 1507) were expressed in both strains, while 614 and 330 genes were specifically expressed in S1616 and M597, respectively (Fig. 5a and 5b). Students’ t-test of 1507 KEGG genes revealed 82 significantly overexpressed (P-value < 0.05 and fold change > 2) genes in the S1616 transcriptome, and 297 significantly overexpressed genes in the M597 transcriptome (Fig. 5c). Regarding amino acid metabolism, 10 genes were overexpressed in S1616 (e.g., K01733, thrC, threonine synthase) while 25 were overexpressed in M597 (e.g., K01920, gshB, glutathione synthase) (Fig. 5d). Reconstruction of the amino acid metabolism-related pathways in the genomic and transcriptomic datasets revealed complete pathways for metabolism of a variety of amino acids, including the transformation of threonine, alanine, and phenylalanine to acetyl-CoA, as well the transformation of citrulline, arginine, glutamine, and glutamate to 2-oxoglutarate (Supplementary Fig. 9). These pathways suggested that these amino acids can probably enter the central carbon metabolism. In addition, pathways for the transformation of intermediates to several amino acids during the central carbon metabolism were also annotated (Supplementary Fig. 9). All these metabolomic, genomic, and transcriptomic results suggest that there is an extracellular amino acid pool, which allows the exchange of amino acids between the two bacterial strains, thus promoting their growth (Fig. 5e).

Microbial coexistence in in situ marine biofilms

Finally, the prevalence and coexistence of the two cross-feeding strains, S1616 and M597, in marine biofilms were assessed. This was done by examining the distribution pattern of these strains in biofilm metagenomes collected from the Pacific Ocean, the Indian Ocean, and the Atlantic Ocean to obtain a global ecological profile. Total 131 biofilms were collected, including 18 on metal panels¹⁵, 13 on rock surfaces¹⁵, 76 on plastic panels¹⁵, and 24 on microplastic or glass particles¹¹. The detailed information of these biofilms, as well as the metagenomic datasets, have been presented in Supplementary Table 10. The results revealed substantial proportion of both strains in nearly all biofilm communities, with maximum relative abundance of 1% and 2% for S1616 and M597, respectively (Supplementary Fig. 10). Interestingly, correlation analysis indicated significant coexistence (Spearman Rho > 0.5 and P-value < 0.05) between these two strains detected in biofilms on all the four substrate types (Supplementary Fig. 11).

Microbial community assembly is a long-standing concern to microbiology²². Here, a systematic metagenomic study was conducted on the influence of carbon source on the assembly of marine biofilm communities. The results showed that the taxonomic composition of the cultured biofilms was not primarily determined by the type of carbon source, and instead, strong bacterial correlation was common across these communities. Subsequently, bacterial isolation, co-culture experiments, 16S rRNA amplicon sequencing, microbial genomics, metabolomics, and transcriptomics were conducted to show that cross-feeding based on amino acid metabolism expanded the range of available carbon sources, thus affecting the structures biofilm community assembly.

Several studies^1,2,3 have supported the influence of carbon source on community assembly. On the contrary, this study showed that, from a more comprehensive perspective, the taxonomic composition of stable marine biofilms could not be classified or predicted according to the type of carbon source. This notion is generally in line with our previous understanding of the principles governing the biofilm development in marine environments^11,23. By deploying six different types of man-made panels into a cold seep environment, it was shown that the effect of substrate type on the presence of microbial species in biofilm communities decreased with biofilm development²³. Investigating biofilm assembly on microplastics also revealed a stronger correlation between substrate size (i.e., the diameter of microplastics) and taxonomic composition in earlier stages, rather than in later stages¹¹. Similarly, a study²⁴ on microbial colonization of chitin particles in the marine environment found that the chitin-degrading taxa were only accumulated selectively during the earlier stages of community assembly, whereas late comers were unable to use the chitin resource. Hence, abiotic factors, including the carbon source, exert effects primarily during initial stages of the biofilm development, while internal factors become increasingly important as the process of biofilm development continues. This is also in agreement with the findings reported in our global ocean biofilm study¹⁵, which demonstrated the existence of a functional “core” in marine biofilms across different locations and substrates.

Mechanistically, this study showed that microbial interactions facilitated the utilization of a broader range of carbon sources, thus reducing the selective force of carbon source on microbial taxa. This finding is consistent with a recent report²⁵ demonstrating that microbial interaction can expand their metabolic niche space, which is especially true for two specialist taxa. However, partially diverging fact from this finding is that both niche expansion and contraction occurred for the tested strains²⁵, whereas here we found that nearly all the pairwise interactions were positive. Moreover, for the carbon source that can support individual bacterial growth, co-culturing with another strain led to the accumulation of more cell density and higher carbon consumption rate. Therefore, in marine biofilms, positive interactions are expected to be more prevalent than in other microbial model systems.

Furthermore, the observed microbial interactions may be attributed to cross-feeding, which allows strains to directly benefit from the biochemical transformation capabilities of each other through an “overflow metabolism”. Microbial cross-feeding has been reported in previous studies^26,27, which suggested remarkable levels of complementary metabolisms between different bacterial species. For instance, in the insect sharpshooters, two bacterial symbionts, Baumannia and Sulcia, are broadly complementary²⁶. While Sulcia contributes to amino acid biosynthesis, Baumannia is responsible for cofactor and vitamin synthesis²⁶. Here, it was shown that biofilm inhabitants that interact with each other may cooperate based on feeding of different amino acids. The formation of an extracellular “amino acid pool” allows the exchange of various amino acids between the two microbial partners, probably establishing a mutualism relationship. It seems that the “consumption” or “secretion” of certain amino acids is not invariable for a given bacterium. Instead, it may be dependent on the carbon source concentration. This notion is derived from the observation that when cultured individually, both S1616 and M597 consumed alanine present in the added co-culture supernatant. Whereas it must be a different case when they were co-cultured; otherwise the alanine could not be accumulated in the co-culture supernatant. In addition, the presence of similar amino acid metabolism pathways in S1616 and M597, as well as the differential expression levels of these genes between the two strains, indicate more significant roles of gene expression regulation in mediating cross-feeding, than genomic complementation.

Finally, the correlation was shown between the two cross-feeding strains in in situ marine biofilms, implying ubiquity in the ecological framework. Considering the existence of a variety of carbon sources in the marine environment, this finding also supports the key conclusion that environmental carbon source could not dictate the presence of key biofilm members. However, limitations still exist in current study. For example, only one pair of coexisted strains was selected for metabolic and molecular study. To address this limitation, understanding the cross-feeding mechanisms of other bacterial pairs would be our future research direction. However, considering the essential role of amino acids in bacterial growth, cross-feeding mediated by amino acid metabolism is speculated to be a widely applicable mechanism.

Overall, this study shows how microorganisms’ assembly into marine biofilms is not controlled by the environmental carbon source. In contrast, internal microbial interactions appear to be a strong and constant force in shaping the taxonomic composition of biofilms. We further demonstrated the novel mechanisms mediating the microbial interaction and utilization of broader range of carbon sources. These findings shed new light on marine microbial ecology and are crucial for efficient utilization of marine biofilms.

Biofilm sampling and enrichment culture

The original community used for enrichment culture was sampled on May 28, 2021 from the subtidal zone around the Maidao Island, Qingdao, China (36.05 N, 120.43 E). The sampling site was connected to the open ocean and the impact of human activities on the site was low. The biofilms on immersed stone surfaces were scraped off using sterile cotton tips and immediately transferred to laboratory. The bacterial cells were washed twice (3687 g, 5 minutes) with sterile seawater to remove excess carbon sources. Fungi, algae, and other irrelevant particles were filtered out using 3-µm cellulose ester membrane filters (Millipore, USA). The bacterial cells were kept at 4 ℃ for 16 hours to allow the consumption of the remained carbon sources.

Liquid media for enrichment culture were prepared by adding sole carbon sources (0.1%, w/v) into a solution containing macro and micro elements. The carbon sources were filtered through 0.22-µm polycarbonate membrane filters (Millipore, USA) to prevent bacterial contamination. The media were autoclaved and the pH was adjusted to 6.5–7.5 and stabilized by adding HEPES (1 mol/L). Afterwards, the media were seeded by starved biofilm bacteria and cultured in glass containers in a static state at 25 ℃ in dark. For aerobic cultivation, the head space of the glass containers was filled with 80% nitrogen and 20% oxygen, while for anaerobic growth, the head space was occupied by high-purity nitrogen (99.999%). During the enrichment culture, biofilms formation occurred at the bottom of the containers and the consortia were harvested when the cell density reached around OD₆₀₀ 1.0 after 20–30 days. The samples used for DNA extraction were stored at -80 ℃ after freezing with liquid nitrogen. Part of each sample was stored in 25% glycerol for bacterial isolation.

Metagenomic sequencing and analyses of the enriched microbiota

DNA extraction was performed using the QIAamp PowerFecal DNA Kit (Qiagen, Germany). The efficiency of several DNA extraction kits were tested and this kit showed the best performance in DNA extraction from marine biofilms. The quality of extracted DNA was examined using gel electrophoresis and NanoDrop Lite spectrophotometer (Thermo Fisher Scientific, USA). The 138 samples were sequenced individually by Illumina metagenomic sequencing. DNA libraries with 350 bp insert length were prepared using the NEBNext® Ultra™ DNA Library Prep Kit (New England Biolabs, USA) following manufacturer’s protocol. Sequencing was performed on the DNASEQ-T7 platform at Annoroad Gene Technology Co., LTD., China. Paired-end sequencing program was conducted to generate more than 20 Gb clean data per sample. DNA samples of the communities grown on same type of carbon sources were pooled together for Nanopore metagenomic sequencing, resulting in four metagenomes (i.e., alcohols, amino acids, organic acids, and saccharides). To construct sequencing library, G-tube was used to fragment nucleic acid. NEBNext FFPE DNA Repair Mix (New England Biolabs, USA) and NEBNext Ultra II End Repair/dA-Tailing Module (New England Biolabs, USA) were used for damage repair and end repair of the nucleic acid fragments by adding poly A. Then, different barcodes were assigned to the samples by using a Native Barcode Expansion kit (Oxford Nanopore Technologies, UK). Sequencing was performed on the PromethION48 platform at the Annoroad Gene technology co., LTD. to generate more than 10 Gb clean data per metagenome.

Metagenomic analyses were performed following the steps used in our previous study on marine biofilms¹⁵ in a Linux system installed in our local server. Quality control of the Illumina metagenomic reads was performed using the software NGS QC Toolkit (version 2.0)²⁸. Adaptor-containing reads, low-quality (quality score < 20) reads, and unpaired reads were filtered out. For taxonomic classification, miTags were extracted from the metagenomic reads using Parallel-META3 software²⁹. For OTU table construction and taxonomic structure profiling, the miTags were clustered at 97% similarity and annotated by mapping to a database that integrates GreenGenes³⁰, RDP³¹, and SILVA³² using Bowtie (version 2.4.2)³³. OTUs detected in both forward and reverse paired-end reads in a metagenome were selected for further steps. PCoA was performed at phylum (class level for Proteobacteria), family, and genus levels based on the Bray-Curtis distance using PAST (version 2.0)³⁴. To validate the miTag results, protein-coding marker gene fragments were extracted from the metagenomes and subjected to taxonomic profiling by using MetaPhlAn3 software (version 3.0)³⁵.

Genome binning and annotation

Prior to genome binning, the pairs of Illumina metagenomes derived from aerobic and anaerobic conditions in the presence of same carbon source were assembled into contigs using MEGAHIT (version 1.2.9)³⁶. Thus, total 69 assemblages were generated. Metagenome binning was performed on contigs longer than 5000 bp using MaxBin (version 2.2.6)³⁷. The completeness and contamination of the resultant MAGs were evaluated by using CheckM (version 1.1.2)³⁸. MAGs with contamination less than 5% and completeness more than 80% were selected for the second-step binning using MetaBAT (version 0.32.5)³⁹. MAGs that meet the above-mentioned criteria were further purified by running the software Concoct (version 1.1.0)⁴⁰. All MAGs obtained from the previous three steps were subjected to reassembly with the corresponding long reads generated by the Nanopore sequencing. The raw MAGs were used as index and the Illumina reads were mapped onto the raw MAGs index using Bowtie³³. The matched reads were extracted using Samtools (version 1.11)⁴¹. The Nanopore reads were used for BLASTn search against the raw MAGs and the matched reads with more than 99% identity were retained. Subsequently, the Illumina and Nanopore reads matched to a given MAG were cross assembled using the software SPAdes (version 3.12.0)⁴² in a single genome mode with default kmers. The reassembled MAGs were further evaluated using CheckM and subjected to de-replication at 99% average nucleotide identity using dRep (version 2.6.2)⁴³. Eventually, 535 non-redundant MAGs with contamination less than 5% and completeness over 90% were obtained. Finally, the assembly quality assessment of the contigs and MAGs was done using QUAST (version 5.0.2)^44,45.

For taxonomic classification of the MAGs, GTDB-Tk (version 0.3.2)⁴⁶ was adopted to annotate the MAGs at the phylum (class for Proteobacteria), family, and genus levels. For phylogenetic analysis, PhyloPhlAn (version 3.0.67)⁴⁷ was used to build a maximum-likelihood tree, which showed the evolutionary relationships among 535 MAGs based on clade-specific phylogenetic markers. MEGA (version 7.0.26)⁴⁸ was used to visualize the tree. For functional annotation and comparison, ORFs of the MAGs were predicted using Prodigal (version 2.60)⁴⁹ and were searched against the KEGG database (2022 version; purchased from the Kanehisa Laboratory, Japan) using BLASTp (E-value < 1e-7). Subsequently, a KEGG-MAG matrix was built by determining the number of genes included in the KEGG orthologs across the 535 MAGs. The KEGG orthologs belonging to strongly-correlated and non-correlated MAGs were compared using Student’s t-test and Hypergeo-test to determine the significantly changed functions.

Ecological network construction

To calculate the relative abundance of these MAGs in the enrichment cultured communities, Illumina reads of the 138 metagenomes were normalized to 1,000,000 reads per sample and mapped to the MAGs using BBMap (version 35.85) with minimum alignment identity of 0.80⁵⁰. Relative abundances of MAGs in the communities were calculated by counting the numbers of mapped reads. The abundance distribution heatmap of the MAGs for different carbon sources and in aerobic and anaerobic conditions was plotted by using the “ComplexHeatmap”^51,52 package of R (version 4.2.1). The Spearman correlation was also calculated by R software with “picante”⁵³, “reshape2”⁵⁴, “psych”⁵⁵, and “dplyr”⁵⁶ packages. P-values were adjusted by “fdr” to reduce false positive rate (FDR). The valid correlations were selected when Spearman Rho (correlation coefficient) between bacterial was greater than 0.9 or less than − 0.9 and the P-value was less than 0.05. The network was visualized using the Gephi software (version 0.9.2)⁵⁷.

Bacterial isolation and co-culture experiments

Bacterial isolation was performed by using the enrichment culture samples stored in glycerol at -80 ℃. The samples were allowed to melt slowly at room temperature and diluted at a series of concentrations (10⁻, 100⁻, 1,000⁻, and 10,000⁻ times) using marine broth 2216E liquid medium (Becton Dickinson and Company, USA). Subsequently, 100 µL aliquots were spread on marine 2216E agar plates (Becton Dickinson and Company, USA) and incubated at 25°C for 24–72 hours. Obtained colonies were examined under a dissecting microscope for analysis of morphological characteristics. Colonies of conspicuous types were picked. After sub-culturing for more than eight generations on marine 2216E agar plates, the pure colonies were seeded into 2216E liquid medium for genomic DNA extraction and taxonomic identification. Genomic DNA was extracted using the TIANamp Genomic DNA Kit (Tiangen Biotech, China), following the manufacturer's protocol. PCR was conducted to amplify the 16S rRNA genes of the extracted DNA using the universal primers 27F (5’-AGAGTTTGATCCTGGCTCAG-3’) and 1492R (5’-GGTTACCTTGTTACGACTT-3’). The 50-µL PCR mix contained 1 µL DNA, 1 µL primers, 17 µL sterile water and 20 µL Taq DNA Polymerase (Sangon Biotech, China). Sanger sequencing was performed at the Beijing Genomics Institute (BGI, China). The 100 bp of unstable base sequences from front and back end were removed to get sequences with higher accuracy. Then the 16S rRNA gene sequence was subjected to the NCBI online website for taxonomic classification. Redundant strains were identified using the software cd-hit with a 16S identity cut-off of 0.999. The identified redundant strains were not used for further study.

Six strains belonging to Pseudoalteromonas, Microbacterium, Roseibium, Leisingera, Sulfitobacter, and Yangia genera were selected for co-culture experiments (corresponding MAGs were present in the constructed network). These strains were initially cultured in marine broth 2216E media and washed twice using carbon free media (3687 g, 5 minutes) to remove the residual carbon sources. Subsequently, the strains were monocultured and co-cultured in pair (two strains with approximately equal cell density) in presence of 21 single carbon sources (0.1%). The other components of the media were prepared following the steps described for the community enrichment culture experiment. This experiment was conducted with three biological replicates.

Complete genomic sequencing and analyses

Complete genome sequencing was performed for the two strains, S1616 and M597. In detail, DNA of these two strains cultured individually in marine broth 2216 was extracted using the TIANamp Genomic DNA Kit (Tiangen Biotech, China). DNA integrity was examined by agarose gel electrophoresis. Library construction for Illumina sequencing was conducted using the NEBNext® UltraTM DNA Library Prep Kit (New England Biolabs, USA). Illumina sequencing was performed on the NovaSeq platform at the Novogene Bioinformatics Institute, China to generate 2 Gb data per strain. PacBio single-molecular real-time (SMRT) sequencing was conducted for library construction using the SMRT bell TM Template kit (version 1.0) to obtain 10-kbp SMRT Bell library. Subsequently, the sequencing of library was performed at Novogene following the circular consensus sequencing (CCS) strategy to generate 1 Gb data per strain. Preliminary genome assembly was performed with SMRT Link v5.0.1 (CCS = 3, minimum accuracy > 0.99) using the PacBio data. Assembled genomes were corrected by the Illumina data using Minimap2 to complete the genome sequencing. Taxonomic classification was done by using GTDB-Tk⁴⁶. The rRNA and tRNA genes in the genomes were predicted using Barrnap (version 0.9, https://github.com/tseemann/barrnap) and Aragorn⁵⁸, respectively. ORFs and corresponding proteins were predicted using Prodigal⁴⁹ in a close-end ORF prediction model. For functional gene annotation, ORFs were used to run BLASTp (E-value < 1e-7) search against the KEGG database, followed by metabolic pathway reconstruction in KEGG Mapper (https://www.genome.jp/kegg/mapper.html).

Taxonomic, metabolomic, and transcriptomic analyses of the co-cultures

To determine the relative abundance of S1616 and M597 in the co-cultured consortia, 16S rRNA gene amplicon sequencing was performed on samples collected on day 0, 2, 4, 6, 8, 10, 12, and 14. DNA from the co-cultures were extracted using the TIANamp Genomic DNA Kit. The V3-V4 regions of the 16S rRNA genes were amplified using the universal primers 16S-341F (5′-CCTACGGGNGGCWGCAG-3′) and 16S-805R (5′-GACTACHVGGGTATCTAATCC-3′) with PlatinumTM Taq DNA Polymerase (DNA free; Thermo Scientific, USA). Libraries were constructed using the TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, USA) and sequencing was conducted on an Illumina PE250 platform in Novogene (Beijing, China). A depth of 10,000 pair-ended reads per sample was generated and merged using FLASH⁵⁹. The merged sequences were searched against the Silva database³² using UCHIME algorithm⁶⁰ for chimera removal. Furthermore, the 16S rRNA gene sequences were classified by mapping with the 16S rRNA genes from the S1616 and M597 genomes using BLASTn (E-value < 1e-7 and identity > 99%). Relative abundance of these two strains were indicated by the numbers of amplicon reads that were normalized to the 16S rRNA gene copy numbers in genomes. This experiment was conducted with two biological replicates.

For metabolomics, the supernatant of co-cultured S1616 and M597 in D-gluconic acid was collected after the consortium entered the stationary phase (144 hours). The culture was subjected to centrifugation at 1,792 g for 5 minutes to remove bacterial cells, followed by filtering through 0.22-µm filter (Millipore, USA). The collected supernatant was amended with 0.1% (v/v) of D-gluconic acid and then seeded with S1616 or M597 for monoculture. After 48 hours of growth, cell density of the two monocultured strains were measured at OD₆₀₀ and the supernatants were collected following the same steps described above. Then, concentration of amino acids in the supernatants of the two monocultures and co-culture were determined by using a 6470A Triple Quadrupole mass spectrometry (Santa Clara, USA) linked to an Agilent 1290 Infinity II UHPLC system (China). The samples were spiked with 10 mL of 6 M HCl and 0.5% β-mercapto ethanol, followed by sealing under nitrogen, and hydrolyzation at 110°C. Total 100 µL of the hydrolysate was dried by adding nitrogen and HCl. The samples were then dissolved in 100 mM sodium carbonate and derivatized using 2% benzoyl chloride. The samples were isometrically mixed with internal standard solutions and subjected to UHPLC-MS/MS analysis. The data were analyzed using the MassHunter Workstation Software (version B.08.00; Agilent, USA). This experiment was conducted with three biological replicates.

For transcriptomics, S1616 and M597were co-cultured in D-gluconic acid up to the early stationary phase (144 hours). Bacterial cells in the cultures were harvested by centrifugation at 3687 g for 10 minutes. Harvested cells were immediately freezed in liquid nitrogen. The RNAprep Pure Cell/Bacteria Kit (Tiangen Biotech, China) was used to extract total RNA following the manufacturer’s protocol. The NEBNext Ultra RNA Library Prep Kit was used to prepare the RNA-Seq library. After quality control, the libraries were sequenced using the Novaseq 6000 system at Novogene (China). To this end, 150-bp paired-end reads were generated and 4 Gb data were obtained for each biological replicate. The NGS QC Toolkit²⁸ was used to filter out the low-quality (quality score < 20) reads from the raw transcriptomic data. ORFs derived from the S1616 and M597 genomes were indexed separately using Bowtie³³. The number of reads mapped to each ORF was summarized using Samtools⁴¹. Afterwards, RPKM values for genes were calculated using the following formula:

RPKM = number of recruited reads / (gene length/1000 × total number of reads/1,000,000)

For ORFs expressed in both strains, significant expression differentiation was identified using Students’ t-test (P-value < 0.05 and fold change of RPKM > 2). The heatmap of the differentially expression genes related to amino acid metabolism (based on the KEGG annotation) was drawn by using Cluster 3.0 (hierarchical clustering with average linkage)⁶¹ and Java TreeView⁶² This experiment was conducted with three biological replicates.

Metagenomic analyses of the coexisted strain pair

The distribution patterns of S1616 and M597 in 131 marine biofilms were analyzed by integrating the genomic and metagenomic analyses. The metagenomic datasets were generated in our previous studies^11,15,16. These metagenomes were classified into four groups based on the substrates of the biofilms. The normalized metagenomic reads (read length of 101 bp and 1,000,000 reads per metagenome) were mapped to the chromosomes of S1616 and M597 using the software BBmap (version 35.85)⁵⁰. The percentages of these two genomes were indicated by the recruited reads, and Spearman correlations in the four biofilm groups were determined by R software as mentioned above.

Statistics and reproducibility

Two-tailed Students’ t-test was performed in EXCEL after confirmation of normal distribution using the Levene's test. Hypergeo-test was performed using STAMP software (version 2.1.3)⁶³. The P-values generated in Students’ t-tests and Hypergeo-tests were adjusted based on the FDR, using the “fdr” function in R⁶⁴. All experiments were performed for at least two batches, and each batch contained two or three independent biological replicates.

Data availability

The metagenomes (138 samples) generated in this study have been deposited in CNCB under accession code CRA006934 (https://bigd.big.ac.cn/gsa/browse/CRA006934). The MAGs (n = 535) generated in this study have been uploaded to the Genome Warehouse in CNCB (https://ngdc.cncb.ac.cn/gwh). The complete genome of S1616 generated in this study has been deposited in NCBI under accession code CP128601-CP128604 (https://www.ncbi.nlm.nih.gov/nuccore/CP128601). The metabolomes data (nine samples) generated in this study have been deposited in CNCB under accession code OMIX004392 (https://ngdc.cncb.ac.cn/omix/release/OMIX004392). The transcriptomic datasets (three samples) generated in this study have been deposited in CNCB under accession code CRA011180 (https://bigd.big.ac.cn/gsa/browse/CRA011180). Other datasets, including the global ocean biofilm metagenomes and the complete genome of M597, have been published associated with our previous studies^11,15,16,18.

Acknowledgements

The authors are grateful to Prof. Weibo Song for Figure editing. This work was supported by the overseas top level talents funding (862105020028 and 862205033005) to Dr. Weipeng Zhang.

Author contributions

W.Z. designed the project and wrote the manuscript; H.C. and S.F. performed the major part of data analyses and experiments; M.S., J.Z., H.Z., S.W. and X.S. were involved in experiments; W.D., R.T., and Y.Z.Z. provided technical supports and comments; all the authors agreed to the final version.

Competing interests

The authors declare no competing interests.

Wawrik, B., Kerkhof, L., Kukor, J. & Zylstra, G. Effect of different carbon sources on community composition of bacterial enrichments from soil. Appl. Environ. Microb. 71, 6776–6783 (2005).
Fu, H., Uchimiya, M., Gore, J. & Moran, M. A. Ecological drivers of bacterial community assembly in synthetic phycospheres. PNAS 117, 3656–3662 (2020).
Goldford, J. E. et al. Emergent simplicity in microbial community assembly. Science 361, 469–474 (2018).
Grokopf, T. & Soyer, O. S. Synthetic microbial communities. Curr. Opin. Microbiol. 18, 72–77 (2014).
Coyte, K. Z., Schluter, J. & Foster, K. R. The ecology of the microbiome: networks, competition, and stability. Science 350, 663–666 (2015).
Kehe, Jared. et al. Positive interactions are common among culturable bacteria. Sci. Adv. 7, 7159 (2021).
LaRowe, D. E. et al. The fate of organic carbon in marine sediments-New insights from recent data and analysis. Earth-Sci Rev. 204, 103146 (2020).
Martin, R. E. & Servais, T. Did the evolution of the phytoplankton fuel the diversification of the marine biosphere? Lethaia 53, 5–31 (2020).
Flemming, H. C. & Wuertz, S. Bacteria and archaea on Earth and their abundance in biofilms. Nat. Rev. Microbiol. 17, 247–260 (2019).
Ding, W. et al. Metagenomic analysis of zinc surface–associated marine biofilms. Microb. Ecol. 77, 406–416 (2019).
Qin, P. et al. Early stage of biofilm assembly on microplastics is structured by substrate size and bacterial motility. iMeta e121, 1–11 (2023).
Lu, J. et al. The landscape of global ocean microbiome: From bacterioplankton to biofilms. Int. J. Mol. Sci. 24, 6491 (2023).
Flemming, H. C. & Wingender, J. The biofilm matrix. Nat. Rev. Microbiol. 8, 623–633 (2010).
Bassler, B. L. & Losick, R. Bacterially speaking. Cell 125, 237–246 (2006).
Zhang, W. et al. Marine biofilms constitute a bank of hidden microbial diversity and functional potential. Nat. Commun. 10, 1–10 (2019).
Ding, W. et al. Expanding our understanding of marine viral diversity through metagenomic analyses of biofilms. Mar. Life Sci. Tech. 3, 395–404 (2021).
Su, X., Cui, H. & Zhang, W. Copiotrophy in a marine-biofilm-derived Roseobacteraceae bacterium can be supported by amino acid metabolism and thiosulfate oxidation. Int. J. Mol. Sci. 24, 8617 (2023).
Ding, W. et al. Anaerobic thiosulfate oxidation by the Roseobacter group is prevalent in marine biofilms. Nat. Commun. 14, 2033 (2023).
Schmid, M., Raschbauer, M., Song, H., Bauer, C. & Neureiter, M. Effects of nutrient and oxygen limitation, salinity and type of salt on the accumulation of poly (3-hydroxybutyrate) in Bacillus megaterium uyuni S29 with sucrose as a carbon source. New Biotechnol. 61, 137–144 (2021).
Liang, K. Y. H., Orata, F. D., Boucher, Y. F. & Case, R. J. Roseobacters in a sea of poly- and paraphyly: Whole genome-based taxonomy of the family Rhodobacteraceae and the proposal for the split of the “Roseobacter Clade” into a novel family, Roseobacteraceae fam. nov. Front. Microbiol. 12, 683109 (2021).
Deng, Y. et al. Network succession reveals the importance of competition in response to emulsified vegetable oil amendment for uranium bioremediation. Environ. Microbiol. 18, 205–218 (2016).
Burke, C., Steinberg, P., Rusch, D. & Thomas, K. T. Bacterial community assembly based on functional genes rather than species. PNAS 108, 14288–14293 (2011).
Zhang, W. et al. Species sorting during biofilm assembly by artificial substrates deployed in a cold seep system. Sci. Rep. 4, 1–7 (2014).
Datta, M.S., Sliwerska, E., Gore, J., Polz, M.F., & Cordero, O.X. Microbial interactions lead to rapid micro-scale successions on model marine particles. Nat. Commun. 7, 11965 (2016).
Oña, L. et al. Obligate cross-feeding expands the metabolic niche of bacteria. Nat. Ecol. Evol. 5, 1224–1232 (2021).
McCutcheon, J. P. & Moran, N. A. Parallel genomic evolution and metabolic interdependence in an ancient symbiosis. PNAS 104, 19392–19397 (2007).
McCutcheon, J. P. & Von Dohlen, C. D. An interdependent metabolic patchwork in the nested symbiosis of mealybugs. Curr. Biol. 21, 1366–1372 (2011).
Patel, R. K. & Jain, M. NGS QC Toolkit: a toolkit for quality control of next generation sequencing data. PLoS One 7, e30619 (2012).
Jing, G. et al. Parallel-META 3: Comprehensive taxonomical and functional analysis platform for efficient comparison of microbial communities. Sci. Rep. 7, 40371 (2017).
DeSantis, T. Z. et al. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microb. 72, 5069–5072 (2006).
Cole, J. R. et al. Ribosomal Database Project: data and tools for high throughput rRNA analysis. Nucleic Acids Res. 42, 633–642 (2013).
Quast, C., Pruesse, E., Yilmaz, P., Gerken, J. & Glckner, F. O. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41, 590–596 (2012).
Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012).
Hammer, Øyvind., Harper, D. A. T. & Ryan, P. D. Past: Paleontological statistics software package for educaton and data analysis. Palaeontol. Electron. 4, 4–9 (2001).
Beghini, F. et al. Integrating taxonomic, functional, and strain-level profiling of diverse microbial communities with biobakery 3. eLife 10, e65088 (2021).
Li, D., Liu, C. M., Luo, R., Sadakane, K. & Lam, T. W. MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31, 1674–1676 (2015).
Wu, Y. W., Simmons, B. A. & Singer, S. W. MaxBin 2.0: an automated binning algorithm to recover genomes from multiple metagenomic datasets. Bioinformatics 32, 605–607 (2016).
Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. & Tyson, G. W. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–1055 (2015).
Kang, D. D., Froula, J., Egan, R. & Wang, Z. MetaBAT, an efficient tool for accurately reconstructing single genomes from complex microbial communities. Peer J. 3, e1165 (2015).
Alneberg, J. et al. Binning metagenomic contigs by coverage and composition. Nat. Methods 11, 1144–1150 (2014).
Li, H., et al. The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).
Olm, M. R., Brown, C. T., Brooks, B. & Banfield, J. F. dRep: a tool for fast and accurate genomic comparisons that enables improved genome recovery from metagenomes through de-replication. ISME J. 11, 2864–2868 (2017).
Gurevich, A., Saveliev, V., Vyahhi, N. & Tesler, G. QUAST: quality assessment tool for genome assemblies. Bioinformatics 29, 1072–1075 (2013).
Mikheenko, A., Saveliev, V. & Gurevich, A. Metaquast: evaluation of metagenome assemblies. Bioinformatics 32, 1088–1090 (2016).
Chaumeil, P. A., Mussig, A. J., Hugenholtz, P. & Parks, D. H. GTDB-Tk: a toolkit to classify genomes with the genome taxonomy database. Bioinformatics, 36, 1925–1927 (2019).
Segata, N., Börnigen, D., Morgan, X. C. & Huttenhower, C. Phylophlan is a new method for improved phylogenetic and taxonomic placement of microbes. Nat. Commun. 4, 2304 (2013).
Kumar, S., Stecher, G. & Tamura, K. MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets. Mol. Biol. Evol. 33, 1870–1874 (2016).
Hyatt, D. et al. Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics 11, 119 (2010).
Bushnell, B. BBMap: a fast, accurate, splice-aware aligner. Lawrence Berkeley National Lab (LBNL), Berkeley, CA, USA, (2014).
Gu, Z., Eils, R. & Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 32, 2847–2849 (2016).
Gu, Z. Complex heatmap visualization. iMeta 1, e43 (2022).
Kembel, S. W. et al. Picante: R tools for integrating phylogenies and ecology. Bioinformatics 26, 1463–1464 (2010).
Wickham, H. Reshape2: flexibly reshape data: a reboot of the reshape package. J. Stat. Softw. 12, 1–20 (2007).
Revelle, W. & Condon, D. M. Reliability from alpha to omega: A tutorial. Psychol. Assess. 31, 1395–1411 (2018).
Wickham, H., François, R., Henry, L., Müller, K. & Vaughan, D. Dplyr: A grammar of data manipulation (UTC. Press, 2019).
Bastian, M., Heymann, S. & Jacomy M. Gephi: an open source software for exploring and manipulating networks. Proceedings of the Third International Conference on Weblogs and Social Media, ICWSM 3, 361–362 (2009).
Laslett, D. & Canback, B. ARAGORN, a program to detect tRNA genes and tmRNA genes in nucleotide sequences. Nucleic Acids Res. 32, 11–16 (2004).
Magoč, T. & Salzberg, S. L. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963 (2011).
Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C., & Knight, R. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27, 2194–2200 (2011).
Hoon, M. J. L. D., Imoto, S., Nolan, J. & Miyano, S. Open source clustering software. Bioinformatics 20, 1453–1454 (2004).
Saldanha, A. Java Treeview-extensible visualization of microarray data. Bioinformatics 20, 3246–3248 (2004).
Parks, D. H., Tyson, G. W., Hugenholtz, P. & Beiko, G. STAMP: Statistical analysis of taxonomic and functional profiles. Bioinformatics 30, 3123–3124 (2014).
Strimmer, K. A unified approach to false discovery rate estimation. BMC Bioinformatics 9, 303 (2008).

There is NO Competing Interest.

SupplementaryTables.xlsx
Supplementary Tables 1-10
SupplementaryFigures.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Mutual cross-feeding drives marine biofilm assembly on various carbon sources

Status:

Version 1

Abstract

Figures

Introduction

Results

Extensive microbial coexistence across the enriched biofilm communities

Positive interactions among biofilm-derived culturable strains

Metabolic and molecular bases for cross-feeding between two representative strains

Discussion

Materials and Methods

Biofilm sampling and enrichment culture

Metagenomic sequencing and analyses of the enriched microbiota

Genome binning and annotation

Ecological network construction

Bacterial isolation and co-culture experiments

Complete genomic sequencing and analyses

Taxonomic, metabolomic, and transcriptomic analyses of the co-cultures

Metagenomic analyses of the coexisted strain pair

Statistics and reproducibility

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1