AEM Accepts, published online ahead of print on 29 August 2014 Appl. Environ. Microbiol. doi:10.1128/AEM.01525-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.

1

High diversity of tailed phages, eukaryotic viruses and virophage-like elements in the

2

metaviromes of Antarctic soils

3 4

Olivier Zablockia, Lonnie van Zylb, Evelien M. Adriaenssensa, Enrico Rubagottia, Marla

5

Tuffinb, Craig Caryc, and Don Cowana#

6 7

Centre for Microbial Ecology and Genomics, University of Pretoria, Pretoria, South Africaa;

8

Institute for Microbial Biotechnology and Metagenomics, University of the Western Cape,

9

Bellville, South Africab; School of Biological Sciences, University of Waikato, Hamilton,

10

New Zealandc

11 12

Running Head: Metaviromics of Antarctic soil communities

13 14

#Address correspondence to Don A. Cowan, [email protected]

15 16 17 18 19 20 21 22 23 1

24

Abstract

25

The metaviromes of two distinct Antarctic hyperarid desert soil communities have

26

been characterized. Hypolithic communities, cyanobacteria-dominated assemblages situated

27

on the ventral surfaces of quartz pebbles embedded in the desert pavement, showed higher

28

virus diversity than surface soils, which correlated with previous bacterial community

29

studies. Prokaryotic viruses (i.e. phages) represented the largest viral component (particularly

30

Mycobacterium phages) in both habitats, with an identical hierarchical sequence abundance

31

of tailed phage families (Siphoviridae > Myoviridae > Podoviridae). No Archaeal viruses

32

were found. Unexpectedly, cyanophages were poorly represented in both metaviromes and

33

were phylogenetically distant from currently characterized cyanophages. Putative phage

34

genomes were assembled and showed a high level of unaffiliated genes, mostly from

35

hypolithic viruses. Moreover, unusual gene arrangements were observed, in which eukaryotic

36

and prokaryotic virus-derived genes were found within identical genome segments.

37

Phycodnaviridae and Mimiviridae viruses were the second most abundant taxa and more

38

numerous within open soil. Novel virophage- like sequences (within the Sputnik clade) were

39

identified. These findings highlight a high virus diversity and novel species discovery

40

potential within Antarctic hyperarid soils and may serve as a starting point for future studies

41

targeting specific viral groups.

42

Introduction

43

Antarctica is the coldest, driest place on Earth (1). Exposed soil areas comprise

44

approximately 0.4% of the continent’s surface and are mainly located in coastal areas,

45

particularly on the Antarctic Peninsula and in the McMurdo Dry Valleys (2). These mineral

46

soils are exposed to a range of “extreme” abiotic factors, including very low temperatures,

47

high soil salinity, low water availability and nutrient levels, high levels of UV radiation and

2

48

strong, cold winds descending from glaciers or mountain tops (katabatic). Due to these

49

conditions, the most morphologically distinct soil communities (i.e. type I, II and III

50

hypoliths) are associated with lithic surfaces (3, 4). Hypolithic communities, occurring on the

51

ventral surfaces of translucent quartz rocks, have been shown to be mostly composed of

52

phototrophic cyanobacterial species (5). These photoautotroph-dominated communities have

53

been attributed crucial roles within the Antarctic soil ecosystem, such as primary productivity

54

and nitrogen input (6, 7). While the composition of these communities is now reasonably well

55

understood (8-12), the associated viruses, with their potential to influence microbial

56

population dynamics and nutrient cycling via viral lysis (13), have yet to be characterised.

57

No comprehensive analyses of the collective viral genomic content

58

(i.e. the metavirome) of Antarctic soils have yet been published, with the limited number of

59

reported Antarctic viral metagenomic studies focusing on aquatic systems (14, 15, 16) and

60

Antarctic mega fauna such as seals (17) and penguins (18). Metaviromic surveys of saline

61

meromictic lakes have shown a high diversity of virus-like particles (mostly phages) and

62

several virophages (15, 16). To date, the few studies of viruses in Antarctic soils have all

63

focused on classical phage isolation and lytic induction experiments from culturable bacterial

64

species (19, 20, 21). Here we report a comprehensive characterization of virus diversity using

65

a metagenomic approach in Antarctic desert soils, with a focus on the dsDNA virus

66

composition of two common microhabitats: open surface soils and hypolithic communities.

67

Materials and Methods

68

Sampling location

69

Samples were collected from the Miers Valley, Ross Dependency in Eastern

70

Antarctica (GPS coordinates: 78º 05.6’ S, 163º 48.6’ E) during the austral summer period of

71

2011. For the open soil sample, 1.5 kg of surface soil (0-2cm depth) was collected from an

3

72

approx. 1m2 area at a single location. The hypolith sample consisted of 0.5 kg of hypolith

73

scrapings gathered aseptically from a collection of cyanobacterial-type hypoliths (n > 50)

74

from an area of approx. 50 m2. The open soil sample was recovered from within this area.

75

Samples were transferred and stored in sterile Whirl-Paks (Nasco, product number

76

B01445WA) at below 0ºC in the field and during transport, and at -80ºC in the laboratory.

77

Sample processing, DNA extraction and sequencing

78

Processing of both sample was performed similar to (22). Both the open soil sample

79

and bulked hypolithic samples were suspended in 3L of de-ionized water and shaken

80

vigorously. Solids were allowed to settle and the supernatant was decanted. The process was

81

repeated and both supernatants were mixed. The supernatant was centrifuged at 1593 ×g for

82

10 min (Beckman, JA10 rotor), decanted and passed through a 0.22μm filter (Millipore,

83

Streicup 500ml, 0.22 μm, Cat. no. SCGPU05RE). Virus particles were collected from the

84

filtrate by centrifugation in a Beckman JA20 rotor at 43,667 ×g for 6 hours, in autoclaved

85

30ml Nalgene PPCO tubes (cat. No. 3119-0030). The 6L were spun down by discarding the

86

supernatant from each 30ml tube (x 8 in a JA20) after a round of centrifugation, then adding

87

another 30ml of the extract to the tube. The individual pellets were resuspended in 3ml

88

successively: the first pellet was resuspended in 3ml TE buffer and the liquid was then

89

transferred to the next tube, the pellet resuspended properly, then transferred to the next tube

90

and so on until all pellets were resuspended. The pellets were treated with DNAseI (EN0521,

91

Fermentas) and RNAseA (EN0531, Fermentas) to a final concentration of 0.1 µg/ml at 37ºC

92

for 1 hour. The presence of bacterial DNA was checked by amplifying the 16S rRNA gene

93

(primers E9F and U1510R; 23, 24) as follows: 1µl of genomic DNA was mixed with 2.5µl of

94

each primer (10mM), 2.5 µl of 2 µM dNTPs, 2.5 µl of 10X DreamTaq buffer (ThermoFisher

95

Scientific, MA, USA), 1µl BSA 10 mg/ml, 0.125 µl DreamTaq polymerase (ThermoFisher

4

96

Scientific, MA, USA) and milliQ water to a total volume of 25 µl. PCR was conducted under

97

the following thermal regime: 95 °C for 5 min, 95°C for 30 sec, 52°C for 30 sec, 72°C for 85

98

sec (30 cycles) and 10 min at 72°C. The virus suspension was treated with Proteinase K

99

(Fermentas) at a final concentration of 1µg/ml at 55°C for 2 hours. 70 µl SDS (20%) was

100

added and incubated at 37 ºC for 1 hour. Nucleic acids were purified by performing two

101

rounds

102

chloroform:isoamyl alcohol (24:1) phase separation. DNA was precipitated by the addition of

103

one tenth volume sodium acetate (3M, pH 5.2) and two volumes 100% ethanol and left

104

overnight at 4ºC. Samples were centrifuged at 29,000×g for 10 min to pellet the DNA, which

105

was resuspended in 30ul of TE buffer. The DNA was further cleaned using the Qiagen Gel

106

Extraction kit (Qiaex II, cat. no. 20021). 10 ng of each sample was then used to perform

107

Phi29 amplification (GE Healthcare GenomiPhi HY DNA amplification kit CAT# 25-6600-

108

20) using the manufacturer's recommendations. Library preparation included a 10% phiX V3

109

spike as per manufacturer’s instructions (Preparation guide, Part #15031942, May 2012

110

revision) with the Illumina Nextera XT library prep kit/MiSeq Reagent kit V2. The amplified

111

DNA was sequenced (2×250bp reads, ~250bp average insert size) on the Illumina MiSeq

112

Sequencer platform located at the University of the Western Cape, Cape Town, South Africa.

113

These sequence data have been submitted to the DDBJ/EMBL/GenBank databases (sequence

114

read archive, SRA) under study accession SRP038018 (hypolith library) and SRP 035457

115

(open soil library).

116

Sequence data analysis

of

phenol:chloroform:isoamyl

alcohol

(25:24:1)

extraction

followed

by

117

Sequence reads were curated for quality control and adapter trimmed using CLC

118

Genomics version 6.0.1 (CLC, Denmark), using the default parameters. Unpaired reads were

119

aligned against each other using Bowtie under default parameters. De novo assembly for each

120

read dataset was performed with both CLC Genomics and DNASTAR Lasergene SeqMan 5

121

assembler suite using the default parameters. Reads and contigs were uploaded to the

122

MetaVir (25) version 2 server (http://metavir-meb.univ-bpclermont.fr/) and MG-RAST (26;

123

http://metagenomics.anl.gov/) for virus diversity estimations (data available from these

124

webservers). Taxonomic composition by MetaVir was computed from a BLAST comparison

125

with the Refseq complete viral genomes protein sequences database from NCBI (release of

126

2013-05-01) using BLASTp with a threshold of 10-5 on the e-value. Assembled reads were

127

searched for open reading frames and compared to the RefSeq complete viral database

128

(through the MetaVir pipeline) and MG-RAST, which include annotations using, for

129

functional and organism assignment: GenBank, IMG, KEGG, PATRIC, RefSeq, SEED,

130

SwissProt, trEMBL and eggnog. The subset of affiliated (i.e. predicted genes with a database

131

match) contigs generated by CLC Genomics were compared to the contigs generated by

132

LaserGene using BLASTx under standard parameters. For the assignment of functional

133

hierarchy, COG, KO and NOG databases were used. Guanine-cytosine (GC) content was

134

determined by importing .fasta files into BioEdit (27). The presence of tRNAs in annotated

135

contigs

136

http://lowelab.ucsc.edu/tRNAscan-SE/ (28). For prediction of phage lifestyle and host Gram

137

stain reaction, whole genome protein sequences of candidate phage genomes were submitted

138

to the online version of PHACTS (http://www.phantome.org/PHACTS/; 29). Aligned marker

139

genes showing sufficient homology (>150bp, MetaVir) against the contigs were recovered

140

and phylogenetic analysis was performed using MEGA5 (http://www.megasoftware.net/).

141

Rooted dendrograms were inferred using the maximum likelihood method with a bootstrap

142

test of a 1000 pseudo-replicates. Phylogenetic analysis for virophage sequences was

143

performed independently from MetaVir. Metavirome virophage amino acid sequences, as

144

well as 9 virophage MCP sequences obtained from the NCBI Genbank database were aligned

was

assessed

with

the

tRNAscanSE

6

software

accessible

through

145

with the online version of MAFFT version 7 (http://mafft.cbrc.jp/alignment/software/). Tree

146

construction was conducted as outlined in Zhou et al., 2013 (30).

147

Results

148

Viral diversity estimations

149

The presence of bacterial contamination was deemed negligible in both metavirome

150

libraries (using the 16S gene fragment), as no discernable bands of amplified products were

151

obtained. MiSeq reads ranged from 236-241 bp, with an overall higher G+C content within

152

reads obtained from the open soil library. Sequencing metadata, assembly metrics, BLASTp

153

searches are summarized in Table 1. BLASTx comparison of contig datasets from both

154

habitat libraries (generated by two separate assemblers) revealed that 99.41% (hypolith

155

library) and 99.5% (open soil library) of affiliated contigs were shared between the two

156

assembled read datasets. Contigs from CLC Genomics were used for the remainder of the

157

analysis. Aligned against each other, libraries contained 66.01% of reads that were unique to

158

each habitat, while 33.99% were shared (a read aligned at least once). In both read datasets,

159

bacteria were the most represented hits (80.7-94.5%). However, these estimations varied

160

depending on the metagenomic platform used and whether reads or contigs were submitted.

161

BLASTp searches from MetaVir using contigs produced significantly more virus-related hits

162

compared to MG-RAST. For example, 1.9% of the open soil contigs were predicted to be

163

viral in origin in MG-RAST, while MetaVir with the same dataset predicted 18.8 %. The

164

same was true for the hypolith library, where MG-RAST predicted 12.8% for viruses, while

165

MetaVir predicted 19.2%. Archaea, Eukaryota and “other” represented the smallest fraction

166

while viruses (particularly in hypolith) were second in terms of contig affiliations (Figure 1).

167

Total diversity (α-diversity) was computed by MG-RAST using normalized values, since

168

unequal distribution of reads between open soil and hypolith libraries were obtained. The

7

169

hypolith library was three-fold more diverse compared to the open soil library (344.5 vs.

170

1058.4 species). In contrast, the open soil showed higher taxonomic abundance (species

171

evenness, γ-diversity) compared to the hypolith (15663 vs. 11480; c.f. Table 2).

172

Proteobacteria and Firmicutes were the most abundant in both libraries, with viruses in

173

hypolith the third most abundant organisms. Rarefaction curves generated by MG-RAST

174

(Supplementary Figure 1) showed the hypolith library was sampled more comprehensively

175

compared to the open soil. Nineteen virus families were identified by MetaVir (Table 3), in

176

which prokaryotic viruses were the most abundant in both habitats (OS = 76.0%; HY =

177

82.3%). Identified phages were dominated by the order Caudovirales in the following

178

abundance ranking (identical for both habitats): Siphoviridae > Myoviridae > Podoviridae.

179

The next most highly represented virus families were Mimiviridae and Phycodnaviridae, both

180

more numerous in the open soil sample. Viral parasites of large dsDNA viruses, i.e.

181

virophages (31), were exclusively identified in the open soil habitat. Signatures from

182

Adenoviridae, Bicaudaviridae, Hytrosaviridae, Retroviridae and Rudiviridae were found in

183

low numbers in the hypolith habitat only. Both habitats contained 13.5-15.1% of sequences

184

identified as unclassified viruses.

185

Due to the lack of universal markers for viruses (such as the 16S rDNA marker used

186

for bacteria or the 18S for eukaryotes), markers targeting virus families/species were used

187

instead as an alternative to improve taxonomic affiliation of the annotated ORFs, from both

188

assembled reads (contigs) and reads alone. Sequences with significant homology to reference

189

markers are shown in Supplementary Table 1. The large terminase subunit (TerL) marker,

190

required for packaging initiation in members of the Caudovirales (32), was the most common

191

match in both habitats. This was consistent with the taxonomic affiliations of contigs in the

192

virus families shown in Table 3. Non-bacterial viruses (Paramecium bursaria chlorella virus

193

and Emiliana huxleyi virus, family: Phycodnaviridae and invertebrate viruses, family:

8

194

Ascoviridae), identified with the major capsid protein (MCP) and DNA polymerase family B

195

(PolB) gene markers, were found exclusively within the open soil community. For virophage-

196

related sequences, 5 candidate ORFs were submitted to a tBLASTn query and showed closest

197

similarity to the Zamilon (33) and Sputnik (31) virophages, both isolated from soil and

198

aquatic environments, respectively. We attempted to determine its phylogenetic relationship,

199

as amongst the very few currently recognised virophages, one has been isolated from Organic

200

Lake, Antarctica. Amongst these ORFs, a partial MCP sequence (344 amino acids long) was

201

identified and aligned with other known virophage MCP sequences (30). The MCP tree in

202

Supplementary Figure 2 shows an identical clustering pattern to that of (30) and indicates that

203

the virophage sequence from open soil (abbreviated Miers Valley Soil Virophage, MVSV)

204

belonged to the Sputnik virophage group (cluster 1), and was not more closely related to its

205

Antarctic counterpart. Its position within the tree suggests that MVSV shares a genetically

206

distant ancestor with Sputnik and Zamilon virophages. No reads with significant homology to

207

the psbA gene (a marine cyanophage photosynthesis-related gene) were identified. However,

208

other cyanophage sequences were detected within the G20 and PhoH phylogenies from the

209

hypolith dataset alone (Figure 2), present as highly divergent sequences at the root of

210

cyanophage sequence clusters. A summary of marker-identified phage species for each

211

marker is shown in Supplementary Table 2.

212

Functional composition of hypoliths and open soil

213

The hypolith dataset was highly uncharacterized (predicted proteins with no

214

significant homologs), with 81.3% compared to the open soil with 58.5%. 26 functional

215

categories were assigned to both libraries (Figure 3), each sub-divided into distinct sub-

216

systems. Apart from the phage category, functional abundance in all categories was greater in

217

open soil. Highest abundance variations between both biotopes included several metabolic

218

pathways involving phosphorus, nitrogen, aromatic compounds and iron metabolism. 9

219

Dormancy and sporulation- related functions were also notably higher in the open soil. A

220

similar trend was found for stress-related functions, including oxidative, osmotic and acid

221

stress. However, found almost exclusively in the hypolith library were desiccation stress-

222

related protein functions. Virus-specific functional components were retrieved manually from

223

the MetaVir server, counted and classified into several virus component categories, shown in

224

Supplementary Table 3. Both habitat samples contained genes encoding numerous virus

225

structural proteins (portal, tape measure, and capsid) and enzymes (terminases, DNA/RNA

226

polymerases, helicases and lysins), consistent with an abundance of tailed phage related

227

components.

228

subset of 26 were selected for further analysis (Supplementary Table 4) based on a

229

combination of criteria: size (≥ 10,000 bp), percentage of annotated ORFs within a contig

230

(11-100%) and predicted circularity of the putative genome. On average, the percentage of

231

homologous genes from public databases was 40.2 ± 21.2% in open soil and 31.9 ± 12.5% in

232

the hypolith sample. Average G+C content in open soil and hypolith contigs was 55.6 ± 7.3%

233

and 44.3 ± 3.2%, respectively. Phage genomes were submitted to PHACTS (29) for lifestyle

234

(temperate or lytic) and host Gram reaction prediction. As a general trend for both habitats,

235

putative temperate phages dominated (61.5%) while predicted host range was 88.5% Gram

236

negative. These predictions are supported by a recent study (34), which reported that Gram

237

negative Proteobacteria were the dominant phylum in hypolithic and open soil habitats within

238

the Dry Valleys.

239

contigs contained genes from two virus families infecting algae (Phycodnaviridae-like) and

240

amoeba (Mimiviridae-like), positioned between phage-related genes. The largest contig

241

(AntarOS_1, 177 571 bp) contained one gene from Acanthamoeba polyphaga mimivirus and

242

one from Paramecium bursaria chlorella virus A1 while the rest of the genes were phage

243

related. Several core genes (35) from the nucleocytoplasmic large DNA viruses (NCLDVs)

Due to the large number of assembled contigs, a

For the open soil habitat alone,

10

244

were identified in the open soil (also to a lesser extent in the hypolith library) contig dataset.

245

These included Topoisomerase II, RNA polymerase subunit 2, guanylyltransferase, RuvC,

246

dUTPase 2, thymidilate kinase, MutT/NUDIX motif and Ankyrin repeats. A hybrid gene

247

arrangement from different viruses was found in another contig, AntarOS_17 (Figure 4). This

248

24 870 bp contig was divided into 30 predicted ORFs, 21 of which showed significant

249

homology to virus genes in the RefSeq database (detailed BLAST result for individual ORFs

250

can be found in Supplementary Table 5). Of the 30 predicted ORFs, from gene 11 to gene 30

251

but excluding gene 18, 67% showed significant homology at the amino acid level with a

252

single microalgae-infecting virus species (unclassified Tetraselmis viridis virus S1,

253

NC_020869.1). Gene 4-8, 10 and 18 showed similarity to several phages infecting

254

Burkholderia, Rhodobacter and Azospirillum species. The genome size was too short to be

255

considered phycodnavirus-like (36), and possessed phage genes that were in usual functional

256

synteny towards phage head maturation (such as TerS, TerL and capsid protein, ORF 4-8). A

257

tBLASTn search using the protein sequence of the TerL gene against the NCBI nr database

258

showed highest similarity (34-35% identity, e-value 6x10-76) to various Streptococcus phi

259

phages including SsUD1, m46.1 and D12. To verify that this contig did not result from read

260

misassembly (i.e. chimeric), two sets of primers were designed to amplify fragments from the

261

overlapping region between ORF10 and ORF 11, which based on the contig annotations,

262

appeared to delineate two gene sets from different viruses. Amplicons of expected sizes were

263

obtained and sequenced bi-directionally by Sanger technology. The sequenced fragments

264

aligned with their respective regions, which indicated that this contig region was correctly

265

assembled.

266

Phage-host associations

267

As both habitats showed a high diversity of phage-related sequences, taxonomic

268

affiliation of the reads (marker gene - independent) were categorized according to host and 11

269

relative sequence abundances in both habitat samples (Figure 5). Phage sequences identified

270

most closely to host species spanning 5 bacterial phyla: Firmicutes (7 bacterial genera),

271

Proteobacteria (8 bacterial genera), Cyanobacteria (3 bacterial genera), Bacteroidetes

272

(Flavobacterium) and Actinobacteria (4 bacterial genera). By comparing the bacterial OTU

273

distributions in the same soil environments generated by (34), we attempted to correlate

274

presence/absence of bacterial OTUs based on the phage sequences obtained. Additionally, we

275

included in our comparison 454 sequencing-based soil metagenomic data from (37), which

276

surveyed moraine soil collected from the margins of a permanent melt water pond located at

277

Mars Oasis on Alexander Island, west of the Antarctic Peninsula. Based on identified phage

278

species, only Firmicutes were found in both soil habitats in this study but not in the

279

16S/TRFLP bacterial data from (34). This discrepancy between phage and bacterial data was

280

also observed for hypoliths in hot desert soils (Adriaenssens, personal communication). The

281

other major bacterial phyla (Proteobacteria, Cyanobacteria, Bacteroidetes and Actinobacteria)

282

were present in both the metavirome data and 16S/TRFLP sequence data. However, bacterial

283

genera indirectly identified by their phages from this study, were all found in the Pearce et al.

284

survey (37).

285

At the level of individual phages, Lactococcus and Mycobacterium phage

286

sequences were most common in the hypolith sample (> 10%), whereas in open soil the

287

largest fraction (> 6%) was composed of Bacillus, Pseudomonas and Mycobacterium phage

288

sequences. Few phage host species could be linked to the 16S/TRFLP data, but at the phylum

289

level, Proteobacteria and Actinobacteria were present in both datasets. Caulobacter and in

290

Flavobacterium were found both in this study (identified by their phage) and 16S/TRFLP

291

data. However, the top 10 virus fraction obtained by Pearce et al. was similar to this study,

292

where Mycobacterium phages ranked first (in the case of the hypolith sample), but also

12

293

included Pseudomonas, Enterobacteria, Flavobacterium and Synechococcus phages.

294 295

Discussion

296

Unlike aquatic ecosystems which have received considerable attention since the

297

advent of viral metagenomics (38), virus diversity of many soil habitats has not been

298

extensively characterized. In studies of Antarctic continental microbiology, only fresh water

299

lake metaviromes have been reported to date (14, 15). Recent phylogenetic studies of

300

Antarctic Dry Valley soils have shown that hypolithic communities represent the most

301

biodiverse and complex biological assemblages in this hyperarid soil biome (3). Diversity

302

estimations generated from our data (including viruses) suggest a similar pattern (three-fold

303

higher diversity compared to open soil). Furthermore, read libraries shared a mere 33%

304

similarity overlap, indicating hypolithic communities as distinct soil habitat from their

305

surroundings. This uniqueness, coupled with its higher microbial diversity, may entitle

306

hypolithic communities as biodiversity “micro hotspots” in this hyperarid desert. A large

307

fraction of ORFs (62.5-84.5%) from both soil habitat samples had no significant homologs in

308

public sequence databases, also observed in another published soil metavirome (39).

309

Rarefaction curves indicated that the open soil biotope has not been sampled sufficiently, and

310

therefore a greater sequencing depth would be advisable in future metagenome experiments

311

for this habitat.

312

Taxonomic/functional affiliation of predicted ORFs and gene marker analyses (e.g.

313

using TerL) were consistent with the conclusion that tailed bacteriophages were the primary

314

virus component in both soil habitats. Furthermore, an identical family-specific hierarchical

315

abundance was observed for both habitats (Siphoviridae >Myoviridae >Podoviridae) but with

316

a higher sequence diversity in hypolith communities. Compared to other virus groups, phage

13

317

sequences were predictably over-represented in both metaviromes, given that the biotic

318

component in these soils is dominated by prokaryotes (9). However, we note that our nucleic

319

acid extraction method would exclude RNA viruses (either single or double stranded) and we

320

therefore do not claim that our data reflect the complete viral diversity in these soil habitats.

321

As a sampling bias, viruses in a prophage state may constitute a large (and unsurveyed)

322

proportion of the dsDNA phage diversity, given that it has been reported (19) that many soil-

323

borne bacteria appear to contain prophages, including those from Antarctic soils. Conversely,

324

our data suggests that ~61% of phage assemblages are temperate.

325

Very few archaeal virus signatures were found in either soil habitat, consistent with

326

previous prokaryote diversity studies (9, 34). Unexpectedly, cyanophages were poorly

327

represented in the hypolith sample (in terms of sequence abundance and diversity). Given the

328

dominance of cyanobacteria in Type I hypoliths (10, 12), it was reasonably predicted that

329

cyanophage would represent a major clade. The apparent success of cyanobacteria as

330

dominant elements of the hypolithic community might possibly be linked to the low

331

abundance of associated viruses (where the phage infection of other bacterial groups such

332

Mycobacteria, Bacillus, Flavobacterium and Pseudomonads was higher and their host

333

populations under tighter predation control). However, this is in contradiction to the general

334

understanding that the most abundant phage groups in any given environment reflect the

335

abundance of microbial community members found in that environment (40). While it is thus

336

possible that cyanophages genuinely represent a minor component of phage diversity, we

337

suggest that this result is an artefact of the substantial under-representation of soil-associated

338

cyanophage genome sequences in public databases, further accentuated by the fact that most

339

characterized cyanophages are of aquatic origin (41). To our knowledge, only a small fraction

340

of cyanophages of soil origin have been described to date (42), which also reported a high

341

phylogenetic distance from “common” marine cyanophages, emphasizing the fact that little is

14

342

known about these cyanophages. In support of this, cyanophage communities in paddy field

343

soils have been shown to be different from those in freshwater, marine water and even paddy

344

floodwater, identifying unique G20 subclusters specific to soil derived cyanophages (42). In

345

the current study, marker gene analysis successfully identified several metavirome sequences

346

at the root of cyanophage clusters, suggesting that these represent novel phage phylotypes

347

with a high genetic distance from currently characterized cyanophages. The high abundance

348

of phages infecting certain bacterial genera such as Mycobacterium, Lactococcus, Bacillus

349

and Pseudomonas phages (~6-10% of all identified phage sequences) in both Antarctic desert

350

soil habitats has been reported by Pearce et al. (2012).

351

Sequences with close homology to large dsDNA eukaryotic virus families such as

352

Mimiviridae and Phycodnaviridae-like genomic elements were found as the second largest

353

virus component (0.88% - 4.33%) in both habitats (excluding the unclassifiable virus fraction

354

(13.5%-15.1%)). Mimivirus -related sequences were unexpected, as a 0.22µm filter size

355

should have excluded large virus particles (~ 0.7 micron; (43)), as well as repeated

356

centrifugation steps. Phycodnaviruses, at ~0.16 ± 0.06 microns (44), would be expected to be

357

recovered in the filtrate. However, detection of MCP components from a novel Sputnik-like

358

virophage, a parasite of large dsDNA viruses (31), provided further indirect evidence for the

359

presence of mimivirus-like populations in the open soil habitat. In addition, the identified

360

virophage sequence was more closely related to geographically distant isolates (France and

361

Tunisia) compared to the other virophage isolate from Organic Lake in Antarctica. A recent

362

study (37) showed sequences belonging to both host genera (Paramecium, Chlorella and

363

Acanthamoeba) and their associated viruses (Chlorovirus and Mimivirus) in moraine

364

Antarctic soil. La Scola et al. (45) first demonstrated the presence of Mimiviruses in soil

365

(previously only isolated from aquatic habitats). The present metavirome sequences,

366

combined with pyrosequencing data of metagenomic libraries from Pearce et al. (37), provide

15

367

additional evidence for the presence of Mimivirus-like genome elements in Antarctic soils.

368

Further sampling to isolate virophages from Antarctic soils would provide further

369

understanding into the ecology and function of these infectious agents, given that their

370

contributions into the regulation of viral populations are starting to be become apparent in

371

other habitats (33). The unusual gene configuration observed within contig AntarOS_17

372

(were phage and eukaryotic viruses were predicted) was confirmed by PCR on the original

373

DNA sample, therefore ruling out misassembly of the reads for this region. Most likely, this

374

was caused by misannotation of the predicted ORFs, caused by a lack of closer homologs in

375

databases. While read misassembly is still a possibility in other generated contigs, confidence

376

level in assembly accuracy was high as the Illumina sequencing control used in both runs was

377

phiX174, (a ~5000 bp ssDNA virus) which was re-assembled almost completely (99.7%) and

378

correctly annotated by the MetaVir pipeline.

379

representing less than 0.5% of sequence abundance (Table 1) included those infecting

380

infected Diptera, arthropods and other invertebrates and were mostly found in the open soil

381

habitat. As these hosts have been shown to occur on the Antarctic Peninsula (46, 47), this

382

may represent an additional pool of uncharacterized viruses within the Antarctic invertebrate

383

fauna.

384

correlation between phage genera from this study and their associated hosts identified in

385

other bacterial diversity studies was established (34, 37, 48, 49). As in previous hypolith/open

386

soil community diversity (α-diversity) comparisons (Makhalanyane et al., 2013), where

387

hypoliths showed a higher degree of diversity compared to open soil, the same was

388

demonstrated to be true for their associated viruses.

389

represents an initial broad survey of virus diversity in Antarctic hyperarid desert soils, and

390

has demonstrated these local virus assemblages are highly diverse and largely

391

uncharacterized. Due to a huge gap in terms of homologous sequences in present databases,

Virus

families

A positive

16

This study

392

the generation of additional metagenomic sequence data is not likely to yield usable

393

information. This emphasises the need for more “traditional” studies, performed in parallel on

394

identical sample sources. These include morphological data from microscopy, lytic induction

395

(e.g. mitomycin C) upon raw soil and Sanger sequencing of clones targeting specific virus

396

families. Unfortunately, a large fraction will most likely remain uncharacterizeable in vitro,

397

as the majority of their host (bacteria in particular) remain unculturable. Larger eukaryotic

398

viruses infecting algae, amoeba and invertebrates have not previously been characterized in

399

this environment, and our data demonstrate that these viruses represent an unknown virus

400

population that awaits characterization. Such data would further advance our understanding

401

of the trophic structure and function of communities inhabiting this cold, hyperarid desert

402

biome.

403

Acknowledgements

404

The authors gratefully acknowledge financial support from the National Research

405

Foundation (SANAP), the University of Waikato NZTABS program, Antarctica New

406

Zealand and the University of Pretoria Genomics Research Institute. The authors also wish to

407

thank

408

(CHPC),

409

Technology of South Africa. E.M Adriaenssens holds a Vice-Chancellor’s fellowship of the

410

University of Pretoria and O. Zablocki is supported by the NRF-DST doctoral Innovation

411

fund. Opinions expressed and conclusions arrived at, are those of the authors and are not

412

necessarily to be attributed to the NRF. The authors declare no conflict of interests.

413

References

414 415

the an

Centre initiative

for supported

High by

the

Performance Department

of

Computing Science

and

1. Oerlemans J, Fortuin JPF. 1990. Parameterization of the annual surface temperature and mass balance of Antarctica. Ann. Glaciol. 14: 78-84. 17

416 417

2. Bockheim JG, McLeod M. 2008. Soil distribution in the McMurdo dry valleys, Antarctica. Geoderma 144: 43-49.

418

3. Cary SC, McDonald IR, Barrett JE, Cowan DA. 2010. On the rocks: the

419

microbiology of Antarctic Dry Valley soils. Nature Rev. Microbiol. 8(2): 129-138.

420

4. Wierzchos J, de los Ríos A, Ascas, C. 2013. Microorganisms in desert rocks: the

421

edge of life on Earth. Int. Microbiol. 15(4): 172-182.

422

5. Wood SA, Rueckert A, Cowan DA, Cary SC. 2008. Sources of edaphic

423

cyanobacterial diversity in the Dry Valleys of Eastern Antarctica. ISME J. 2(3): 308-

424

320.

425

6. Tracy CR, Streten‐Joyce C, Dalton R, Nussear KE, Gibb KS, Christian KA.

426

2010. Microclimate and limits to photosynthesis in a diverse community of hypolithic

427

cyanobacteria in northern Australia. Appl. Environ. Microbiol. 12(3): 592-607.

428

7. Cowan DA, Sohm JA, Makhalanyane TP, Capone DG, Green TGA, Cary SC,

429

Tuffin IM. 2011. Hypolithic communities: important nitrogen sources in Antarctic

430

desert soils. Environ. Microbiol. Rep. 3(5): 581-586.

431 432

8. Smith MC, Bowman JP, Scott FJ, Line ME. 2000. Sublithic bacteria associated with Antarctic quartz stones. Antarct. Sci. 12(2):177-184.

433

9. Pointing SB, Chan Y, Lacap DC, Lau MC, Jurgens JA, Farrell RL. 2009. Highly

434

specialized microbial diversity in hyper-arid polar desert. Proc. Natl. Acad. Sci. U. S.

435

A 106(47):19964-19969.

436 437

10. Cowan DA, Khan N, Pointing S, Cary SC. 2010. Diverse hypolithic refuge communities in Antarctic Dry Valleys. Antarct. Sci. 22: 714–720.

438

11. Cowan DA, Pointing SB, Stevens MI, Cary SC, Stomeo F, Tuffin IM. 2011.

439

Distribution and abiotic influences on hypolithic microbial communities in an

440

Antarctic Dry Valley. Polar Biol. 34(2): 307-311.

18

441

12. Khan N, Tuffin M, Stafford W, Cary C, Lacap DC, Pointing SB, Cowan DA.

442

2011. Hypolithic microbial communities of quartz rocks from Miers Valley,

443

McMurdo Dry Valleys, Antarctica. Polar. Biol. 34(11): 1657-1668.

444

13. Laybourn-Parry J. 2009.No place too cold. Science 324(5934): 1521-1522.

445

14. Säwström C, Lisle J, Anesio AM, Priscu JC, Laybourn-Parry J. 2008

446

Bacteriophage in polar inland waters. Extremophiles 12(2):167-175.

447

15. López-Bueno A, Tamames J, Velázquez D, Moya A, Quesada A, Alcamí A. 2009.

448

High diversity of the viral community from an Antarctic lake. Science 326(5954):

449

858-861.

450

16. Yau S, Lauro FM, DeMaere MZ, Brown MV, Raftery MJ, Andrews-Pfannkoch

451

C, Lewis M, Hoffman JM, Gibson JA, Cavicchioli R. 2011. Virophage control of

452

antarctic algal host–virus dynamics. Proc. Natl. Acad. Sci. U. S. A 108 (15): 6163-

453

6168.

454

17. Kennedy S, Kuiken T, Jepson PD, Deaville R, Forsyth M, Barrett T, Wilson S.

455

2000. Mass die-off of Caspian seals caused by canine distemper virus. Emerg. Infect.

456

Dis. 6(6): 637.

457

18. Wallensten A, Munster VJ, Osterhaus ADME, Waldenström J, Bonnedahl J,

458

Broman T, Olsen B. 2006. Mounting evidence for the presence of influenza A virus

459

in the avifauna of the Antarctic region. Antarctic Science 18(3): 353.

460

19. Williamson KE, Radosevich M, Smith DW, Wommack KE. 2007. Incidence of

461

lysogeny within temperate and extreme soil environments. Environ. Microbiol. 9(10):

462

2563-2574.

463

20. Meiring TL, Tuffin IM, Cary C, Cowan DA. 2012. Genome sequence of temperate

464

bacteriophage Psymv2 from Antarctic Dry Valley soil isolate Psychrobacter sp. MV2.

465

Extremophiles 16(5): 715-726.

19

466

21. Swanson MM, Reavy B, Makarova KS, Cock PJ, Hopkins DW, Torrance L,

467

Taliansky M. 2012. Novel bacteriophages containing a genome of another

468

bacteriophage within their genomes. PloS One 7(7): e40683.

469

22. Adriaenssens EM, Van Zyl L, De Maayer P, Rubagotti E, Rybicki E, Tuffin M,

470

Cowan DA. 2014. Metagenomic analysis of the viral community in Namib Desert

471

hypoliths. Environ. Microbiol. DOI: 10.1111/1462-2920.12528

472

23. Hansen MC, Tolker‐Nielsen T, Givskov M, Molin S. 1998. Biased 16S rDNA PCR

473

amplification caused by interference from DNA flanking the template region. FEMS

474

Microbiol. Ecol. 26(2):141-149.

475 476

24. Reysenbach AL, Pace NR, Robb FT, Place AR. 1995. Archaea: a laboratory manual—thermophiles. Cold Spring Harb. Protoc. 16: 101-107.

477

25. Roux S, Faubladier M, Mahul A, Paulhe N, Bernard A, Debroas D, Enault F.

478

2011. Metavir: a web server dedicated to virome analysis. Bioinformatics 27(21):

479

3074-3075.

480

26. Meyer F, Paarmann D, D'Souza M, Olson R, Glass EM, Kubal M, Paczian T,

481

Rodrigez A, Stevens R, Wilke A, Wilkening J, Edwards RA. 2008. The

482

metagenomics RAST server–a public resource for the automatic phylogenetic and

483

functional analysis of metagenomes. BMC bioinformatics, 9(1): 386.

484 485

27. Hall TA. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic. Acids. Symp. Ser. 41:95-98.

486

28. Schattner P, Brooks AN, Lowe TM. 2005. The tRNAscan-SE, snoscan and snoGPS

487

web servers for the detection of tRNAs and snoRNAs. Nucleic Acids Res. 33: W686-

488

W689.

489 490

29. McNair K, Bailey BA, Edwards RA. 2012. PHACTS, a computational approach to classifying the lifestyle of phages. Bioinformatics 28(5): 614-618.

20

491 492

30. Zhou J, Zhang W, Yan S, Xiao J, Zhang Y, Li B, Pan Y, Wang Y. 2013. Diversity of virophages in metagenomic data sets. J. Virol. 87(8): 4225-4236.

493

31. La Scola B, Desnues C, Pagnier I, Robert C, Barrassi L, Fournous G, Raoult D.

494

2008. The virophage as a unique parasite of the giant mimivirus. Nature 455(7209):

495

100-104.

496 497

32. Black LW. 1995. DNA packaging and cutting by phage terminases: control in phage T4 by a synaptic mechanism. Bioessays 17(12): 1025-1030.

498

33. Gaia M, Benamar S, Boughalmi M, Pagnier I, Croce O, Colson P, Raoult D, La

499

Scola B. 2014. Zamilon, a novel virophage with Mimiviridae host specificity. PLOS

500

One 9(4): e94923.

501

34. Makhalanyane TP, Valverde A, Birkeland NK, Cary SC, Tuffin IM, Cowan DA.

502

2013. Evidence for successional development in Antarctic hypolithic bacterial

503

communities. ISME J. 7(11):2080-2090.

504 505 506 507

35. Lyer LM, Aravind L, Koonin EV. 2001. Common origin of four diverse families of large eukaryotic DNA viruses. J. Virol. 75(23): 11720-11734. 36. van Ettel JL, Graves MV, Müller DG, Boland W, Delaroque N. 2002. Phycodnaviridae–large DNA algal viruses. Arch. Virol. 147(8): 1479-1516.

508

37. Pearce DA, Newsham KK, Thorne MA, Calvo-Bado L., Krsek M, Laskaris P,

509

Hodson A, Wellington EM. 2012. Metagenomic analysis of a southern maritime

510

Antarctic soil. Front. Microbiol. 3.

511

38. Srinivasiah S, Bhavsar J, Thapar K, Liles M, Schoenfeld T, Wommack KE.

512

2008. Phages across the biosphere: contrasts of viruses in soil and aquatic

513

environments. Res. Microbiol. 159(5): 349-357.

514

39. Fierer N, Breitbart M, Nulton J,Salamon P, Loozupone C, Jones R, Robeson M,

515

Edwards RA, Felts B, Rayhawk S, Knight R, Rohwer F, Jackson RB. 2007.

21

516

Metagenomic and small-subunit rRNA analyses reveal the genetic diversity of

517

bacteria, fungi, and viruses in soil. Appl. Environ. Microbiol. 73(21):7059-7066.

518

40. Breitbart M, Rohwer F. 2005. Here a virus, there a virus, everywhere the same

519

virus? Trends Microbiol. 13(6): 278-284.

520

41. Hurst CJ. 2011. Studies in viral ecology, vol. 1. New Jersey, USA: Wiley-Blackwell.

521

42. Wang G, Asakawa S, Kimura M. 2011. Spatial and temporal changes of

522

cyanophage communities in paddy field soils as revealed by the capsid assembly

523

protein gene g20. FEMS Microbiol. Ecol. 76(2):352-9.

524

43. Claverie JM, Grzela R, Lartigue A, Bernadac A, Nitsche S, Vacelet J, Abergel C.

525

2009. Mimivirus and Mimiviridae: giant viruses with an increasing number of

526

potential hosts, including corals and sponges. J. Invertebr. Patho. 101(3): 172-180.

527

44. Dunigan DD, Fitzgerald LA, Van Etten JL. 2006. Phycodnaviruses: a peek at

528

genetic diversity. Virus. Res. 117(1): 119-132.

529

45. La Scola B, Campocasso A, N’Dong R, Fournous G, Barrassi L, Flaudrops C,

530

Raoult D. 2010. Tentative characterization of new environmental giant viruses by

531

MALDI-TOF mass spectrometry. Intervirology 53(5): 344-353.

532

46. Convey P, Gibson JA, Hillenbrand CD, Hodgson DA, Pugh PJ, Smellie JL,

533

Stevens MI. 2008. Antarctic terrestrial life–challenging the history of the frozen

534

continent? Biol. Rev. Camb. Philos. Soc. 83(2): 103-117.

535 536

47. Treonis AM, Wall DH, Virginia RA. 1999. Invertebrate biodiversity in Antarctic Dry Valley soils and sediments. Ecosystems 2(6): 482-492.

537

48. Friedmann EI. 1993. Antarctic Microbiology. Wiley-Liss. New York, p 634.

538

49. Smith JJ, Tow LA, Stafford W, Cary C, Cowan DA. 2006. Bacterial diversity in

539

three different Antarctic cold desert mineral soils. Microb. Ecol. 51(4): 413-421.

540

22

541 542

543 544 545 546 547 548

FIGURES

549 550 551 552

Figure 1: Taxonomic affiliation of the predicted ORFs in both habitat libraries, assigned by MG-RAST.

553 554

23

555 556

Figure 2: Selected cyanophage sub-tree phylogenies from G20 (A) and PhoH (B) marker genes based on

557

protein alignments retrieved from the MetaVir 2.0 analysis server (metagenomic read selection and tree

558

construction methods are outlined in Roux et al. (2011).

559 560 561

24

Functional Abundance Virulence, Disease and Defense Sulfur Metabolism Stress Response Secondary Metabolism Respiration Regulation and Cell signaling RNA Metabolism Protein Metabolism Potassium metabolism

Functional categories

Photosynthesis Phosphorus Metabolism Phages, Prophages, Transposable elements, Plasmids Nucleosides and Nucleotides Nitrogen Metabolism Motility and Chemotaxis Metabolism of Aromatic Compounds Membrane Transport Iron acquisition and metabolism Fatty Acids, Lipids, and Isoprenoids Dormancy and Sporulation DNA Metabolism Cofactors, Vitamins, Prosthetic Groups, Pigments Clustering-based subsystems Cell Wall and Capsule Cell Division and Cell Cycle Carbohydrates Amino Acids and Derivatives 0

0.2

0.4

0.6

0.8

1

Normalized abundance ratio

Open soil

Hypolith

562 563

Figure 3: Functional assignment of predicted ORFs compared in both soil habitats. (Functional annotation

564

performed by MG-RAST using a 60% similarity cut-off).

565 25

1.2

566 567

Figure 4: Predicted genome organization of phage AntarOS_17, assembled from open soil reads. Arrows represent

568

open reading frames (ORFs) and their orientation. Purple indicates ORFs without taxonomic affiliations or ORFs

569

that were predicted without a known function (ORFan). Red indicates ORFs which were more closely related to

570

bacteriophage genes and green indicates ORFs more closely related to Tetraselmis viridis virus S1, a phycodnavirus.

571

Numbers below the continuous black line (whole contig length) represent the nucleotide number at a given point.

572

Numbers on top of each ORF arrow denotes gene number. HP denotes a hypothetical protein.

573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 26

589 590

Figure 5: Relative abundance of identified phage isolate sequences based on predicted ORFs identified by MetaVir

591

(e-value cut-off: 10-5) against the RefSeq database. Phage isolates were clustered and counted per their bacterial

592

hosts in their respective habitats (hypolith: dark grey); open soil: light grey). Black lines below a subset of phage-

593

infecting bacterial species indicate the bacterial host phylum.

594 595 596 597 598 599 600 601 602 603 27

TABLES

604 605

Table 1: NGS metadata, including assembly, annotation, and diversity statistics produced by CLC

606

Genomics and MG-RAST.

607 608

Pre-QC nbr. of reads Post-QC nbr. of reads Average read length (post-QC) Mean G+C % Nbr. / % of reads not assembled into contigs Nbr. of contigs generated Minimum length (nt) Maximum length (nt) N25/N50/N75 (nt) % unknown proteins % annotated proteins α-diversity (total, not limited to viruses)

Open Soil library 1,622,598 1,597,524 236.95 52±12 111,385 (6.97)

Hypolith library 3,771,948 3,729,606 241.25 47±8 274,272 (7.35)

22,237 200 177,571 865/446/267 58.5 39.2 344.508

53,695 200 50,044 945/553/354 81.3 14.5 1058.398

609 610

Table 2: Relative abundance of the most represented phyla in both biotopes identified by MG-RAST.

611

OS: open soil; HY: hypolith. Top Phyla Proteobacteria Firmicutes Viruses Actinobacteria Bacteroidetes Cyanobacteria Choroflexi Verrucomicrobia Chordata Unclassified Eukaryotes Planctomycetes

Relative abundance OS HY 9493 3492 2722 2647 302 1469 521 1039 1133 876 173 443 60 121 165 110 91 109 85 106 103

87

% from total abundance OS HY 60.6 30.4 17.4 23 1.9 12.8 3.3 9 7.2 7.6 1.1 3.9 0.4 1.1 1.1 1 0.6 0.9 0.5 0.9 0.7

0.8

612 28

613

Table 3: Taxonomic abundance of identified viral ORFs (BLASTp with a threshold of 10-5 on e-value)

614

identified by MetaVir in both Antarctic biotopes. Virus order Caudovirales

615

Virus family

Host

Myoviridae Podoviridae Siphoviridae Herpesviridae Adenoviridae Ascoviridae Asfarviridae

Hypolith (%/sequence hits) 20.9 (1305) 9.04 (565) 52.6 (3287) 0.11 (7) 0.02 (1) 0.03 (2) 0.03 (2)

Bacteria, Archaea Bacteria Bacteria, Archaea Herpesvirales Vertebrates Families not Vertebrates assigned into an Invertebrates order1 Swine, arthropod borne Baculoviridae Invertebrates 0.08 (5) Bicaudaviridae Archaea 0.05 (3) Hytrosaviridae Diptera (flies) 0.02 (1) Inoviridae Bacteria 0.16 (10) Iridoviridae Amphibians, fishes, 0.11 (7) invertebrates Microviridae Bacteria 0.30 (19) Mimiviridae Amoeba 0.88 (55) Phycodnaviridae Algae 1.74 (109) Polydnaviridae Parasitoid wasps 0.02 (1) Poxviridae Humans, 0.06 (4) arthropods, vertebrates Retroviridae Vertebrates 0.02 (1) Rudiviridae Thermophilic 0.03 (2) Archaea Viruses not Unclassified viruses N/A 13.5 (844) assigned into Sputnik virophage Mimivirus-infected 0.00 families1 amoeba 1 Virus families not yet assigned into an order as per the ICTV 2012 release.

29

Open soil (%/sequence hits) 26.1 (415) 11.0 (175) 38.3 (610) 0.13 (2) 0.00 (0) 0.37 (6) 0.06 (1) 0.25 (4) 0.00 (0) 0.00(0) 0.06 (1) 0.44 (7) 0.50 (8) 2.32 (37) 4.33 (69) 0.00 (0) 0.56 (9) 0.00 (0) 0.00 (0) 15.1 (241) 0.37 (6)