Glacier ice archives nearly 15,000-year-old microbes and phages

Abstract

Background: Glacier ice archives information, including microbiology, that helps reveal paleoclimate histories and predict future climate change. Though glacier-ice microbes are studied using culture or amplicon approaches, more challenging metagenomic approaches, which provide access to functional, genome-resolved information and viruses, are under-utilized, partly due to low biomass and potential contamination.

Results: We expand existing clean sampling procedures using controlled artificial ice-core experiments and adapted previously established low-biomass metagenomic approaches to study glacier-ice viruses. Controlled sampling experiments drastically reduced mock contaminants including bacteria, viruses, and free DNA to background levels. Amplicon sequencing from eight depths of two Tibetan Plateau ice cores revealed common glacier-ice lineages including Janthinobacterium, Polaromonas, Herminiimonas, Flavobacterium, Sphingomonas, and Methylobacterium as the dominant genera, while microbial communities were significantly different between two ice cores, associating with different climate conditions during deposition. Separately, ~355- and ~14,400-year-old ice were subject to viral enrichment and low-input quantitative sequencing, yielding genomic sequences for 33 vOTUs. These were virtually all unique to this study, representing 28 novel genera and not a single species shared with 225 environmentally diverse viromes. Further, 42.4% of the vOTUs were identifiable temperate, which is significantly higher than that in gut, soil, and marine viromes, and indicates that temperate phages are possibly favored in glacier-ice environments before being frozen. In silico host predictions linked 18 vOTUs to co-occurring abundant bacteria (Methylobacterium, Sphingomonas, and Janthinobacterium), indicating that these phages infected ice-abundant bacterial groups before being archived. Functional genome annotation revealed four virus-encoded auxiliary metabolic genes, particularly two motility genes suggest viruses potentially facilitate nutrient acquisition for their hosts. Finally, given their possible importance to methane cycling in ice, we focused on Methylobacterium viruses by contextualizing our ice-observed viruses against 123 viromes and prophages extracted from 131 Methylobacterium genomes, revealing that the archived viruses might originate from soil or plants.

Conclusions: Together, these efforts further microbial and viral sampling procedures for glacier ice and provide a first window into viral communities and functions in ancient glacier environments. Such methods and datasets can potentially enable researchers to contextualize new discoveries and begin to incorporate glacier-ice microbes and their viruses relative to past and present climate change in geographically diverse regions globally. Video Abstract.

Keywords: Guliya ice cap; Ice microbes; Ice viruses; Janthinobacterium; Methylobacterium; Mountain glacier ice; Sphingomonas; Surface decontamination.

Fig. 1

Establishment of decontamination protocol. a Schematic of layered removal of the outer core surface to obtain clean inner ice (top panel) and experimental approach to establish decontamination procedures using sterile artificial ice core sections coated with mock “contaminants” (down panel). Cut, wash, and inner represent ice samples collected from band saw scrapping, water washing, and the inner ice, respectively. Mix represents a sample from the melted ice of a control ice core section prepared without decontamination processing. The mock contaminants were detected by qPCR and nested PCR (see “Materials and methods”) in (b) and (c). b Total bacterial (dark teal color) and viral (purple color) numbers were quantified by qPCR using strain-designed primers in all samples collected in (a). c Lambda DNA was detected using nested PCR with designed outer and inner primer sets for lambda DNA. PCR products from inner primer sets were visualized by agarose gel electrophoresis; 1, 100bp DNA ladder; 2–7 represent 1.9×10⁴, 10³, 10², 10¹, 10⁰, and 10⁻¹ (10-times dilution from standards) copies of lambda DNA, respectively, used as templates for nested PCR; 8, Control_Negative (no template); 9, Sample Cut1; 10, Wash1;11, Inner1; 12, Cut2; 13, Wash2; 14, Inner2; 15, 100bp DNA ladder (same as 1); 16, Control_Mix; 17, Control_Negative (same as 8)

Fig. 2

Sampling sites of glacier ice and an overview of experimental design. a Location of the Guliya ice cap; b drilling sites of the S3 and PS ice cores in Guliya ice cap; c sampling depths of eight ice samples used to investigate the microbial and viral communities; and d an overview of experimental design for microbial and viral investigations of collected ice samples. S3 and PS cores were drilled from the summit and plateau of Guliya ice cap, respectively (b). The drill date and length of the two ice cores and the approximate age of each sample are indicated (c). The sample names are coded by depth, e.g., for D13.3 is from 13.3 m below the glacier surface. All samples were subjected to microbial investigations, and two samples D25 and D49 (light blue) were selected for viral investigation

Fig. 3

Distinct microbial profiles between PS and S3 ice cores. a Microbial profiles of the 26 most abundant genera in PS and S3 ice core samples. Profiles are illustrated as a percent of the total 16S rRNA gene amplicon sequences. The key indicates genera, preceded by family, or order in cases where family is not assigned. Genera labeled “Other” represent sequences with unknown genus-level taxonomy, i.e., distinct from taxonomically assigned genera in the reference database. The 26 most abundant genera, defined as those comprising at least 1.0% of the sequences in at least one ice sample, collectively represented >95.1% of each community. The total relative abundance of these genera was normalized to 100%. b PCoA showing sample clustering based on microbial communities at OTU (~species, 97% identity) level. Samples from the same ice core are marked with the same color. Sample names are indicated next to each symbol. PCoA was performed on the weighted UniFrac metric, which accounts for the relative abundance and inferred relatedness of the lineages present

Fig. 4

Taxonomies (a), communities (b), and host linkages (c–f) of 33 vOTUs recovered from two glacier ice samples. a Viral taxonomy was assigned by comparing genome-content-based network analysis of the 33 glacier vOTUs and 2304 known viral genomes in the NCBI RefSeq database using vConTACT v2 (see “Materials and methods”). vOTUs were classified into three groups: “Singletons” (gray) that had no close relatives; “Exclusive VCs” (black) that were viral clusters (VCs) of exclusively glacier ice vOTUs; and “Classified VCs” (blue) which included glacier ice vOTUs and Refseq viral genomes. b The normalized coverage of these 33 vOTUs was generated by mapping quality-controlled reads to vOTUs, and was normalized to per gigabase of metagenome. c–f Relative abundances of three abundant (>1.0%) microbial genera and their viruses: c Methylobacterium in D25, d Methylobacterium in D49, e Janthinobacterium in D25, and f Sphingomonas in D49. Relative abundances of microbes are based on 16S rRNA amplicon sequencing, and vOTUs are based on their coverages generated by mapping quality-controlled reads to vOTUs. Viruses were linked to hosts in silico by three methods: Blastn, VirHostMatcher, and CRISPR matches (see “Materials and methods”)

Fig. 5

Genome content–based network (a) and genome organization (b–c) of viruses infecting Methylobacterium. a A gene content–based network was built to evaluate the relationship of five glacier-ice viruses to 484 viruses from other environments, all predicted to infect Methylobacterium (see “Materials and methods”). For clarity, viruses that were not connected to any of the five glacier-ice viruses were excluded from the network. Each node represents a virus, with glacier-ice viruses and others shaped in triangle and circle, respectively. The edge between nodes indicates the distances between two viruses. Viral clusters (VCs) are generated by vConTACT v2, and viruses that belonged to the same VC are indicated in the same color. In each VC, the name and source environment of each member are indicated, with glacier-ice virus at the top. All gray nodes represent viruses from other environments that did not share VC with any glacier-ice virus. b–c Genomic organization and comparison of Methylobacterium viruses that are longer than 15kb in VC0_0 and VC8_0 from (a). Only glacier viruses and their closely related viruses with genome size more than 15kb were illustrated, including four and four viruses from VC0_0 (b) and VC8_0 (c), respectively. Viral contigs were compared in terms of gene similarity, order, and direction (i.e., leftward or rightward arrow). Genes are coded in color based on their putative biological function. Potential microbial genes were identified by CheckV (see “Materials and methods”) and marked in green color. The predicted protein with no functional annotation is classified as “Hypothetical protein” and colored in gray. The gray lines indicate the amino acid identities between genes, as illustrated in the scale bar. Abbreviations: TransR, transcriptional regulator; MTase, mRNA methyltransferase; tRNASL, tRNA-splicing ligase; terS, terminase small subunit; terL, terminase large subunit; Mu N, Mu N gene product; DNARP, DNA repair protein; DNAM, DNA methylase; DNAP, DNA polymerase; RNAP, RNA polymerase; LytT, lytic transglycosylases; TransmP, transmembrane protein; DigC, diguanylate cyclase