The genetic potential for key biogeochemical processes in Arctic frost flowers and young sea ice revealed by metagenomic analysis (original) (raw)
Abstract
Newly formed sea ice is a vast and biogeochemically active environment. Recently, we reported an unusual microbial community dominated by members of the Rhizobiales in frost flowers at the surface of Arctic young sea ice based on the presence of 16S gene sequences related to these strains. Here, we use metagenomic analysis of two samples, from a field of frost flowers and the underlying young sea ice, to explore the metabolic potential of this surface ice community. The analysis links genes for key biogeochemical processes to the Rhizobiales, including dimethylsulfide uptake, betaine glycine turnover, and halocarbon production. Nodulation and nitrogen fixation genes characteristic of terrestrial root-nodulating Rhizobiales were generally lacking from these metagenomes. Non-Rhizobiales clades at the ice surface had genes that would enable additional biogeochemical processes, including mercury reduction and dimethylsulfoniopropionate catabolism. Although the ultimate source of the observed microbial community is not known, considerations of the possible role of eolian deposition or transport with particles entrained during ice formation favor a suspended particle source for this microbial community.
Introduction
Frost flowers are hypersaline ice structures common to the surface of newly formed (young) sea ice when atmospheric temperatures are sufficiently cold (Style & Worster, ). Frost flowers, having higher salinity, greater surface area (per unit mass), and lower temperature than the underlying young sea ice, have garnered significant attention for their potential influence on atmospheric chemistry (Rankin et al., ; Alvarez-Aviles et al., ). This influence is a product of their ability to concentrate salt and organic matter at the critical ice–atmosphere interface (Perovich & Richter-Menge, ; Bowman & Deming, ; Aslam et al., ; Beine et al., ; Douglas et al., ).
Frost flowers also concentrate bacteria, with the more saline frost flowers typically containing the largest number of bacteria, suggesting that they are populated from sea ice brines rejected to the surface of young sea ice during ice growth (Bowman & Deming, ). In an earlier study, we used 16S rRNA gene sequences to investigate the microbial community composition and structure of frost flowers and underlying young sea ice growing on a small lead offshore of Barrow, Alaska (Bowman et al., ), part of the larger Barrow flaw-lead. The Barrow flaw-lead is itself an extension of the Arctic-wide circumpolar flaw-lead system. Surprisingly, the microbial community was dominated by members of the order Rhizobiales. While the Rhizobiales are not unknown in the marine environment or in other extreme environments, they had not been previously reported in Arctic waters or in association with sea ice.
To further develop hypotheses regarding the source of the observed frost flower Rhizobiales and to explore the potential biogeochemical impact of this community at the critical ice–ocean–atmosphere interface, we sequenced metagenomes from two of the samples from the earlier study, one from a small field of frost flowers and the other from the underlying young sea ice. A quantitative approach was developed to assess the taxonomic composition of these communities and the presence of key functional genes. We also considered available data on wind magnitude, sea ice sediment concentration, and bacterial abundance to evaluate whether aerial deposition after ice formation or entrainment of suspended particles during ice formation was the more likely delivery mechanism of the frost flower microbial community.
Materials and methods
Sample collection and sequencing
Samples were collected in April, 2010, from a small lead offshore of Pt. Barrow, Alaska, USA, and the DNA extracted as described previously (Bowman et al., ). One young ice and one frost flower sample corresponding to samples YI4 and FF1 of Bowman et al. () were selected for shotgun sequencing on the Illumina platform following whole-genome amplification (WGA). For WGA, sheared DNA (200–500 bp) was amplified with the SEQPLEX kit (Sigma-Aldrich) following the manufacturer's recommended protocol, including the enzymatic primer removal step. WGA was carried out under conditions of high sterility to reduce the risk of contamination. Products from WGA were visualized on a gel and purified using the Gene-Jet PCR Purification Kit (Fermentas). Paired-end sequencing at 2 × 125 bp with a 175-bp insert was conducted at the Argonne National Laboratory on the Illumina Hi-Seq platform using one lane for each metagenome.
Sequence processing and taxonomic analysis
Raw sequence reads were screened for duplicated reads using the fastq_nodup command in seastar (Iverson et al., ). The command was called with the flags -z -l 15 -d 1 -e 2 -v. Remaining primer sequence was removed using an in-house script (primer_purge.py). De-duplicated reads without primers were trimmed for quality using the seastar command trimfastq with the flags -z –mates_file -p 0.5 -l 34 -m 34 –add_len. The resulting trimmed 16S rRNA gene reads were classified using the Bayesian classifier in Mothur (Schloss et al., ) with a kmer size of 8 against the greengenes reference taxonomy (www.mothur.org/wiki/Greengenes-formatted_databases). Reads that classified at the level of phylum with a bootstrap score > 60 were considered to be 16S rRNA genes, and the taxonomic classification was retained.
All quality-controlled (QC'd) reads were assembled iteratively with the Velvet assembler (Zerbino, ) following the method of Iverson et al. (). Kmer lengths between 15 and 29 were evaluated, with the optimal assembly using a kmer length of 23. Following the initial round of assembly, QC'd reads were mapped to the assembled contigs using the BWA aligner (Li & Durbin, ) with the flags -n 0.001 -l 18. The next round of assembly used only the contigs assembled in the previous round and successfully mapped reads. After five iterations, the N50 value and maximum contig length stabilized at 1905 and 18 102, respectively, for the frost flower sample and 329 and 4992 for the young sea ice sample. Open reading frames (ORFs) longer than 300 bp were identified in frost flower contigs longer than 10 000 bp using an in-house script (annotate_contigs.py). ORFs were evaluated as potential coding sequences (CDS) by searching translations against the PFAM-A database (Punta et al., ) with the hmmscan command in hmmer v3.0 (Eddy, ) and delta blast (Boratyn et al., ) against the NCBI cdd database. Contigs were searched for 16S rRNA genes using blastn (Altschul et al., ) against the NCBI 16S Microbial database.
Metabolic profiles
We created five databases composed of high-quality, full-length or near-full-length protein sequences from UniProt (Bairoch et al., ) derived from 20 different genes or gene families diagnostic of 10 different metabolic processes (Table 1). All QC'd reads were aligned against these databases using blastx (Altschul et al., ). Reads that successfully mapped to a sequence in the database were extracted as amino acid translations of the alignment. For the frost flower metagenome, extracted read translations were further classified using the phylogenetic placement program pplacer (Matsen et al., ) and a reference tree of the database sequences. Alignment of the reference sequences, and the query sequences to the reference alignment, was carried out with three iterations in clustalo v1.2 (Sievers et al., ). Reference trees were constructed using fasttree2 (Price et al., ) or raxml v7.2.8 (Stamatakis, ). The relative proportion of reference genes in the two metagenomes was estimated by dividing the combined length of the recruited read translations by the mean length of the reference sequences. Because the 16S rRNA gene often has multiple copies in bacterial genomes, we normalized coverage to proteins from the single-copy recA gene. The significance of differences in normalized coverage between the two metagenomes was assessed using the two-proportion z-test, with a sample size of 1000. Thus, a hypothetical RecA-normalized coverage value of 0.019 would represent 19 observations of 1000 in the z-test.
Table 1
Number of reads, coverage, and normalized coverage by protein groups (left column). Protein groups with statistically significant differences in abundance are shown in bold. Negative differences indicate protein groups with higher normalized abundance in the young sea ice dataset. Protein group function shown below
Mean length | Frost flower | Young sea ice | Comparison | |||||||
---|---|---|---|---|---|---|---|---|---|---|
Translated reads | Coverage | Normalized coverage | Translated reads | Coverage | Normalized coverage | Δ | z | P | ||
Betaine methyltransferasea | 337 | 1450 | 137 | 0.187 | 436 | 41 | 0.198 | −0.011 | −0.624 | 0.5326 |
CmuAb | 356 | 69 | 6 | 0.008 | 25 | 2 | 0.01 | −0.002 | −0.474 | 0.6355 |
DMSP lyase/demethylasec | 260 | 1425 | 175 | 0.239 | 451 | 55 | 0.266 | −0.027 | −1.39 | 0.1645 |
DMS catabolismd | 465 | 2187 | 150 | 0.205 | 354 | 24 | 0.116 | 0.089 | 5.422 | < 0.0002 |
Haloperoxidasee | 240 | 889 | 118 | 0.161 | 105 | 14 | 0.068 | 0.093 | 6.531 | < 0.0002 |
Ice structuringf | 337 | 46 | 4 | 0.005 | 105 | 9 | 0.043 | −0.038 | −5.552 | < 0.0002 |
Indoleacetomide hydrolaseg | 457 | 626 | 43 | 0.059 | 198 | 13 | 0.063 | −0.004 | −0.374 | 0.7084 |
MerAh | 454 | 1417 | 99 | 0.135 | 369 | 26 | 0.126 | 0.009 | 0.597 | 0.5505 |
NifHi | 279 | 51 | 5 | 0.007 | 30 | 3 | 0.014 | −0.007 | −1.536 | 0.1245 |
NodAj | 160 | 62 | 12 | 0.016 | 2 | 0 | 0 | 0.016 | na | na |
NodBk | 136 | 638 | 150 | 0.205 | 57 | 13 | 0.063 | 0.142 | 9.321 | < 0.0002 |
NodCl | 245 | 743 | 97 | 0.133 | 211 | 27 | 0.13 | 0.003 | 0.133 | 0.8942 |
NodDm | 249 | 118 | 15 | 0.021 | 21 | 2 | 0.01 | 0.011 | 1.991 | 0.0465 |
NodEn | 312 | 719 | 73 | 0.1 | 277 | 28 | 0.135 | −0.035 | −2.43 | 0.0151 |
NodFo | 90 | 12 | 4 | 0.005 | 5 | 1 | 0.005 | 0 | 0 | 1 |
NodGp | 194 | 1905 | 314 | 0.43 | 422 | 69 | 0.333 | 0.097 | 4.465 | < 0.0002 |
NodHq | 242 | 13 | 1 | 0.001 | 5 | 0 | 0 | 0.001 | na | na |
NodIr | 309 | 15836 | 1639 | 2.242 | 2609 | 270 | 1.304 | 0.938 | na | na |
NodJs | 268 | 702 | 83 | 0.114 | 150 | 17 | 0.082 | 0.032 | 2.407 | 0.0161 |
RecAt | 276 | 6306 | 731 | 1 | 1789 | 207 | 1 | 0 | na | na |
Mean length | Frost flower | Young sea ice | Comparison | |||||||
---|---|---|---|---|---|---|---|---|---|---|
Translated reads | Coverage | Normalized coverage | Translated reads | Coverage | Normalized coverage | Δ | z | P | ||
Betaine methyltransferasea | 337 | 1450 | 137 | 0.187 | 436 | 41 | 0.198 | −0.011 | −0.624 | 0.5326 |
CmuAb | 356 | 69 | 6 | 0.008 | 25 | 2 | 0.01 | −0.002 | −0.474 | 0.6355 |
DMSP lyase/demethylasec | 260 | 1425 | 175 | 0.239 | 451 | 55 | 0.266 | −0.027 | −1.39 | 0.1645 |
DMS catabolismd | 465 | 2187 | 150 | 0.205 | 354 | 24 | 0.116 | 0.089 | 5.422 | < 0.0002 |
Haloperoxidasee | 240 | 889 | 118 | 0.161 | 105 | 14 | 0.068 | 0.093 | 6.531 | < 0.0002 |
Ice structuringf | 337 | 46 | 4 | 0.005 | 105 | 9 | 0.043 | −0.038 | −5.552 | < 0.0002 |
Indoleacetomide hydrolaseg | 457 | 626 | 43 | 0.059 | 198 | 13 | 0.063 | −0.004 | −0.374 | 0.7084 |
MerAh | 454 | 1417 | 99 | 0.135 | 369 | 26 | 0.126 | 0.009 | 0.597 | 0.5505 |
NifHi | 279 | 51 | 5 | 0.007 | 30 | 3 | 0.014 | −0.007 | −1.536 | 0.1245 |
NodAj | 160 | 62 | 12 | 0.016 | 2 | 0 | 0 | 0.016 | na | na |
NodBk | 136 | 638 | 150 | 0.205 | 57 | 13 | 0.063 | 0.142 | 9.321 | < 0.0002 |
NodCl | 245 | 743 | 97 | 0.133 | 211 | 27 | 0.13 | 0.003 | 0.133 | 0.8942 |
NodDm | 249 | 118 | 15 | 0.021 | 21 | 2 | 0.01 | 0.011 | 1.991 | 0.0465 |
NodEn | 312 | 719 | 73 | 0.1 | 277 | 28 | 0.135 | −0.035 | −2.43 | 0.0151 |
NodFo | 90 | 12 | 4 | 0.005 | 5 | 1 | 0.005 | 0 | 0 | 1 |
NodGp | 194 | 1905 | 314 | 0.43 | 422 | 69 | 0.333 | 0.097 | 4.465 | < 0.0002 |
NodHq | 242 | 13 | 1 | 0.001 | 5 | 0 | 0 | 0.001 | na | na |
NodIr | 309 | 15836 | 1639 | 2.242 | 2609 | 270 | 1.304 | 0.938 | na | na |
NodJs | 268 | 702 | 83 | 0.114 | 150 | 17 | 0.082 | 0.032 | 2.407 | 0.0161 |
RecAt | 276 | 6306 | 731 | 1 | 1789 | 207 | 1 | 0 | na | na |
aGlycine betaine catabolism, bmethylhalide uptake, cdimethylsulfoniopropionate catabolism, ddimethylsulfide uptake, emethylhalide production, fice binding/nucleating, gindole-3-acetate production, hmercury reduction, initrogen fixation, j–snodulation factor production, tDNA repair.
Table 1
Number of reads, coverage, and normalized coverage by protein groups (left column). Protein groups with statistically significant differences in abundance are shown in bold. Negative differences indicate protein groups with higher normalized abundance in the young sea ice dataset. Protein group function shown below
Mean length | Frost flower | Young sea ice | Comparison | |||||||
---|---|---|---|---|---|---|---|---|---|---|
Translated reads | Coverage | Normalized coverage | Translated reads | Coverage | Normalized coverage | Δ | z | P | ||
Betaine methyltransferasea | 337 | 1450 | 137 | 0.187 | 436 | 41 | 0.198 | −0.011 | −0.624 | 0.5326 |
CmuAb | 356 | 69 | 6 | 0.008 | 25 | 2 | 0.01 | −0.002 | −0.474 | 0.6355 |
DMSP lyase/demethylasec | 260 | 1425 | 175 | 0.239 | 451 | 55 | 0.266 | −0.027 | −1.39 | 0.1645 |
DMS catabolismd | 465 | 2187 | 150 | 0.205 | 354 | 24 | 0.116 | 0.089 | 5.422 | < 0.0002 |
Haloperoxidasee | 240 | 889 | 118 | 0.161 | 105 | 14 | 0.068 | 0.093 | 6.531 | < 0.0002 |
Ice structuringf | 337 | 46 | 4 | 0.005 | 105 | 9 | 0.043 | −0.038 | −5.552 | < 0.0002 |
Indoleacetomide hydrolaseg | 457 | 626 | 43 | 0.059 | 198 | 13 | 0.063 | −0.004 | −0.374 | 0.7084 |
MerAh | 454 | 1417 | 99 | 0.135 | 369 | 26 | 0.126 | 0.009 | 0.597 | 0.5505 |
NifHi | 279 | 51 | 5 | 0.007 | 30 | 3 | 0.014 | −0.007 | −1.536 | 0.1245 |
NodAj | 160 | 62 | 12 | 0.016 | 2 | 0 | 0 | 0.016 | na | na |
NodBk | 136 | 638 | 150 | 0.205 | 57 | 13 | 0.063 | 0.142 | 9.321 | < 0.0002 |
NodCl | 245 | 743 | 97 | 0.133 | 211 | 27 | 0.13 | 0.003 | 0.133 | 0.8942 |
NodDm | 249 | 118 | 15 | 0.021 | 21 | 2 | 0.01 | 0.011 | 1.991 | 0.0465 |
NodEn | 312 | 719 | 73 | 0.1 | 277 | 28 | 0.135 | −0.035 | −2.43 | 0.0151 |
NodFo | 90 | 12 | 4 | 0.005 | 5 | 1 | 0.005 | 0 | 0 | 1 |
NodGp | 194 | 1905 | 314 | 0.43 | 422 | 69 | 0.333 | 0.097 | 4.465 | < 0.0002 |
NodHq | 242 | 13 | 1 | 0.001 | 5 | 0 | 0 | 0.001 | na | na |
NodIr | 309 | 15836 | 1639 | 2.242 | 2609 | 270 | 1.304 | 0.938 | na | na |
NodJs | 268 | 702 | 83 | 0.114 | 150 | 17 | 0.082 | 0.032 | 2.407 | 0.0161 |
RecAt | 276 | 6306 | 731 | 1 | 1789 | 207 | 1 | 0 | na | na |
Mean length | Frost flower | Young sea ice | Comparison | |||||||
---|---|---|---|---|---|---|---|---|---|---|
Translated reads | Coverage | Normalized coverage | Translated reads | Coverage | Normalized coverage | Δ | z | P | ||
Betaine methyltransferasea | 337 | 1450 | 137 | 0.187 | 436 | 41 | 0.198 | −0.011 | −0.624 | 0.5326 |
CmuAb | 356 | 69 | 6 | 0.008 | 25 | 2 | 0.01 | −0.002 | −0.474 | 0.6355 |
DMSP lyase/demethylasec | 260 | 1425 | 175 | 0.239 | 451 | 55 | 0.266 | −0.027 | −1.39 | 0.1645 |
DMS catabolismd | 465 | 2187 | 150 | 0.205 | 354 | 24 | 0.116 | 0.089 | 5.422 | < 0.0002 |
Haloperoxidasee | 240 | 889 | 118 | 0.161 | 105 | 14 | 0.068 | 0.093 | 6.531 | < 0.0002 |
Ice structuringf | 337 | 46 | 4 | 0.005 | 105 | 9 | 0.043 | −0.038 | −5.552 | < 0.0002 |
Indoleacetomide hydrolaseg | 457 | 626 | 43 | 0.059 | 198 | 13 | 0.063 | −0.004 | −0.374 | 0.7084 |
MerAh | 454 | 1417 | 99 | 0.135 | 369 | 26 | 0.126 | 0.009 | 0.597 | 0.5505 |
NifHi | 279 | 51 | 5 | 0.007 | 30 | 3 | 0.014 | −0.007 | −1.536 | 0.1245 |
NodAj | 160 | 62 | 12 | 0.016 | 2 | 0 | 0 | 0.016 | na | na |
NodBk | 136 | 638 | 150 | 0.205 | 57 | 13 | 0.063 | 0.142 | 9.321 | < 0.0002 |
NodCl | 245 | 743 | 97 | 0.133 | 211 | 27 | 0.13 | 0.003 | 0.133 | 0.8942 |
NodDm | 249 | 118 | 15 | 0.021 | 21 | 2 | 0.01 | 0.011 | 1.991 | 0.0465 |
NodEn | 312 | 719 | 73 | 0.1 | 277 | 28 | 0.135 | −0.035 | −2.43 | 0.0151 |
NodFo | 90 | 12 | 4 | 0.005 | 5 | 1 | 0.005 | 0 | 0 | 1 |
NodGp | 194 | 1905 | 314 | 0.43 | 422 | 69 | 0.333 | 0.097 | 4.465 | < 0.0002 |
NodHq | 242 | 13 | 1 | 0.001 | 5 | 0 | 0 | 0.001 | na | na |
NodIr | 309 | 15836 | 1639 | 2.242 | 2609 | 270 | 1.304 | 0.938 | na | na |
NodJs | 268 | 702 | 83 | 0.114 | 150 | 17 | 0.082 | 0.032 | 2.407 | 0.0161 |
RecAt | 276 | 6306 | 731 | 1 | 1789 | 207 | 1 | 0 | na | na |
aGlycine betaine catabolism, bmethylhalide uptake, cdimethylsulfoniopropionate catabolism, ddimethylsulfide uptake, emethylhalide production, fice binding/nucleating, gindole-3-acetate production, hmercury reduction, initrogen fixation, j–snodulation factor production, tDNA repair.
To further investigate both the metabolic potential of the frost flower and the young sea ice microbial communities and the taxonomic affiliation of the detected strains, we used the BWA aligner to align all QC'd reads against the 4710 prokaryotic genomes (chromosomes and plasmids) available from NCBI as complete assemblies as of October, 2012. The BWA aln command was called with the flags -n 0.001 -l 12 -k 2 following a call to the BWA index command using -a is. These alignments were conducted independently; that is, each index created for the alignment contained only a single genome. Genome coverage and breadth, and the fraction of positions in the reference sequence with a mapped read, were calculated using in-house scripts (tally_reads_in_sam.py, read_length_recruit_from_sam.py, and get_coverage_gaps.r). To assess functional differences between the genomes present in the frost flower metagenome and sequenced Rhizobiales, we analyzed regions of the top-five recruiting Rhizobiales reference genomes (according to coverage × breadth), and the genome of Candidatus Pelagibacter ubique, longer than 5000 nt that did not recruit any reads using an in-house script (get_coverage_gaps.py). Translated ORFs from these gap regions were evaluated using blastp against the NCBI refseq_protein database with an E-value cutoff of 10−30.
The metagenomic sequences were submitted to the MG-RAST database (Meyer et al., ) with MG-RAST ID numbers 4537103.3–4537105.3. Scripts used in this analysis are available at https://github.com/bowmanjeffs/barrow_metagenome_scripts.
Results
Sequencing produced 21 012 180 read pairs for the frost flower sample and 9 882 191 read pairs for the young sea ice sample. Approximately 30% of the reads passed QC for both libraries, with an average length of 96 nt, producing 1.1 Gb for the frost flower library and 560 Mb from the young sea ice library. About 0.0012% of the reads classified as 16S rRNA gene. A majority of the 16S sequences (52%, 7247) from the frost flower library classified as Rhizobiales, whereas about 8% (541) of those from young sea ice classified as Rhizobiales (Table 2). 16S rRNA gene reads classified as derived from chloroplasts were recovered from both the frost flower (464, 2%) and young sea ice (1815, 22%) libraries. A large fraction of the young sea ice 16S rRNA gene reads could not be classified at the level of order.
Table 2
Most abundant orders according to classification of 16S rRNA gene reads
Order | Number of reads | Percent of total 16S reads |
---|---|---|
Frost flower | ||
Rhizobiales | 7247 | 51.96 |
Actinomycetales | 332 | 2.38 |
Flavobacteriales | 181 | 1.30 |
Rickettsiales | 177 | 1.27 |
Stramenopiles | 134 | 0.96 |
Euglenozoa | 127 | 0.91 |
Oceanospirillales | 111 | 0.80 |
Rhodobacterales | 80 | 0.57 |
Sphingomonadales | 41 | 0.29 |
Mycoplasmatales | 41 | 0.29 |
Unclassified | 4676 | 33.53 |
Young sea ice | ||
Stramenopiles | 1441 | 21.56 |
Actinomycetales | 585 | 8.75 |
Rhizobiales | 541 | 8.10 |
Euglenozoa | 128 | 1.92 |
Rickettsiales | 64 | 0.96 |
Flavobacteriales | 61 | 0.91 |
Oceanospirillales | 35 | 0.52 |
Chlorophyta | 31 | 0.46 |
Clostridiales | 19 | 0.28 |
Mycoplasmatales | 15 | 0.22 |
Unclassified | 3763 | 56.30 |
Order | Number of reads | Percent of total 16S reads |
---|---|---|
Frost flower | ||
Rhizobiales | 7247 | 51.96 |
Actinomycetales | 332 | 2.38 |
Flavobacteriales | 181 | 1.30 |
Rickettsiales | 177 | 1.27 |
Stramenopiles | 134 | 0.96 |
Euglenozoa | 127 | 0.91 |
Oceanospirillales | 111 | 0.80 |
Rhodobacterales | 80 | 0.57 |
Sphingomonadales | 41 | 0.29 |
Mycoplasmatales | 41 | 0.29 |
Unclassified | 4676 | 33.53 |
Young sea ice | ||
Stramenopiles | 1441 | 21.56 |
Actinomycetales | 585 | 8.75 |
Rhizobiales | 541 | 8.10 |
Euglenozoa | 128 | 1.92 |
Rickettsiales | 64 | 0.96 |
Flavobacteriales | 61 | 0.91 |
Oceanospirillales | 35 | 0.52 |
Chlorophyta | 31 | 0.46 |
Clostridiales | 19 | 0.28 |
Mycoplasmatales | 15 | 0.22 |
Unclassified | 3763 | 56.30 |
Table 2
Most abundant orders according to classification of 16S rRNA gene reads
Order | Number of reads | Percent of total 16S reads |
---|---|---|
Frost flower | ||
Rhizobiales | 7247 | 51.96 |
Actinomycetales | 332 | 2.38 |
Flavobacteriales | 181 | 1.30 |
Rickettsiales | 177 | 1.27 |
Stramenopiles | 134 | 0.96 |
Euglenozoa | 127 | 0.91 |
Oceanospirillales | 111 | 0.80 |
Rhodobacterales | 80 | 0.57 |
Sphingomonadales | 41 | 0.29 |
Mycoplasmatales | 41 | 0.29 |
Unclassified | 4676 | 33.53 |
Young sea ice | ||
Stramenopiles | 1441 | 21.56 |
Actinomycetales | 585 | 8.75 |
Rhizobiales | 541 | 8.10 |
Euglenozoa | 128 | 1.92 |
Rickettsiales | 64 | 0.96 |
Flavobacteriales | 61 | 0.91 |
Oceanospirillales | 35 | 0.52 |
Chlorophyta | 31 | 0.46 |
Clostridiales | 19 | 0.28 |
Mycoplasmatales | 15 | 0.22 |
Unclassified | 3763 | 56.30 |
Order | Number of reads | Percent of total 16S reads |
---|---|---|
Frost flower | ||
Rhizobiales | 7247 | 51.96 |
Actinomycetales | 332 | 2.38 |
Flavobacteriales | 181 | 1.30 |
Rickettsiales | 177 | 1.27 |
Stramenopiles | 134 | 0.96 |
Euglenozoa | 127 | 0.91 |
Oceanospirillales | 111 | 0.80 |
Rhodobacterales | 80 | 0.57 |
Sphingomonadales | 41 | 0.29 |
Mycoplasmatales | 41 | 0.29 |
Unclassified | 4676 | 33.53 |
Young sea ice | ||
Stramenopiles | 1441 | 21.56 |
Actinomycetales | 585 | 8.75 |
Rhizobiales | 541 | 8.10 |
Euglenozoa | 128 | 1.92 |
Rickettsiales | 64 | 0.96 |
Flavobacteriales | 61 | 0.91 |
Oceanospirillales | 35 | 0.52 |
Chlorophyta | 31 | 0.46 |
Clostridiales | 19 | 0.28 |
Mycoplasmatales | 15 | 0.22 |
Unclassified | 3763 | 56.30 |
Assembly of the frost flower reads produced 13 contigs longer than 10 000 bp, with the longest reaching 18 102 bp (Table 3). No 16S rRNA gene genes were recovered from these contigs. Two contigs displayed high similarity to sequences within the NCBI nt database, Contig_1473 aligned to the genome of Sinorhizobium meliloti Rm41 (score = 4947) and Contig_171 aligned to plasmid pSymB of S. meliloti 1021 (score = 2105).
Table 3
blastn results for assembled contigs longer than 10 000 bp, searched against nt as single queries
Contig | Number of putative CDS | Top hit in nt | Max score |
---|---|---|---|
NODE_1473 | 13 | Sinorhizobium meliloti Rm41 complete genome | 4947 |
NODE_171 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 2105 |
NODE_229 | 6 | Rhodopseudomonas palustris DX-1, complete genome | 1274 |
NODE_448 | 8 | Agrobacterium tumefaciens str. C58 circular chromosome, complete sequence | 874 |
NODE_870 | 17 | Sinorhizobium meliloti 1021 plasmid pSymA, complete sequence | 807 |
NODE_2 | 10 | Rhizobium tropici CIAT 899, complete genome | 740 |
NODE_100 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 621 |
NODE_213 | 8 | Rhizobium leguminosarum insertion sequence ISR1F7-2 putative transposase (Tnp) gene, complete cds | 522 |
NODE_525 | 8 | Rhizobium leguminosarum bv. viciae plasmid pRL7 complete genome, strain 3841 | 515 |
NODE_14 | 12 | Sinorhizobium fredii NGR234, complete genome | 497 |
NODE_389 | 6 | Caulobacter sp. K31 plasmid pCAUL01, complete sequence | 345 |
NODE_411 | 8 | Sinorhizobium meliloti SM11, complete genome | 345 |
NODE_388 | 2 | Burkholderia vietnamiensis G4 plasmid pBVIE03, complete sequence | 69.8 |
Contig | Number of putative CDS | Top hit in nt | Max score |
---|---|---|---|
NODE_1473 | 13 | Sinorhizobium meliloti Rm41 complete genome | 4947 |
NODE_171 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 2105 |
NODE_229 | 6 | Rhodopseudomonas palustris DX-1, complete genome | 1274 |
NODE_448 | 8 | Agrobacterium tumefaciens str. C58 circular chromosome, complete sequence | 874 |
NODE_870 | 17 | Sinorhizobium meliloti 1021 plasmid pSymA, complete sequence | 807 |
NODE_2 | 10 | Rhizobium tropici CIAT 899, complete genome | 740 |
NODE_100 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 621 |
NODE_213 | 8 | Rhizobium leguminosarum insertion sequence ISR1F7-2 putative transposase (Tnp) gene, complete cds | 522 |
NODE_525 | 8 | Rhizobium leguminosarum bv. viciae plasmid pRL7 complete genome, strain 3841 | 515 |
NODE_14 | 12 | Sinorhizobium fredii NGR234, complete genome | 497 |
NODE_389 | 6 | Caulobacter sp. K31 plasmid pCAUL01, complete sequence | 345 |
NODE_411 | 8 | Sinorhizobium meliloti SM11, complete genome | 345 |
NODE_388 | 2 | Burkholderia vietnamiensis G4 plasmid pBVIE03, complete sequence | 69.8 |
Table 3
blastn results for assembled contigs longer than 10 000 bp, searched against nt as single queries
Contig | Number of putative CDS | Top hit in nt | Max score |
---|---|---|---|
NODE_1473 | 13 | Sinorhizobium meliloti Rm41 complete genome | 4947 |
NODE_171 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 2105 |
NODE_229 | 6 | Rhodopseudomonas palustris DX-1, complete genome | 1274 |
NODE_448 | 8 | Agrobacterium tumefaciens str. C58 circular chromosome, complete sequence | 874 |
NODE_870 | 17 | Sinorhizobium meliloti 1021 plasmid pSymA, complete sequence | 807 |
NODE_2 | 10 | Rhizobium tropici CIAT 899, complete genome | 740 |
NODE_100 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 621 |
NODE_213 | 8 | Rhizobium leguminosarum insertion sequence ISR1F7-2 putative transposase (Tnp) gene, complete cds | 522 |
NODE_525 | 8 | Rhizobium leguminosarum bv. viciae plasmid pRL7 complete genome, strain 3841 | 515 |
NODE_14 | 12 | Sinorhizobium fredii NGR234, complete genome | 497 |
NODE_389 | 6 | Caulobacter sp. K31 plasmid pCAUL01, complete sequence | 345 |
NODE_411 | 8 | Sinorhizobium meliloti SM11, complete genome | 345 |
NODE_388 | 2 | Burkholderia vietnamiensis G4 plasmid pBVIE03, complete sequence | 69.8 |
Contig | Number of putative CDS | Top hit in nt | Max score |
---|---|---|---|
NODE_1473 | 13 | Sinorhizobium meliloti Rm41 complete genome | 4947 |
NODE_171 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 2105 |
NODE_229 | 6 | Rhodopseudomonas palustris DX-1, complete genome | 1274 |
NODE_448 | 8 | Agrobacterium tumefaciens str. C58 circular chromosome, complete sequence | 874 |
NODE_870 | 17 | Sinorhizobium meliloti 1021 plasmid pSymA, complete sequence | 807 |
NODE_2 | 10 | Rhizobium tropici CIAT 899, complete genome | 740 |
NODE_100 | 11 | Sinorhizobium meliloti 1021 plasmid pSymB, complete sequence | 621 |
NODE_213 | 8 | Rhizobium leguminosarum insertion sequence ISR1F7-2 putative transposase (Tnp) gene, complete cds | 522 |
NODE_525 | 8 | Rhizobium leguminosarum bv. viciae plasmid pRL7 complete genome, strain 3841 | 515 |
NODE_14 | 12 | Sinorhizobium fredii NGR234, complete genome | 497 |
NODE_389 | 6 | Caulobacter sp. K31 plasmid pCAUL01, complete sequence | 345 |
NODE_411 | 8 | Sinorhizobium meliloti SM11, complete genome | 345 |
NODE_388 | 2 | Burkholderia vietnamiensis G4 plasmid pBVIE03, complete sequence | 69.8 |
Small proportions (frost flower: 1.68%, young sea ice: 1.9%) of the QC'd metagenomic reads mapped to available reference genomes. In both metagenomes, the chromosome of Candidatus Pelagibacter ubique was the deepest and most broadly covered. In the frost flower metagenome, P. ubique coverage was 3.62 and breadth was 0.60 (Fig. 1), whereas in the young sea ice, P. ubique coverage was 1.38 and breadth was 0.41, indicating the lower overall sequence depth of the young sea ice sample. When ranked by coverage × breadth, the other well-covered genomes in the frost flower sample were either other members of the Rhizobiales, particularly Ochrobactrum anthropi ATCC 49188, Sinorhizobium fredii NGR234, and Sinorhizobium meliloti GR4, or the alphaproteobacterium Candidatus Pelagibacter sp. IMCC9063. Other than Pelagibacter spp., the first non-Rhizobiales in the mapping order was Sphingobium japonicum UT26S, which had sequence coverage in relatively few regions (coverage = 0.48, breadth = 0.01). The most highly covered genome in the young sea ice sample after P. ubique was Candidatus Hodgkinia cicadicola (coverage = 0.29, breadth = 0.01).
Fig. 1
Coverage of chromosomes and plasmids in the frost flower metagenome. (a) All available chromosomes and plasmids; (b) all Rhizobiales chromosomes and plasmids; (c) distribution of breadth for chromosomes and plasmids of the Rhizobiales; (d) distribution of coverage for chromosomes and plasmids of the Rhizobiales.
To further evaluate the taxonomic makeup of the community, we considered only those reads that mapped uniquely to a single genome or to multiple genomes from a single species. These uniquely mapped reads were considered diagnostic of the presence of closely related strains within the metagenome samples. The species most covered by uniquely mapped reads from the frost flower metagenome was Candidatus Pelagibacter ubique at 2.61, followed by Candidatus Pelagibacter sp. IMCC9063 (0.21), Agrobacterium vitis (0.08), Rhodococcus erythropolis (0.05), and Candidatus Nitrosopumilus sp. AR2 (0.05). The species most covered by uniquely mapped reads from the young sea ice metagenome was Candidatus P. ubique at 0.97, followed by Pelagibacter sp. IMCC9063 (0.10) and Nitrosopumilus sp. AR2 (0.03). Among the Rhizobiales, many genes required for plant infection and symbiosis are contained on plasmids. By Student's _t_-test on both breadth (T = 2.53, P = 0.012, d.f. = 254) and coverage (T = 10.10, P ≪ 10−6, d.f. = 235), plasmids were poorly covered compared with chromosomes among the Rhizobiales (Fig. 1b–d). The exceptions to this trend were plasmids from Agrobacterium vitis S4 (NC_011986), Ochrobactrum anthropic ATCC49188 (NC_009670 and NC_009671), Methylobacterium radiotolerans JCM2831 (NC_010510), and Sinorhizobium fredii HH103 (NT_187146) (Fig. 2).
Fig. 2
Coverage for chromosomes and plasmids for strains with well-covered plasmids.
The distribution of reads across the reference genomes was analyzed to determine whether potential systematic differences existed between the organisms detected in our samples and their cultured relatives. As a negative control for this comparison, we included Candidatus Pelagibacter ubique, with previously described variable genomic regions (Rusch et al., ; Wilhelm et al., ). At a resolution of 5000 bp, a bin larger than the average gene, we observed only five regions of low sequence coverage in the genome of P. ubique (Fig. 3), with the largest corresponding to the expected position of the HVR2 hypervariable region (Wilhelm et al., ). Genes within these regions primarily encoded proteins involved in substrate utilization, energy transfer, and biosynthesis. In contrast, the chromosome of Sinorhizobium fredii NGR234, the most covered root-nodulating Rhizobiales and fourth most covered genome overall, contained an abundance of coverage gaps distributed along its genome (Fig. 3). Similar to the P. ubique analysis, some genes within these coverage gaps encoded proteins for biosynthesis and energy transfer. Importantly, when the gap regions from the five best covered Rhizobiales genomes were analyzed, 32 genes encoding nodulation factors were found to be lacking in the observed Rhizobiales, as well as 230 transposases, integrases, and other genes involved in DNA transfer (Table S1).
Fig. 3
Coverage across the chromosomes of Candidatus Pelagibacter ubique (top) and _Sinorhizbium fredii_HH103 (bottom). Gap regions, highlighted by the red boxes near to the top of the plot, are defined as any region > 5000 nt with zero coverage.
In blast analysis of the nodulating proteins (NodA–J), RecA-normalized coverage varied by protein (Table 1). NodIG were well covered at 2.242 and 0.43, respectively, while NodBCEJ were more moderately covered between 0.1 and 0.133. NodADFH had relatively low coverage. Several of the groups of predicted proteins were present in different proportions between the frost flower and young sea ice metagenomes. Considering RecA-normalized coverage, the NodDGJ proteins were more represented in the frost flower dataset (Table 1).
Metabolic profiles
blastx analysis of the frost flower metagenome identified a large number of reads with significant homology to proteins of biogeochemical interest (Table 1). These alignments signify potential for the catabolism of glycine betaine (a compatible solute used under hypersaline conditions), dimethylsulfide (DMS), and dimethylsulfoniopropionate (DMSP) and for mercury resistance and methylhalide production. Relatively few reads aligned to the NifH nitrogenase, CmuA methyltransferase (which functions in methylhalide catabolism), or ice structuring databases (consisting of ice binding and ice nucleating proteins).
Phylogenetic placement of translated reads from the frost flower metagenome placed more than half of the haloperoxidase and DMS degradation reads with reference sequences from the Rhizobiales (Table 4). For DMS dehydrogenase and monooxgenase, which have relatively few representative sequences in UniProt, all Rhizobiales placements were to an edge corresponding to Hyphomicrobium sulfonivorans. For haloperoxidase, the placements were more varied, with edges corresponding to Rhizobium etli and Agrobacterium spp. accounting for most of the Rhizobiales placements. Proportionally fewer betaine methyltransferase reads were associated with the Rhizobiales, although the absolute number was still high. A large number of reads were placed to a close homolog of the Sinorhizobium meliloti betaine methyltransferase in Cupriavidus nectar and to the bifurcating node for these sequences (Fig. 4). Fewer indoleacetomide hydrolase (IAH) reads placed with the Rhizobiales and all of those placements were to Brucella spp. There are no reported Rhizobiales DMSP catabolism or degradation genes. Reads associated with DMSP lyase (which liberates the gas DMS) and DMSP demethylase were placed with non-Rhizobiales species, primarily Ruegeria pomeroyi (DddQ, DddW, DmdA), Pelagibacter ubique (DmdA), and Aspergillus sydowii (DddP). Considering RecA-normalized coverage, the proteins for DMS catabolism and haloperoxidases were more represented in the frost flower dataset. The ice structuring proteins were better covered in the young sea ice dataset (Table 1).
Table 4
Putative peptides placed on edges corresponding to proteins from the Rhizobiales
Protein group | Number of reads | Fraction of total |
---|---|---|
Indoleacetomide hydrolase | 73 | 0.116 |
Betaine methyltransferase | 406 | 0.280 |
DMS catabolism | 1137 | 0.520 |
Haloperoxidase | 467 | 0.525 |
MerA | 0 | 0 |
Protein group | Number of reads | Fraction of total |
---|---|---|
Indoleacetomide hydrolase | 73 | 0.116 |
Betaine methyltransferase | 406 | 0.280 |
DMS catabolism | 1137 | 0.520 |
Haloperoxidase | 467 | 0.525 |
MerA | 0 | 0 |
Table 4
Putative peptides placed on edges corresponding to proteins from the Rhizobiales
Protein group | Number of reads | Fraction of total |
---|---|---|
Indoleacetomide hydrolase | 73 | 0.116 |
Betaine methyltransferase | 406 | 0.280 |
DMS catabolism | 1137 | 0.520 |
Haloperoxidase | 467 | 0.525 |
MerA | 0 | 0 |
Protein group | Number of reads | Fraction of total |
---|---|---|
Indoleacetomide hydrolase | 73 | 0.116 |
Betaine methyltransferase | 406 | 0.280 |
DMS catabolism | 1137 | 0.520 |
Haloperoxidase | 467 | 0.525 |
MerA | 0 | 0 |
Fig. 4
Phylogenetic placement of betaine methyltransferase read translations from the frost flower metagenome on a reference tree of UniProt sequences. The width of each edge is proportional to the number of reads placed on the edge. Exact values are given in the histogram, by node number on the reference tree. Node numbers for selected locations are given in white on the reference tree. Confidence values for the reference tree are given in black at each bifurcation.
Discussion
Our analysis of the frost flower and young sea ice metagenomes suggests a partitioning of metabolic function between these two environments, with genes for DMS catabolism and haloperoxidase production abundant and increased within frost flowers. The presence of these genes appears to be connected to the occurrence of bacteria belonging to the Rhizobiales, a bacterial order most commonly (but not exclusively) associated with soil. Despite the taxonomic overlap with soil, the genetic composition of the frost flower microbial community is generally inconsistent with a terrestrial origin. Root-nodulating bacteria are found in tundra soil (Bordeleau & Prévost, ), but there is little evidence of nodulating bacteria in the frost flower metagenome. The complex process by which symbiotic rhizobia infect terrestrial plants and carry out a mutualistic interaction is mediated by over 50 proteins (Downie, ). In this analysis, we searched for genes encoding ten key nodulation proteins, NodA-J. Several of these proteins were almost entirely absent by our analysis, most notably NodABCDJ, which are present in all root-nodulating Rhizobiales (Downie, ). NodFH, respectively, a lipochitin oligonucleotide production protein and a sulfation protein that are taxonomically restricted to the Sinorhizobium and Rhizobium, were also present only at low abundance. Of the remaining nodulation proteins, NodGI were well represented in these metagenomes. NodG is a ribitol dehydrogenase and a close homolog to the housekeeping protein FadG (López-Lara & Geiger, ). Placement of translated frost flower NodG reads on a tree of NodG and FadG proteins suggests that most of the NodG blastx hits are in fact FadG read translations (Fig. 5). NodI, a membrane transporter, could represent a similar case of a protein required for, but not indicative of, symbiosis.
Fig. 5
Phylogenetic placement of NodG and FabG read translations from the frost flower metagenome on a reference tree of UniProt sequences. Reference sequences annotated as FabG are shown in red, reference sequences annotated as NodG are shown in gray, along with ambiguous nodes.
The observed Rhizobiales also appear deficient in genes encoding IAH and NifH. IAH catalyzes the final step in the conversion of tryptophan to indole-3-acetate (IAA), a known signaling molecule between bacteria and plants in the terrestrial environment (Patten & Glick, ). Although a number of translated reads were classified with IAH (Table 1), none of them were placed with IAH proteins from the root-nodulating genera Rhizobium, Bradyrhizobium, or Agrobacterium. A majority of the IAH reads grouped with IAH homologs belonging to Rhodococcus, Cupriavidus, and Ralstonia spp. NifH enables nitrogen fixation, widely believed to be the primary benefit plants derive from root-nodulating bacteria. The few NifH read translations in the frost flower metagenome classified with NifH proteins from the Archaea and the genus Paenibacillus, suggest that the frost flower Rhizobiales community was not equipped to fix nitrogen.
Many Rhizobiales genes required for plant infection and symbiosis are contained on plasmids. An overall absence of known symbiosis plasmids in our samples further supports an alternative ecology for these putatively marine Rhizobiales. Limited sequence homology was detected within our frost flower sample to predicted proteins encoded over a short segment of the symbiosis plasmid pSymB of Sinorhizobium meliloti 1021. The homologous proteins were amino acid transporters and a transcription regulator rather than proteins diagnostic of a root-nodulating lifestyle commonly found on the plasmids such as NodA-J. The assembled contig likely represents a conserved fragment of a much altered plasmid or a chromosomal fragment in a bacterium that shares some ancestry (whether vertical or horizontal) with S. meliloti 1021.
Despite the general absence of key nodulation genes known from terrestrial plant symbioses, other evidence indicates that the microbial community present in frost flowers was equipped to live in association with an aquatic or marine photosynthetic community. This evidence includes detection of genes for the catabolism of the important algal exudates glycine betaine, DMSP, and DMS, all of which have important biogeochemical consequences (Welsh, ). Betaine homocysteine methyltransferase mediates the first step in the bacterial conversion of glycine betaine to pyruvate and ammonia. Frost flower betaine homocysteine methyltransferase reads classified primarily with Sinorhizoium meliloti and a close homolog in the ß-proteobacterium Cupriavidus necator (a genus not otherwise present) suggesting that this function belonged predominantly to the Rhizobiales. DMSP catabolism is nearly ubiquitous among marine bacteria, and our observed predicted DMSP sequences were associated with typical marine strains. DMS catabolism is less common, and the DMS sequences were associated primarily with the aquatic Rhizobiales genus Hyphomicrobium.
There is some similarity between the frost flower microbial community and aquatic microbial communities. Previously, we noted that 16S rRNA genes recovered from frost flowers are closely homologous to the free-living aquatic bacterium Rhizobium aggregatum and Rhizobium capsulatus, formerly of the genus Blastobacter (Bowman et al., ). Although genomes are not available for either of these Rhizobium, our recovery of DMS dehydrogenase reads with close homology to that of Hyphomicrobium sulfinovorans, a Rhizobiales known from Antarctic lake sediment (Moosvi et al., ), is suggestive, as is the observation of _Hyphomicrobium_-like morphologies on marine diatoms (Kaczmarska et al., ). Although we did not find 16S rRNA gene or cmuA gene sequence (Nadalig et al., ) evidence for H. sulfinovorans in our metagenomes, the taxonomic relationship with Rhizobium, and overlapping habitat with R. aggregatum and R. capsulatus, would have provided opportunities for the vertical or horizontal transfer of DMS dehydrogenase genes to the Rhizobium genus.
Because of the presence of genes for key processes and the demonstrated association between Rhizobiales and phytoplankton in other environments (Bidle & Azam, ; Makk et al., ; Jasti et al., ; Stevenson & Waterbury, ; Kazamia et al., ), we previously hypothesized that the observed Rhizobiales were originally associated with marine phytoplankton (Bowman et al., ). In this model, bacteria physically attached to phytoplankton cells or associated with the phycosphere (the region around a phytoplankton cell dominated by cell exudates) would have been entrained along with their hosts in newly formed sea ice. Once entrained, the larger cells would become immobilized. A dissociation of hosts and symbionts, induced by stress or physical force, would leave these smaller cells free to transport upwards through the ice. Physical attachment to large algal cells has long been considered a mechanism by which bacteria can enter sea ice (Grossmann & Dieckmann, ), but the adequacy of phycosphere association for bacterial entrainment remains to be tested. Analysis of the young sea ice 16S rRNA genes (Table 2) indicated the presence of a significant number of Stramenopile (diatom) chloroplasts that were much reduced by comparison in the frost flower metagenome.
An alternative to entrainment from the water column is eolian transport. Although the available data on bacterial abundance of springtime Arctic air or dust is very limited, it is possible to make a rough estimate of these contributions. We found dust deposition in combination with airborne bacterial abundance to be inadequate to explain the observed enrichment of bacteria in frost flowers relative to the underlying young sea ice (Supporting Information, Data S1). Supporting this conclusion are the low bacterial abundances in snow over sea ice adjacent to the frost flower field (Data S1) and in the vicinity a month earlier (Ewert et al., ). A similar calculation, however, suggests a role for lithogenic particles in the delivery of bacteria to young sea ice. Frazil ice can scavenge seafloor sediment along with biotic and abiotic suspended particles, delivering this material to young sea ice. Although frazil ice formation is typically considered to be a fall phenomenon (Kempema et al., ), the water column was sufficiently cool to allow frazil ice formation during the frost flower growth period. A sediment concentration of 100 g L−1, at the low end of the range of values given for visible sediment layers in Eicken et al. (), would be sufficient to explain the observed accumulation of bacteria. Considering that a sediment layer was not visible in the young sea ice and that a number of chloroplasts were present in the young sea ice metagenome, a mix of less sediment and phytoplankton seems more plausible (see Data S1).
The surface of young sea ice can be a particularly extreme environment for bacteria regardless of their origin, especially in winter when temperatures are very low at the ice–air interface and salinities very high due to expelled brines (Bowman & Deming, ). The young ice surface is also a highly oxidizing environment (Douglas et al., ) and some Rhizobiales produce methylhalides as a byproduct of antioxidant activity (Amachi et al., ; Bengtson et al., ). Although atmospheric halide chemistry above Arctic polynyas has received much attention, measurements of halocarbons were only recently made in young sea ice. In that study, a biologic source was invoked to explain the observed flux of halocarbons from newly formed sea ice (Granfors et al., ), although the responsible organisms were not identified. Our observation of bacterial haloperoxidases in both the young sea ice and frost flower metagenomes, and particularly associated with the Rhizobiales in frost flowers, is suggestive of a potential role in volatile halide production.
Heavy metal toxicity is another biologic stressor noted by previous geochemical studies of the young sea ice environment, where mercury levels in particular were elevated in frost flowers (Douglas et al., ). Genes coding for MerA, the mercury reductase that confers mercury resistance, have been observed in sea ice (Poulain et al., ; Møller et al., ); the detoxification mechanism volatilizes the mercury, impacting atmospheric chemistry and the fate of the mercury. Consistent with the observations of MerA in sea ice, a large number of read translations were aligned to MerA proteins in our blastx analysis, although none of these were homologous to known Rhizobiales MerA. Instead, these read translations placed with the Archaeal genus Nitrososphaera.
Conclusions
A comparative analysis of two metagenomes, one from a sampling of frost flowers and one from the underlying young sea ice, confirmed our previous 16S rRNA gene-based observation that the frost flower sample was enriched in members of the bacterial order Rhizobiales. Detailed analysis, including recruitment of reads to the available Rhizobiales genomes, suggests that the observed bacteria were genetically distinct from the well-studied strains responsible for root nodulation in terrestrial temperate environments. Specifically, the observed bacteria appeared to lack symbiosis plasmids and key genes for nitrogen fixation and root nodulation that characterize members of the terrestrial rhizosphere. Instead, the community appeared adapted to live in close association with marine or aquatic phytoplankton, carrying genes for the catabolism of common algal exudates.
Regardless of their origin, the presence of the metabolically diverse Rhizobiales in the young sea ice environment could have implications for biogeochemical cycles there. The observed Rhizobiales have the potential for DMS and betaine glycine catabolism, and halocarbon production, with the larger community additionally capable of DMSP catabolism and mercury reduction. Our findings are limited by the fact that these metagenomes were sourced from samples that represent a single location in the flaw-lead system of the coastal Arctic. Observations of other microbial communities in the biologically under-studied young sea ice environment are needed to determine the broader implications of our results for the polar marine and atmospheric environments, the efficacy of the proposed transport mechanisms, and whether gene potential is translated to activity at the sea ice–atmosphere interface.
Acknowledgements
This project was funded by NSF OPP award 0908724 to JWD and by the Walters Endowed Professorship. JSB was supported by an NSF IGERT fellowship through the UW Astrobiology Program and an EPA STAR fellowship. EVA was supported by the Moore Foundation Marine Microbial Investigator program. Field support was provided by Umiaq and the Barrow Arctic Science Consortium; we are particularly grateful to Nok Acker and Lewis Brower for their assistance. The manuscript was substantially improved by input from two anonymous reviewers.
References
Altschul
SF
Gish
W
Miller
W
Myers
EW
Lipman
DJ
(
1990
)
Basic local alignment search tool
.
J Mol Biol
215
:
403
–
410
.
Alvarez-Aviles
L
Simpson
WR
Douglas
TA
Sturm
M
Perovich
D
Domine
F
(
2008
)
Frost flower chemical composition during growth and its implications for aerosol production and bromine activation
.
J Geophys Res
113
:
D21304
.
Amachi
S
Kamagata
Y
Kanagawa
T
Muramatsu
Y
(
2001
)
Bacteria mediate methylation of iodine in marine and terrestrial environments
.
Appl Environ Microbiol
67
:
2718
–
2722
.
Aslam
S
Underwood
G
Kaartokallio
H
et al. (
2012
)
Dissolved extracellular polymeric substances (dEPS) dynamics and bacterial growth during sea ice formation in an ice tank study
.
Polar Biol
35
:
661
–
676
.
Bairoch
A
Bougueleret
L
Altairac
S
et al. (
2009
)
The Universal Protein Resource (UniProt) 2009
.
Nucleic Acids Res
37
:
D169
–
D174
.
Beine
H
Anastasio
C
Domine
F
et al. (
2012
)
Soluble chromophores in marine snow, seawater, sea ice and frost flowers near Barrow, Alaska
.
J Geophys Res
117
:
D00R15
.
Bengtson
P
Bastviken
D
Öberg
G
(
2012
)
Possible roles of reactive chlorine II: assessing biotic chlorination as a way for organisms to handle oxygen stress
.
Environ Microbiol
15
:
991
–
1000
.
Bidle
KD
Azam
F
(
2001
)
Bacterial control of silicon regeneration from diatom detritus: significance of bacterial ectohydrolases and species identity
.
Limnol Oceanogr
46
:
1606
–
1623
.
Boratyn
GM
Schaffer
A
Agarwala
R
Altschul
SF
Lipman
DJ
Madden
TL
(
2012
)
Domain enhanced lookup time accelerated BLAST
.
Biol Direct
7
:
12
.
Bordeleau
L
Prévost
D
(
1994
)
Nodulation and nitrogen fixation in extreme environments
.
Plant Soil
161
:
115
–
125
.
Bowman
JS
Deming
JW
(
2010
)
Elevated bacterial abundance and exopolymers in saline frost flowers with implications for atmospheric chemistry and microbial dispersal
.
Geophys Res Lett
37
:
L13501
.
Bowman
JS
Larose
C
Vogel
TM
Deming
JW
(
2013
)
Selective occurrence of Rhizobiales in frost flowers on the surface of young sea ice near Barrow, Alaska and distribution in the polar marine rare biosphere
.
Environ Microbiol Rep
5
:
575
–
582
.
Douglas
TA
Sturm
M
Simpson
WR
Brooks
S
Lindberg
SE
Perovich
DK
(
2005
)
Elevated mercury measured in snow and frost flowers near Arctic sea ice leads
.
Geophys Res Lett
32
:
L04502
.
Douglas
TA
Domine
F
Barret
M
et al. (
2012
)
Frost flowers growing in the Arctic ocean-atmosphere-sea ice-snow interface: 1. Chemical composition
.
J Geophys Res
117
:
D00R09
.
Downie
JA
(
1998
)
Functions of rhizobial nodulation genes
.
The Rhizobiaceae
(
Spaink
HP
Kondorosi
A
Hooykaas
PJJ
, eds), pp.
387
–
402
.
Springer
,
the Netherlands
.
Eddy
SR
(
1998
)
Profile hidden Markov models
.
Bioinformatics
14
:
755
–
763
.
Eicken
H
Gradinger
R
Gaylord
A
Mahoney
A
Rigor
I
Melling
H
(
2005
)
Sediment transport by sea ice in the Chukchi and Beaufort Seas: increasing importance due to changing ice conditions?
Deep Sea Res Part II
52
:
3281
–
3302
.
Ewert
M
Carpenter
S
Colangelo-Lillis
J
Deming
J
(
2013
)
Bacterial and extracellular polysaccharide content of brine-wetted snow over Arctic winter first-year sea ice
.
J Geophys Res Oceans
18
:
726
–
735
.
Granfors
A
Andersson
M
Chierici
M
et al. (
2013
)
Biogenic halocarbons in young Arctic sea ice and frost flowers
.
Mar Chem
155
:
124
–
134
.
Grossmann
S
Dieckmann
GS
(
1994
)
Bacterial standing stock, activity, and carbon production during formation and growth of sea ice in the Weddell Sea, Antarctica
.
Appl Environ Microbiol
60
:
2746
–
2753
.
Iverson
V
Morris
RM
Frazar
CD
Berthiaume
CT
Morales
RL
Armbrust
EV
(
2012
)
Untangling genomes from metagenomes: revealing an uncultured class of marine Euryarchaeota
.
Science
335
:
587
–
590
.
Jasti
S
Sieracki
ME
Poulton
NJ
Giewat
MW
Rooney-Varga
JN
(
2005
)
Phylogenetic diversity and specificity of bacteria closely associated with Alexandrium spp. and other phytoplankton
.
Appl Environ Microbiol
71
:
3483
–
3494
.
Kaczmarska
I
Ehrman
JM
Bates
SS
Green
DH
Léger
C
Harris
J
(
2005
)
Diversity and distribution of epibiotic bacteria on Pseudonitzschia multiseries (Bacillariophyceae) in culture, and comparison with those on diatoms in native seawater
.
Harmful Algae
4
:
725
–
741
.
Kazamia
E
Czesnick
H
Nguyen
TTV
et al. (
2012
)
Mutualistic interactions between vitamin B12-dependent algae and heterotrophic bacteria exhibit regulation
.
Environ Microbiol
14
:
1466
–
1476
.
Kempema
E
Reimnitz
E
Barnes
PW
(
1989
)
Sea ice sediment entrainment and rafting in the Arctic
.
J Sediment Res
59
:
308
–
317
.
Li
H
Durbin
R
(
2009
)
Fast and accurate short read alignment with Burrows–Wheeler transform
.
Bioinformatics
25
:
1754
–
1760
.
López-Lara
IM
Geiger
O
(
2001
)
The nodulation protein NodG shows the enzymatic activity of an 3-oxoacyl-acyl carrier protein reductase
.
Mol Plant Microbe Interact
14
:
349
–
357
.
Makk
J
Beszteri
B
Ács
É
Márialigeti
K
Szabó
K
(
2003
)
Investigations on diatom-associated bacterial communities colonizing an artificial substratum in the River Danube
.
Arch Hydrobiol
14
:
249
–
265
.
Matsen
F
Kodner
R
Armbrust
EV
(
2010
)
pplacer: linear time maximum-likelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree
.
BMC Bioinformatics
11
:
538
.
Meyer
F
Paarmann
D
D'Souza
M
et al. (
2008
)
The metagenomics RAST server – a public resource for the automatic phylogenetic and functional analysis of metagenomes
.
BMC Bioinformatics
9
:
386
.
Møller
A
Barkay
T
Hansen
M
et al. (
2013
)
Mercuric reductase genes (merA) and mercury resistance plasmids in High Arctic snow, freshwater and sea-ice brine
.
FEMS Microbiol Ecol
73
:
2230
–
2238
.
Moosvi
AS
Mcdonald
IR
Pearce
DA
Kelly
DP
Wood
AP
(
2005
)
Molecular detection and isolation from Antarctica of methylotrophic bacteria able to grow with methylated sulfur compounds
.
Syst Appl Microbiol
28
:
541
–
554
.
Nadalig
T
Farhan
M
Haque
U
Roselli
S
Schaller
H
(
2011
)
Detection and isolation of chloromethane-degrading bacteria from the Arabidopsis thaliana phyllosphere, and characterization of chloromethane utilization genes
.
FEMS Microbiol Ecol
77
:
438
–
448
.
Patten
CL
Glick
BR
(
1996
)
Bacterial biosynthesis of indole-3-acetic acid
.
Can J Microbiol
42
:
207
–
220
.
Perovich
DK
Richter-Menge
JA
(
1994
)
Surface characteristics of lead ice
.
J Geophys Res
99
:
16341
–
16350
.
Poulain
AJ
Ni Chadhain
SM
Ariya
PA
et al. (
2007
)
Potential for mercury reduction by microbes in the high Arctic
.
Appl Environ Microbiol
73
:
2230
–
2238
.
Price
M
Dehal
P
Arkin
A
(
2010
)
FastTree 2 – Approximate maximum likelihood trees for large alignments
.
PLoS One
5
:
e9490
.
Punta
M
Coggill
PC
Eberhardt
RY
et al. (
2012
)
The Pfam protein families database
.
Nucleic Acids Res
40
:
D290
–
D301
.
Rankin
AM
Wolff
EW
Martin
S
(
2002
)
Frost flowers: implications for tropospheric chemistry and ice core interpretation
.
J Geophys Res
107
:
4683
–
4695
.
Rusch
DB
Halpern
AL
Sutton
G
et al. (
2007
)
The Sorcerer II global ocean sampling expedition: Northwest Atlantic through eastern tropical Pacific
.
PLoS Biol
5
:
e77
.
Schloss
PD
Westcott
SL
Ryabin
T
et al. (
2009
)
Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities
.
Appl Environ Microbiol
75
:
7537
–
7541
.
Sievers
F
Wilm
A
Dineen
D
et al. (
2011
)
Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega
.
Mol Syst Biol
7
:
539
.
Stamatakis
A
(
2006
)
RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models
.
Bioinformatics
22
:
2688
–
2690
.
Stevenson
BS
Waterbury
JB
(
2006
)
Isolation and identification of an epibiotic bacterium associated with heterocystous Anabaena cells
.
Biol Bull
210
:
73
–
77
.
Style
RW
Worster
MG
(
2009
)
Frost flower formation on sea ice and lake ice
.
Geophys Res Lett
36
:
L11501
.
Welsh
DT
(
2000
)
Ecological significance of compatible solute accumulation by micro-organisms: from single cells to global climate
.
FEMS Microbiol Rev
24
:
263
–
290
.
Wilhelm
L
Tripp
HJ
Givan
S
Smith
D
Giovannoni
S
(
2007
)
Natural variation in SAR11 marine bacterioplankton genomes inferred from metagenomic data
.
Biol Direct
2
:
27
.
Zerbino
DR
(
2010
)
Using the Velvet de novo assembler for short-read sequencing technologies
.
Curr Protoc Bioinformatics
31
:
11.5.1
–
11.5.12
.
Author notes
Editor: Johanna Laybourn-Parry
NSF
0908724
Walters Endowed
NSF IGERT
UW Astrobiology
EPA STAR
Moore Foundation Marine Microbial Investigator
Umiaq and the Barrow Arctic Science Consortium
© 2014 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd