Article Figures & Data
Figures
- FIG 1
(A) Shannon entropy along the 18S rRNA gene alignment numbered according to corresponding positions in the 18S rRNA gene of Saccharomyces cerevisiae. (B) Average Shannon entropy per variable region compared to amplicon length with various primer combinations. Dashed lines in panel B indicate the overlap achievable with Illumina MiSeq v2 chemistry; diamonds indicate the primer sets selected for this study from the V4 (blue; alternative V4 primer sets are indicated by colored circles) and V8-V9 (red) hypervariable regions; the location of the selected set along the 18S rRNA gene is indicated by the dashed lines in panel A. The V4 region has the highest entropy within the overlap capabilities of the v2 chemistry, followed by the V8-V9 region. Although multiple primer sets are possible using the V4 region, we selected a set that has been used in previous studies. The V6 region is not highly variable and often is not included in discussion of the 18S hypervariable region.
- FIG 2
Shannon entropy per alignment position without (open symbols) and with (filled symbols) heterogeneity spacers for V4 and V8-V9 as estimated using algal sequences extracted from the Silva 119 database: blue, V4; red, V8-V9; A, B, E, and F, forward read; C, D, G, and H, reverse read. Nucleotide diversity (as represented by Shannon entropy) is close to zero for the conserved primer regions and the reverse read of the V4 primer set (C) exhibits scattered entropy across the alignment. The addition of nucleotide spacers increases total entropy across the alignment for all primers.
- FIG 3
Composition of the seven mock communities, ranging from freshwater (MC1) to marine (MC7) only. Ratios show the theoretical freshwater/marine compositions of each community based on the number of 18S rRNA copies present.
- FIG 4
Distance from the theoretical community (Jaccard dissimilarity) at various overlaps between forward and reverse reads and clustering of OTUs based on sequence similarity cutoffs for the V4 (A) and V8-V9 (B) regions using the “even” community (MC4). The left-hand y axis and gray line show the loss of reads as the minimum read overlap increases. Increasing the read overlap decreases the Jaccard dissimilarity but at the expense of loss of reads. The V4 and V8-V9 regions are able to achieve a Jaccard dissimilarity of <0.20 at 5% clustering (right-hand y axis) using 70- and 50-bp overlaps, respectively.
- FIG 5
Effect of quality processing on sequencing errors for the V4 (blue) and V8-V9 (red) regions across the seven mock communities evaluated (MC1 to MC7). All reads were processed in mothur using no base quality cutoff (A), no base quality cutoff and singletons removed (B), and a base quality cutoff at a phred score of 20 and read overlaps of 70 bp (V4) and 50 bp (V8-V9) with singletons removed (C). Removing singletons has the greatest effect on reducing sequencing error.
- FIG 6
Mean relative abundances of freshwater dominant (A and B), even (C and D), and marine dominant (E and F) communities corresponding to mock communities 2, 4, and 6 (MC2, MC4, and MC6), respectively, as represented by the V4 (A, C, and E; blue) and V8-V9 (B, D, and F; red) regions. The dashed lines indicate the mean relative abundance of the theoretical community. The V4 primer set consistently underrepresented the marine haptophyte members P. parvum and I. galbana. Although the V8-V9 primer set struggled to represent members when in low abundance, it more accurately represented the overall community structure.
- FIG 7
(A) Observed OTUs; (B) nonparametric Shannon index; (C) Jaccard distance from theoretical; (D) Bray-Curtis distance from theoretical. The V4 (blue) and V8-V9 (red) regions show similar alpha-diversity metrics in panels A and B (P = 0.82 and 0.46, respectively, Welch's t test; gray dashes indicate theoretical values), but the V8-V9 region more closely represents the theoretical community using beta-diversity metrics in panels C and D.
- FIG 8
Community composition of the V4 (top) and V8-V9 (bottom) regions at a taxonomic rank of 3 in mothur, which corresponds to the class level. Samples containing F, W, and M in their sample name belong to freshwater, wastewater, and coastal marine samples, respectively. Data used to generate this figure are available in Table S7 in the supplemental material.
- FIG 9
Observed OTUs (A) and nonparametric Shannon index (B) of environmental samples by the V4 (blue) and V8-V9 (red) regions. Wastewater samples were taken from primary effluent (PE), treatment, or secondary effluent (SE) streams from wastewater treatment plants. Both hypervariable regions show significantly higher numbers of OTUs (P < 0.003, Welch's t test) in the freshwater and coastal samples than in wastewater and similar representations of metrics across all samples.
- FIG 10
Bray-Curtis distance between the V4 and V8-V9 representations of the same sample community using a phylotype-based approach. Community representation by the V4 and V8-V9 regions varies in agreement on a sample-by-sample basis, rather than across sample types (i.e., freshwater, coastal, or wastewater).
Tables
- TABLE 1
Primers evaluated for 18S rRNA-based amplicon sequencing on the Illumina MiSeq platform
Primer identification S. cerevisiae position Target region Sequence Reference or source 550r 550 V4 GGRCMAGBCTGGTGCCAG 22 563f 563 V4 GCCAGCAVCYGCGGTAAY 22 574f 574 V4 CGGTAAYTCCAGCTCYAV 22 Reuk454FWD1 565 V4 CCAGCASCYGCGGTAATTCC 41a,b ReukREV3 981 V4 ACTTTCGTTCTTGATYRA 41a V4r 981 V4 ACTTTCGTTCTTGAT This studyb 1132r 1150 V4-V5 CCGTCAATTHCTTYAART 22 V8f 1422 V8 ATAACAGGTCTGTGATGCCCT This studyb 1422f 1422 V8 ATAACAGGTCTGTGATGC 23 1424f 1424 V8 AACAGGTCHGWRATGCCC 22 1510r 1797 V9 CCTTCYGCAGGTTCACCTAC 7b
Supplemental material
- Supplemental file 1 -
Algal species included in the construction of mock communities used in this study (Table S1); environmental samples evaluated using V4- and V8-V9-based amplicon sequencing of the 18S rRNA gene (Table S2); EukA (forward) and EukB (reverse) primers used for amplifying full-length 18S rRNA gene copies for cloning and mock community construction (Table S3); forward and reverse dual-index primers (with heterogeneity spacers) used for the V4 region (Table S4) and for the V8-V9 region (Table S5) of the 18S rRNA gene; priming locations for the 5 primers used in this study, aligned to the 18S rRNA gene from S. cerevisiae (Fig. S1); SILVA TestPrime results from in silico PCR using the original V4 primer set, modified V4 primer set, and V8-V9 primer set (Fig. S2); fluorescence intensities per cycle for the four nucleotides during the first MiSeq run (Fig. S3); proportion of reads with a phred score of greater than 30 and proportion of the four nucleotides detected per cycle during the second sequencing run (Fig. S4); evidence that the distribution of the four nucleotides per sequencing run was greatly improved with the inclusion of heterogeneity spacers for the third sequencing run (Fig. S5); mean relative abundance of mock community obtained from sequencing run 2 (V4 primers Reuk454FWD1 and ReukRev3) for ?even? community MC4 (Fig. S6); read overlap versus similarity cutoff for V4 and V8-V9 primer sets and the effect on Jaccard distance (Fig. S7); mean relative abundance of each mock community member for MC1, MC3, MC5, and MC7, determined by targeting the V4 and V8-V9 hypervariable regions of the 18S rRNA gene (Fig. S8); community composition for the V4 and V8-V9 regions at the genus level across all environmental samples (Fig. S9); PCoA plot of environmental samples using Bray-Curtis metrics for V4 and V8-V9 (Fig. S10); results for AMOVA test comparing sequencing libraries for environmental samples for the V4 and V8-V9 hypervariable regions of the 18S rRNA gene (Table S6); mean relative abundances among environmental samples for the V4 and V8-V9 regions at the class level (Table S7); principal-coordinate analyses using Bray-Curtis distance metrics indicating that amplicon data cluster according to sample type (Fig. S11).
PDF, 1.7M
- Supplemental file 1 -
