Metagenomic 16S rDNA Illumina tags are a powerful alternative to amplicon sequencing to explore diversity and structure of microbial communities
Sequencing of 16S rDNA polymerase chain reaction (PCR) amplicons is the most common approach for investigating environmental prokaryotic diversity, despite the known biases introduced during PCR. Here we show that 16S rDNA fragments derived from Illumina‐sequenced environmental metagenomes (ₘᵢtags)...
Saved in:
| Published in: | Environmental microbiology Vol. 16; no. 9; pp. 2659 - 2671 |
|---|---|
| Main Authors: | , , , , , , , , , , , , , , , , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
England
Blackwell Science
01-09-2014
Blackwell Publishing Ltd Wiley Subscription Services, Inc Society for Applied Microbiology and Wiley-Blackwell |
| Subjects: | |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Sequencing of 16S rDNA polymerase chain reaction (PCR) amplicons is the most common approach for investigating environmental prokaryotic diversity, despite the known biases introduced during PCR. Here we show that 16S rDNA fragments derived from Illumina‐sequenced environmental metagenomes (ₘᵢtags) are a powerful alternative to 16S rDNA amplicons for investigating the taxonomic diversity and structure of prokaryotic communities. As part of the Tara Oceans global expedition, marine plankton was sampled in three locations, resulting in 29 subsamples for which metagenomes were produced by shotgun Illumina sequencing (ca. 700 Gb). For comparative analyses, a subset of samples was also selected for Roche‐454 sequencing using both shotgun (ₘ₄₅₄tags; 13 metagenomes, ca. 2.4 Gb) and 16S rDNA amplicon (₄₅₄tags; ca. 0.075 Gb) approaches. Our results indicate that by overcoming PCR biases related to amplification and primer mismatch, ₘᵢtags may provide more realistic estimates of community richness and evenness than amplicon ₄₅₄tags. In addition, ₘᵢtags can capture expected beta diversity patterns. Using ₘᵢtags is now economically feasible given the dramatic reduction in high‐throughput sequencing costs, having the advantage of retrieving simultaneously both taxonomic (Bacteria, Archaea and Eukarya) and functional information from the same microbial community. |
|---|---|
| Bibliography: | http://dx.doi.org/10.1111/1462-2920.12250 istex:C850ABD937EC35C15E879C69670FEEC35F82FB55 ark:/67375/WNG-5DQ5SZ77-8 MicroOcean PANGENOMICS - No. CGL2011-26848/BOS ArticleID:EMI12250 Fig. S1. Pipeline flowchart showing all steps performed for quality filtration, classification and OTU assignation of mitags. Fig. S2. Coverage analysis of mitags against one reference 16S rDNA sequence from Escherichia coli (Brosius et al., 1978). The upper part of the figure shows the distribution of mitags (horizontal lines) along the 16S rRNA gene, taking into consideration the nucleotide coordinates and the percentage of identity (left scale) with regard to the reference gene as well as their density (grey scale; clear grey indicates low density). The lower part of the figure shows the coverage as the number of times that a nucleotide position in the reference gene is covered by a mitag (right scale). Horizontal red lines were plotted in accordance with the nucleotide coordinates of the hypervariable regions (V1 to V9). Fig. S3. RDP classification. Mean classification confidence values across taxonomic levels calculated with RDP classifier for mitags, m454tags and 454tags. The dashed lines indicate the standard deviation (±). Only confidence values > 0.5 were considered. Fig. S4. Rarefaction analysis of 16S mitags for the size fractions 0.2-1.6 and 0.8-5 μm. All mitags were used, covering all possible V regions of the 16S rRNA gene. Fig. S5. Venn diagrams. miTags are indicated in salmon and 454tags in green. A. Shared and unique taxonomic ranks (phylum, class, order, family and genus) recovered with mitags and 454tags using the entire dataset (i.e. mitags spanning the entire 16S rDNA and nontrimmed 454tags; no subsampling has been carried out). Classifications were done using the RDP classifier (see Methods). B. Shared and unique OTUs recovered with mitags and 454tags; dataset restricted to the V1-V3 region, with (right side) and without (left side) subsampling. Fig. S6. Phylum coverage of the pair set primers (27Fmod/533R) used for 16S rDNA amplicon tags (454tags). Fig. S7. Rank-abundance curves using different datasets from six samples retrieved from the bacterial fraction (see Table 1). A. All miTags are contrasted with 454tags considering full datasets as well as those with subsampling (2000 reads per sample) and trimmed 454tags (100-150 bp), i.e. using TARA-ALL and TARA-TRIMMED OTs with and without subsampling. B. Here, results are presented for mitags falling into the V1-V3 region, while the rest of the comparisons are the same as in A, i.e. using TARA-V1-V3 and TARA-V1-V3-TRIMMED OTs with and without subsampling. Fig. S8. Comparison between mitags, 454tags and flow cytometry (FC). A best linear fit was adjusted and Pearson's correlation coefficient was calculated for each plot. A. Quantitative comparison of relative abundances of mitags and FC Prochlorococcus counts. Pearson's r = 0.782; P < 0.001. B. miTags vs FC Synechococcus counts. Pearson's r = 0.603; P < 0.001. Fig. S9. Dendrogram based on UPGMA clustering of Bray-Curtis distances between samples analysed with mitags and 454tags. All the analysed samples belong to four size fractions (0.2-1.6, 0.8-5, 5-20 and 20-180 μm), and all were subsampled to 2000 reads per sample to correct for unequal sampling efforts (a number of samples, in particular all those analysed with m454tags, did not reach that number, and for that reason they were excluded from this analysis). The four analysed datasets are shown in panels A-D. In all datasets, singletons and OTUs present in only one sample were included. Jackknife support (subsampling = 2000) is indicated with an asterisk (50-75% support) or double asterisk (> 75% support). In each panel, the position for the 454tags and mitags is indicated with a box. The size fraction from which each sample originated is indicated with colored dots. A. miTags spanning the entire 16S rDNA. B. Same dataset as in A but including trimmed 454tags. C. miTags belonging to the V1-V3 rDNA region only. D. miTags belonging to the V1-V3 region plus trimmed 454tags. The two branches in blue indicate the only two samples that clustered together despite having been analysed with different approaches and platforms.Table S1. General description of the investigated Tara Oceans samples.Table S2. List of the Tara Oceans metagenomes sequenced by Illumina used for this study.Table S3. List of the Tara Oceans metagenomes sequenced by 454 Titanium pyrosequencing used for this study.Table S4. List of the Tara Oceans samples sequenced by 454 Titanium pyrosequencing using 16S amplicons (454tags). For comparative analysis, clustering analyses were done using two approaches: (i) using an OTU reference database (SILVA 108 release); and (ii) de novo clustering for two independent V regions (V1 and V3).Table S5. OTU comparison of the six samples sequenced with both the 454 and Illumina sequencing platforms.Table S6. Estimation of several diversity indexes for different Tara Oceans datasets.Table S7. Unique phylums, classes, orders, families and genera detected by 16S mitags that were not detected by 16S 454tags.Table S8. Unique genera detected by tags (454tags) that were not detected by mitags.Table S9. Contingency table of OTU detection versus presence of primer mismatches.Table S10. Correlation and linear fit of OTU abundances estimated by mitag, 454tag and m454tag. Scientific Research Flanders (FWO) Agence Nationale de la Recherche - No. Prometheus ANR-09-GENM-031; No. Poseidon ANR-09-BLAN-0348; No. Tara-GirusANR-09-PCS-GENM-218 Agència de Gestió d'Ajusts Universitaris i Reserca - No. CTM2010-12317-E; No. CONES 2010-0036 ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| ISSN: | 1462-2912 1462-2920 |
| DOI: | 10.1111/1462-2920.12250 |