Small RNAs from plants, bacteria and fungi within the order Hypocreales are ubiquitous in human plasma

Abstract

Background: The human microbiome plays a significant role in maintaining normal physiology. Changes in its composition have been associated with bowel disease, metabolic disorders and atherosclerosis. Sequences of microbial origin have been observed within small RNA sequencing data obtained from blood samples. The aim of this study was to characterise the microbiome from which these sequences are derived.

Results: Abundant non-human small RNA sequences were identified in plasma and plasma exosomal samples. Assembly of these short sequences into longer contigs was the pivotal novel step in ascertaining their origin by BLAST searches. Most reads mapped to rRNA sequences. The taxonomic profiles of the microbes detected were very consistent between individuals but distinct from microbiomes reported at other sites. The majority of bacterial reads were from the phylum Proteobacteria, whilst for 5 of 6 individuals over 90% of the more abundant fungal reads were from the phylum Ascomycota; of these over 90% were from the order Hypocreales. Many contigs were from plants, presumably of dietary origin. In addition, extremely abundant small RNAs derived from human Y RNAs were detected.

Conclusions: A characteristic profile of a subset of the human microbiome can be obtained by sequencing small RNAs present in the blood. The source and functions of these molecules remain to be determined, but the specific profiles are likely to reflect health status. The potential to provide biomarkers of diet and for the diagnosis and prognosis of human disease is immense.

Figure 1

Schema of the strategy for analysis of sequencing data. Reads that did not align to human sequences or other known microRNAs were assembled into contigs. These were annotated by BLAST alignment to the NCBI nr database and phylogenetic analysis performed with the gi numbers of the top resulting hits.

Figure 2

Distribution of human reads by gene type and other reads by organism. Each individual is represented by a number: 1–3 this study (whole plasma); 4–6 Huang et al. [28] (plasma exosomal RNAs). The library preparation method is indicated as follows: a = Illumina; b: NEB; c = Bioo Scientific). (A) 100% stacked columns illustrating the proportions of reads annotated to human genes, non-human microRNAs or unannotated. (B) The proportions of unannotated reads (from (a)) subsequently assigned to superkingdom or kingdom. (C) Bacterial reads assigned to Phyla (those comprising <0.5% in all samples are not illustrated). (D) The proportions of fungal reads by phyla. (E) The proportions of reads assigned to classes within the phylum Ascomycetes. (F) The proportions of reads assigned to orders within the class Sordariomycetes.

Figure 3

Small RNAs derived from the non-coding hY4 RNA present in plasma. (A) The predicted secondary structure of hY4 is shown in dot-bracket notation above the sequence and the reads mapping to the 5p and 3p arms indicated below (numbers refer to the reads detected in sample 1a). The positions of the most abundant 5p and 3p reads (and much less frequent short reads) are indicated by arrows adjacent to the hY4 structure. (B) Custom Taqman small RNA assays targeting the hY4-5p or 3p RNAs corresponding to the most abundant reads amplified products several threshold cycles before individual microRNAs (eg miR-22 in sample 1a). (C) RT-PCR with primers specific for the putative hY4 fragments and performed upon RNA that had been polyadenylated, amplified products with lengths consistent with the presence of the small RNA templates detected in the sequencing rather than full length hY4 RNA. A product of the predicted size (79 bp) was detected with the hY4 5p primer, whereas a longer product of 143 bp would have been amplified from full length hY4 RNA. M: Marker, sizes in bp; Lane 1, hY4 5p; Lane 2, No RT; Lane 3, hY4 3p; Lane 4, No RT.

Figure 4

Alignment of contigs with sequences that could potentially be derived from food. Selected BLAST hits aligned using MAFFT and visualised with Jalview, coloured by BLOSUM62 score. A) Contig 2129 exhibits complete identity across the kingdom Viridiplantae 28S rRNA. Alignment with potential dietary plant and meat foodstuffs and the human rRNA gene. B) Contig 2062 is very similar to many chloroplast rRNA sequences and is shown aligned to several of the best hits and potential dietary sources. 1 - Pseudendoclonium akinetum: 2 - Trichosarcina mucosa: 3 - Lycopodium clavatum: 4 - Zygnema: 5 - Solanum Lycopersicum: 6 - Solanum tuberosum: C) All the lineages to which Contig 1748 has a perfect match, including many potential food sources. Representative sequences from each species are aligned and coloured by percentage identity (1: Fragaria vesca, 2: Medicago truncatula, 3: Lotus japonicus, 4: Glycine max, 5: Arabidopsis thaliana, 6: Solanum lycopersicum, NB: A 30 bp insertion present in Glycine max immediately 5 prime of the contig 1748 sequence is omitted to facilitate visualisation). Full lineage of core eudicotyledons is [root; cellular organisms; Eukaryota; Viridiplantae; Streptophyta; Streptophytina; Embryophyta; Tracheophyta; Euphyllophyta; Spermatophyta; Magnoliophyta; eudicotyledons].

Figure 5

Order-level phylogenetic profile of fungal small RNAs. The tree illustrates the taxonomic composition of the contigs derived from small RNAs isolated from the plasma samples of six individuals. All orders within the kingdom Fungi which have matching sequences are illustrated. The numbers of contigs assigned to each taxonomic group are indicated within the tree. The numbers on the right are the total number of reads assigned to each order; the order Hypocreales, highlighted in green, is the most abundant.

Figure 6

Taxonomic profile and relative expression between individuals of abundant contigs. (A) The top 20 contigs ranked according to the total number of reads aligned to them from all samples. All the contigs matched rRNA and the top BLAST hit is shown. The lowest common taxonomic rank was assigned by analysis of the BLAST hits with scores within 5% of the top hit. The proportion of reads mapping to each contig in individuals and overall is indicated. (B) Phylogenetic tree of the top 20 contigs generated with MEGAN. The number of contigs assigned at each node is indicated.

Figure 7

Alignment of contig 44 to rRNA sequences. (A) BLAST alignment of contig 44 with Cordyceps gunnii 28S ribosomal RNA gene. (B) Section of multiple alignment between contig 44 and rRNA sequences from exemplar species in the orders Hypocreales or Malasseziales and human rRNA. (C) Phylogram illustrating the divergence between Hypocreales/contig 44, Malasseziales and human rRNA sequences.

Figure 8

Distribution of reads along Hypocreales rRNA gene. The positions of the most abundant contigs along the rDNA are indicated at the top of the figure. The read coverage for contig 44 is shown. Abbreviations: SSU: Small subunit; LSU: Large subunit; ITS: Internal Transcribed Sequence.