Hundreds of viral families in the healthy infant gut

Abstract

The gut microbiome (GM) is shaped through infancy and plays a major role in determining susceptibility to chronic diseases later in life. Bacteriophages (phage) are known to modulate bacterial populations in numerous ecosystems, including the gut. However, virome data is difficult to analyse because it mostly consists of unknown viruses, i.e. viral dark matter. Here, we manually resolved the viral dark matter in the largest human virome study published to date. Fecal viromes from a cohort of 647 infants at 1 year of age were deeply sequenced and analysed through successive rounds of clustering and curation. This uncovered more than ten thousand viral species distributed over 248 viral families falling within 17 viral order-level clusters. Most of the defined viral families and orders were novel and belonged to the Caudoviricetes viral class. Bacterial hosts were predicted for 79% of the viral species using CRISPR spacers in metagenomes from the same fecal samples. While Bacteroides-infecting Crassphages were present, novel viral families were more predominant, including phages infecting Clostridiales and Bifidobacterium. Phage lifestyles were determined for more than three thousand caudoviral species. Lifestyles were homogeneous at the family level for 149 caudiviral families. 32 families were found to be virulent, while 117 families were temperate. Virulent phage families were more abundant but temperate phage families were more diverse and widespread. Together, the viral families found in this study represent a major expansion of current bacteriophage taxonomy, and the sequences have been put online for use and validation by the community.

UG/Abi: a highly diverse family of prokaryotic reverse transcriptases associated with defense functions

“…In an attempt to assign taxonomy to these viral sequences or test whether they belong to a known family, we compared them against reference databases and a compendium of new phage families ( 54–56 ) using ANI (average nucleotide identity) and gene-content network information (Methods). At the ANI level, some UG27-containing viral genomes share similarities with recently proposed orders of bacterial viruses, including Crassvirales and Friedlandervirales ( 55 ) and viruses from the proposed Quimbyviridae family ( 54 ) ( Supplementary Table S5 ). Most of the UG27-containing genomes however showed very distant to zero similarity to any currently classified bacteriophage family even at the protein level, so we constructed an aggregate protein similarity (APS) tree ( 55 ) to determine whether the phages belonged to known or novel viral families.…”

“…At the ANI level, some UG27-containing viral genomes share similarities with recently proposed orders of bacterial viruses, including Crassvirales and Friedlandervirales ( 55 ) and viruses from the proposed Quimbyviridae family ( 54 ) ( Supplementary Table S5 ). Most of the UG27-containing genomes however showed very distant to zero similarity to any currently classified bacteriophage family even at the protein level, so we constructed an aggregate protein similarity (APS) tree ( 55 ) to determine whether the phages belonged to known or novel viral families. After cutting the APS tree at the viral family level, by using the six proposed Crassvirales families ( 56 ) as references, some of the UG27 containing phages co-clustered with Quimbyviridae ( 54 ) and various families within the proposed Crassvirales , Twortvirales and Freidlandervirales viral orders ( 55 ) ( Supplementary Table S6 ).…”

“…Most of the UG27-containing genomes however showed very distant to zero similarity to any currently classified bacteriophage family even at the protein level, so we constructed an aggregate protein similarity (APS) tree ( 55 ) to determine whether the phages belonged to known or novel viral families. After cutting the APS tree at the viral family level, by using the six proposed Crassvirales families ( 56 ) as references, some of the UG27 containing phages co-clustered with Quimbyviridae ( 54 ) and various families within the proposed Crassvirales , Twortvirales and Freidlandervirales viral orders ( 55 ) ( Supplementary Table S6 ). Four main clusters (205, 206, 255 and 265) accounted for 1350 out of 1447 UG27-containing genomes.…”

“…After this, vContact2 ( 59 ) was used with the –db ‘None’ option and default parameter. To increase the resolution at the gene sharing level, an aggregate protein similarity tree was built as described earlier ( 55 ). To obtain family-level viral clades, the tree was rooted and cut at the levels reproducing the six proposed Crassvirales families ( 56 ).…”

Extensive gut virome variation and its associations with host and environmental factors in a population-level cohort

“…Owing to the lack of a high-quality genome catalogue of human gut phages until recently, the major clades of phages in the human gut are still largely unknown, other than a few groups, such as crAss-like phages 5,7,8,10,38,39 . To explore the major and abundant phage clades in the human gut, we quantified the abundance of each vOTU/VC by mapping the metagenomic reads of the 4,198 individuals to the catalogue ( Methods ).…”