Prokaryotic-virus-encoded auxiliary metabolic genes throughout the global oceans

Abstract

Background: Prokaryotic microbes have impacted marine biogeochemical cycles for billions of years. Viruses also impact these cycles, through lysis, horizontal gene transfer, and encoding and expressing genes that contribute to metabolic reprogramming of prokaryotic cells. While this impact is difficult to quantify in nature, we hypothesized that it can be examined by surveying virus-encoded auxiliary metabolic genes (AMGs) and assessing their ecological context.

Results: We systematically developed a global ocean AMG catalog by integrating previously described and newly identified AMGs and then placed this catalog into ecological and metabolic contexts relevant to ocean biogeochemistry. From 7.6 terabases of Tara Oceans paired prokaryote- and virus-enriched metagenomic sequence data, we increased known ocean virus populations to 579,904 (up 16%). From these virus populations, we then conservatively identified 86,913 AMGs that grouped into 22,779 sequence-based gene clusters, 7248 (~ 32%) of which were not previously reported. Using our catalog and modeled data from mock communities, we estimate that ~ 19% of ocean virus populations carry at least one AMG. To understand AMGs in their metabolic context, we identified 340 metabolic pathways encoded by ocean microbes and showed that AMGs map to 128 of them. Furthermore, we identified metabolic "hot spots" targeted by virus AMGs, including nine pathways where most steps (≥ 0.75) were AMG-targeted (involved in carbohydrate, amino acid, fatty acid, and nucleotide metabolism), as well as other pathways where virus-encoded AMGs outnumbered cellular homologs (involved in lipid A phosphates, phosphatidylethanolamine, creatine biosynthesis, phosphoribosylamine-glycine ligase, and carbamoyl-phosphate synthase pathways).

Conclusions: Together, this systematically curated, global ocean AMG catalog and analyses provide a valuable resource and foundational observations to understand the role of viruses in modulating global ocean metabolisms and their biogeochemical implications. Video Abstract.

Keywords: Tara Oceans; AMGs; Prokaryotic.

Fig. 1

The global ocean virus and auxiliary metabolic genes (AMGs) datasets. A Global map showing the location of 62 Tara Oceans sampling stations where paired prokaryote- and virus-enriched size-fractionated metagenomes were available. Numbers indicate sampling stations. B Bar charts showing the percentage frequency of virus populations (≥ 5 kb) carrying AMGs across the South Pacific Ocean (SPO), North Pacific Ocean (NPO), South Atlantic Ocean (SAO), Indian Ocean (IO), Red Sea (RS), Mediterranean Sea (MS), Southern Ocean (SO), North Atlantic Ocean (NAO), and Arctic Ocean (AO) regions, and across the surface, epipelagic (EPI) and mesopelagic (MES) layers with depths up to 150 m and 1000 m, respectively. C Pie charts showing the percentage of confidently identified viruses against the total number of contigs (≥ 5 kb) in prokaryote- and virus-enriched fractions. D Stacked bar charts showing the observed proportion of virus populations (≥ 5 kb) carrying AMGs (left) and the estimated actual proportion of virus populations carrying AMGs (right). A conversion factor of 2.1 times was estimated from in silico mock community modeling experiments that sought to extrapolate from observed AMG frequencies in fragmented genomes to the likely actual frequencies in complete genomes (see Methods).We began by developing a robust, semi-automated method to confidently identify virus contigs and virus regions within contigs (e.g., prophages in microbial genomes). While previous studies have developed benchmarked approaches for identifying viruses in ocean metagenome data [–15], additional rigor is required to ensure that cellular metabolic genes are not incorrectly labeled as AMGs [29]. We used VirSorter2 [37] to identify viruses, which leverages machine learning to identify known and unknown viruses in large-scale metagenomic datasets and has been extensively benchmarked as a top performing virus identification tool [37, 50]. Among 2,108,015 contigs (≥ 5 kb in length) from the prokaryote- and virus-enriched datasets, approximately one-third (n = 689,679) were identified as high-confidence virus contigs (see Methods and Fig. S1). These virus contigs were dereplicated by clustering into 579,904 virus populations (approximately equivalent to species-ranked taxa [15, 38, 39]) with a median genome length of 12,608 bp. This increases the number of virus populations by 16% (from 488,130 to 579,904) over those previously described [15]. While 52% (n = 303,570) of our dataset was reported in the GOV 2.0 dataset (which relied on VirSorter1) [15], 241,999 virus populations were newly identified by using VirSorter2, and another 37,335 new virus populations were identified by assessing the prokaryote-enriched fraction metagenomes. Comparison of virus populations revealed, only 1% (n = 4604) of the virus populations were shared between the prokaryote- and virus-enriched fractions. This supports existing literature, where “virus fraction” populations are different compared to those of the “cellular fractions” [51, 52]

Fig. 2

Metabolic pathways detected in global ocean microbes and viruses

Fig. 3

Complete metabolic pathways targeted by novel AMGs

Fig. 4

Metabolic pathway steps augmented by AMGs. Through enrichment analysis, the abundance of AMGs and their microbial homologs were assessed in virus versus microbial contigs across 146 AMGs with KEGG annotation. Shown here (A–E) are microbial metabolic pathways where at least one step was enriched for viruses, as determined by the abundance of a virus-encoded AMG (red text) being higher than that of their microbial homologs. Virus-enriched AMGs are starred (yellow). References (black) are enzymes involved in reaction steps per pathway as defined in the MetaCyc database. Reaction steps are indicated via circled numbers, and those lacking a number are independent of enzymatic reactions. F Summary data for virus-enriched AMGs and their levels of enrichment (the number on top of each bar) are provided as a bar chart