Diversity of Virophages in Metagenomic Data Sets

Abstract

Virophages, e.g., Sputnik, Mavirus, and Organic Lake virophage (OLV), are unusual parasites of giant double-stranded DNA (dsDNA) viruses, yet little is known about their diversity. Here, we describe the global distribution, abundance, and genetic diversity of virophages based on analyzing and mapping comprehensive metagenomic databases. The results reveal a distinct abundance and worldwide distribution of virophages, involving almost all geographical zones and a variety of unique environments. These environments ranged from deep ocean to inland, iced to hydrothermal lakes, and human gut- to animal-associated habitats. Four complete virophage genomic sequences (Yellowstone Lake virophages [YSLVs]) were obtained, as was one nearly complete sequence (Ace Lake Mavirus [ALM]). The genomes obtained were 27,849 bp long with 26 predicted open reading frames (ORFs) (YSLV1), 23,184 bp with 21 ORFs (YSLV2), 27,050 bp with 23 ORFs (YSLV3), 28,306 bp with 34 ORFs (YSLV4), and 17,767 bp with 22 ORFs (ALM). The homologous counterparts of five genes, including putative FtsK-HerA family DNA packaging ATPase and genes encoding DNA helicase/primase, cysteine protease, major capsid protein (MCP), and minor capsid protein (mCP), were present in all virophages studied thus far. They also shared a conserved gene cluster comprising the two core genes of MCP and mCP. Comparative genomic and phylogenetic analyses showed that YSLVs, having a closer relationship to each other than to the other virophages, were more closely related to OLV than to Sputnik but distantly related to Mavirus and ALM. These findings indicate that virophages appear to be widespread and genetically diverse, with at least 3 major lineages.

INTRODUCTION

Virophages, a group of circular double-stranded DNA (dsDNA) viruses, are icosahedral in shape and approximately 50 to 100 nm in size (1–4). Virophages have three unique features (2). First, the nuclear phase is absent during the infection cycle of virophages. Second, the replication of virophages takes place in a viral factory of the giant host DNA viruses. Third, they depend on enzymes from host viruses instead of host cells. Accordingly, virophages are considered to be parasites of giant DNA viruses, e.g., mimiviruses and phycodnaviruses (1–3). Giant DNA viruses possess huge genome sizes (up to ≈1,259 kb), some of which are even larger than those of certain bacteria (5–7). The infection and propagation of virophages lead to a significant decrease in host virus particles and, consequently, an increase in host cell survival (1–3). Additionally, exchanges of genes may occur between virophages and giant DNA viruses (1–3, 8, 9). Therefore, virophages are potential mediators of lateral gene transfer between large DNA viruses (8, 9).

Thus far, four virophages have been identified in distinct locations (Table 1). The first reported virophage, Sputnik, was isolated from an Acanthamoeba species infected with the large mamavirus in a water-cooling tower in Paris, France (2). The second virophage, Mavirus, was observed in a marine phagotrophic flagellate (Cafeteria roenbergensis) in the presence of the host virus, Cafeteria roenbergensis virus, originating from the coastal waters of Texas (1, 10). The third virophage, Organic Lake virophage (OLV), discovered in a hypersaline meromictic lake in Antarctica, is thought to parasitize large DNA viruses infecting microalgae (3, 11). At the time of this report, a fourth virophage, Sputnik 2, together with its host virus, Lentille, has been detected in the contact lens solution of a patient with keratitis in France (12). The fact that virophages exist in a wide range of virus and eukaryotic hosts, as well as in a variety of unique habitats, implies the possibility that they are more widely distributed and diverse than previously thought.

Table 1.

Features of virophages

Virophage	Location	Host		Genome
		Virus	Eukaryote	Size (bp)	No. of ORFs	C+G content (%)
Sputnik	A cooling tower in Paris, France	Acanthamoeba polyphaga mimivirus	A. polyphaga	18,343	21	27.0
Mavirus	Coastal waters of Texas	Cafeteria roenbergensis virus	Marine phagotrophic flagellate (C. roenbergensis)	19,063	20	30.3
OLV	Organic Lake, a hypersaline meromictic lake in Antarctica	Large DNA viruses	Prasinophytes (phototrophic algae)	26,421	26	39.1
Sputnik 2	Contact lens fluid of a patient with keratitis, France	Lentille virus	A. polyphaga	18,338	20	28.5
YSLV1	Yellowstone Lake	Phycodna- or mimiviruses?	Microalgae?	27,849	26	33.4
YSLV2	Yellowstone Lake	Phycodna- or mimiviruses?	Microalgae?	23,184	21	33.6
YSLV3	Yellowstone Lake	Phycodna- or mimiviruses?	Microalgae?	27,050	23	34.9
YSLV4	Yellowstone Lake	Phycodna- or mimiviruses?	Microalgae?	28,306	34	37.2
ALM	Ace Lake in Antarctica	mimiviruses?	Phagotrophic protozoan?	17,767	22	26.7

To obtain greater insight into the unusual diversity of the global distribution and abundance of virophages, in this study, metagenomic databases on the Community cyberinfrastructure for Advanced Microbial Ecology Research and Analysis (CAMERA) 2.0 Portal (https://portal.camera.calit2.net/) (13) were analyzed comprehensively. Four complete genomic sequences of virophages and one nearly complete sequence were assembled based on the metagenomic DNA sequences of Yellowstone Lake, Wyoming, and Ace Lake, Antarctica. Comparative genomics and phylogenetic analyses were performed in order to better understand the genomic sequence features, phylogeny, and evolution of virophages.

MATERIALS AND METHODS

Analysis of metagenomic databases.

The gene sequences of the three known virophages, Sputnik, Mavirus, and OLV (Sputnik 2 was excluded in the analysis since it was a new strain of Sputnik), were downloaded from the NCBI genome database and blasted against the NCBI nr database. The genomic sequence of another Sputnik, strain 3, was also available in GenBank; however, because Sputnik 2 and Sputnik 3 actually have the same sequence, Sputnik 3 was also not included in the analysis. Genes showing blastp hits to virophages only or no hits (E-value<10⁻⁵) were considered virophage-specific marker genes and were used to evaluate the global distribution and abundance of virophages. The genes were searched (tblastx, E-value<10⁻⁵) against databases of all metagenomic pyrosequencing reads and all Sanger reads on the CAMERA 2.0 Portal. The screened virophage-related sequences were further confirmed based on a blast similarity search against the NCBI nr databases. Mapping of the global distribution pattern of virophages was visualized through MapInfo Professional (version 11.0; Pitney Bowes Software, Inc.). The abundance of virophages is presented as the ratio of the number of virophage-like sequences in a given metagenomic data set and the total number of sequences in that respective data set, normalized to 1,000,000.

Analysis of virophage conserved genes.

All gene sequences of virophages Sputnik, Mavirus, and OLV were compared to the NCBI nr database using both blastp and PSI-BLAST searches (14, 15). Homologous genes shared among these three virophages were considered to be conserved. Their sequence similarities were also proofed based on multiple sequence alignment using MUSCLE (16) on Geneious Pro (version 5.5.7; Biomatters Ltd.).

Assembling of genomic sequences of new virophages.

Major capsid protein (MCP), the homolog of MV18 (Mavirus), V20 (Sputnik), and OLV09 (OLV), was searched (tblastx, E-value<10⁻⁵) against all metagenomic pyrosequencing read databases and all Sanger read databases on the CAMERA 2.0 Portal. Sequences significantly similar to these three MCPs were screened, downloaded, and treated as virophage MCP-related sequences. Subsequently, they were assembled to obtain MCP-related contigs. Each contig served as a reference sequence to which all reads from the corresponding metagenomic database were assembled. Once an extended sequence with a relatively longer size and higher coverage was obtained after assembly, it was used as the next reference to assemble all reads from metagenomic databases. This procedure was repeated until the assembled sequence stopped extending. If there was a repeat region of approximately 100 bp at both ends of the sequence obtained, it was eventually self-assembled to a circular DNA sequence. All sequence assemblies were performed using Geneious Pro. The sequence assembly parameters used in this study were a minimum overlap of 25 bp with >90% sequence identity, as well as 50% maximum mismatches per read.

Prediction and annotation of ORFs.

The prediction and annotation of virophage open reading frames (ORFs) followed the procedures described in the literature (17, 18). Each predicted ORF encompassed a start codon of ATG, minimum size of 135 bp, standard genetic code, and a stop codon. The blastp, tblastx, and PSI-BLAST programs were used for sequence similarity comparisons of the predicted ORFs to NCBI nr databases (14, 15). A local database that contained the translated protein sequences of all predicted ORFs in Sputnik, Mavirus, and OLV, as well as the five new virophages described in this study, was also included in the blast search. ORFs were searched for characteristic sequence signatures using the InterProScan program (19).

Phylogenetic analysis.

Amino acid sequences were aligned using MUSCLE (16), and the phylogenetic trees were reconstructed by using PhyML (version 3.0, Méthodes et Algorithmes pour la Bioinformatique, LIRMM, CNRS—Université de Montpellier; http://www.atgc-montpellier.fr/phyml/) (20).

Nucleotide sequence accession numbers.

The genomic sequences of the four Yellowstone Lake virophages (YSLVs) and Ace Lake Mavirus (ALM) have been deposited in GenBank under the accession numbers KC556924 (YSLV1), KC556925 (YSLV2), KC556926 (YSLV3), KC556922 (YSLV4), and KC556923 (ALM).

RESULTS AND DISCUSSION

Diversity of global distribution and abundance of virophages.

The blast similarity search (E-value<10⁻⁵) indicated that a total of 44 ORFs turned out to be virophage-specific marker genes, comprising 16 ORFs of Sputnik, 13 of Mavirus, and 15 of OLV (Table 2). These genes were used as query sequences and searched against all metagenomic data deposited in the CAMERA database. The CAMERA database is a web-based analysis portal that allows for depositing, locating, analyzing, visualizing, and sharing microbial data obtained from various environments, such as marine, soil, freshwater, wastewater, hot springs, animal hosts, and other habitats (13). Therefore, the general tendency of the global distribution and abundance of virophages can be predicted according to the virophage-related sequence information of blast hits provided by the CAMERA 2.0 Portal. The search found 1,766 pyrosequencing reads and 204 Sanger reads related to Sputnik, 203 pyrosequencing reads and 253 Sanger reads akin to Mavirus, and more than 50,000 pyrosequencing reads and Sanger reads similar to OLV (see Table S1 in the supplemental material). The redundant reads were incorporated and removed. Finally, 23,599 virophage-related sequences were obtained. Among them, 148 were Mavirus hits, 812 were Sputnik hits, and 22,639 were OLV hits, accounting for 95% of the total sequences associated with virophages (23,599). It appeared that OLV and its relatives were more abundant than Sputnik and Mavirus virophages in the environments.

Table 2.

Virophage-specific genes

Virophage	Genes
Sputnik	V01, V02, V03, V04, V05, V07, V08, V09, V14, V15, V16, V17, V18, V19, V20, V21
Mavirus	MV04, MV05, MV07, MV08, MV09, MV10, MV11, MV12, MV14, MV15, MV16, MV17, MV18
OLV	OLV01, OLV02, OLV03, OLV04, OLV05, OLV06, OLV07, OLV08, OLV09, OLV10, OLV11, OLV15, OLV21, OLV24, OLV26

As depicted in Figure 1, virophages were distributed widely throughout the world, including almost all geographical zones. The habitats of virophages were also localized in a variety of environments, ranging from the deep ocean to inland (Fig. 2). The abundance of virophages tended to increase from the ocean to land environments, was the highest in freshwater habitats, and was relatively greater in ocean sediment than in deep seawater (Fig. 2A). As for vertical distribution, in general, virophage abundance decreased with the increase in ocean depth (Fig. 2B). The epipelagic zone seemed to be enriched with virophages. This was probably because this illuminated zone at the surface of the sea is colonized by the most living organisms in the sea. Interestingly, although there is a large difference between the conditions of the abyssopelagic and the mesopelagic zones, it seemed that the numbers of virophage-related sequences observed in these two zones were quite similar (Fig. 2B). Whether real virophage enrichment was present in the abyssopelagic zone or whether it was a result of the virophage-infected host viruses and/or host cells settling to the deep sea remains to be studied further. In terms of geographical zones, the frigid zones turned out to have the greatest abundance of virophages, followed by the tropical zones (Fig. 2C). Obvious limitations and biases of the data deposited in CAMERA exist, and caution should be taken during attempts to interpret the global distribution and abundance of virophages. However, these findings open a new window into further exploration and survey of the diversity of unique virophages worldwide.

Fig 1.

Geographic distribution and corresponding abundance of virophages. Colored dots indicate distinct abundances of virophages in metagenomic data sets obtained from a specific area of latitude and longitude (see Table S1 in the supplemental material). Abundance was normalized to 1,000,000.

Fig 2.

Abundance of virophages in different environments (A), ocean depths (B), and latitudes (C). Abundance was normalized to 1,000,000.

In addition, unexpectedly, a small number of virophage-related sequences was detected in nonaquatic environments, e.g., 65 sequences from the human gut, 11 from animal-associated habitats, 7 from soils, 4 from glacier metagenomes, and 1 from air in the East Coast of Singapore. Thus far, little is known with regard to such unusual diversity (21). Taken together, comparative analyses of metagenomic databases revealed the global distribution and distinct abundance of virophage-related sequences, which suggested that virophages are common entities on Earth. Large-scale sampling and analyses are necessary to obtain a complete picture of the diversity of virophages.

Four complete genomes of Yellowstone Lake virophages and one nearly complete genome of Ace Lake Mavirus.

Major capsid protein is generally considered to be a conserved protein among viruses, and it is widely used to reconstruct phylogenetic trees. It was also conserved in virophages, based on blast sequence similarity searches and sequence alignment. In our study, four complete virophage genomes and one nearly complete virophage genome were obtained from two metagenomic databases named Yellowstone Lake: Genetic and Gene Diversity in a Freshwater Lake and Antarctica Aquatic Microbial Metagenome, which were downloaded from the CAMERA 2.0 Portal. These virophages were tentatively named YSLV1, YSLV2, YSLV3, YSLV4, and ALM. Detailed results of the metagenome assembly, i.e., genome coverage, the number of reads recruited to each genome, and the size of the data sets from which the metagenomes originated, are shown in Table 3; see also Figures S1 and S2 in the supplemental material.

Table 3.

Data on metagenomic assemblies of the five new virophages

Name	No. of reads recruited to each genome	No. of identical sites	Pairwise identity (%)	Genome coverage			Size of dataset (Gb)
				Mean	Minimum	Maximum
YSLV1	5,544	22,271	98.0	67.9	16	127	11.1
YSLV2	834	21,453	97.7	13.1	3	27	11.1
YSLV3	1,098	25,529	98.2	15.1	3	35	11.1
YSLV4	1,119	25,732	97.2	14.5	4	32	11.1
ALM	494	13,654	95.4	14.4	4	26	32.4

They were all dsDNA viruses, with G+C contents of 33.4% (YSLV1), 33.6% (YSLV2), 34.9% (YSLV3), 37.2% (YSLV4), and 26.7% (ALM) (Table 1). Their genomes were 27,849 bp in length with 26 predicted ORFs (YSLV1), 23,184 bp with 21 predicted ORFs (YSLV2), 27,050 bp with 23 predicted ORFs (YSLV3), 28,306 bp with 34 predicted ORFs (YSLV4), and 17,767 bp with 22 predicted ORFs (ALM) (Table 1 and Fig. 3 and 4). The YSLVs and OLV were generally alike in genome size, number of ORFs, and G+C content (Table 1).

Fig 3.

Circular maps of the complete genomes of Yellowstone Lake virophages. Homologous genes are indicated in the same color, the five conserved genes are labeled with red asterisks, and the inner circles represent G+C content plots. The dashed-line boxes represent the conserved gene cluster in all eight virophages, the dotted-line boxes represent the gene cluster shared by YSLVs 2, 3, and 4 and OLV, and the dash-dot-dot–line boxes represent the gene cluster present in YSLVs 3 and 4.

Fig 4.

Linear genomic map of ALM and Mavirus. Homologous genes are shown in the same color, while syntenic regions are presented in green, light blue, and orange. The five virophage conserved genes are labeled with red asterisks, and the conserved gene cluster is marked with dashed-line boxes.

Among 126 predicted ORFs from these five new virophages, 59 showed significant similarity to 33 of 67 ORFs of three known virophages, 11 showed similarity to the nucleocytoplasmic large DNA viruses (NCLDs) of eukaryotes (including phycodnaviruses, Marseilleviruses, and mimiviruses), and 3 showed similarity to sequences of unicellular eukaryotic organisms (marine choanoflagellate Monosiga brevicollis and ciliated protozoan Tetrahymena thermophila); 67 ORFs had no sequence hits to current NCBI databases (Table 4). Given that the virus and eukaryotic hosts of the virophages obtained in this study may be the NCLDs and the protists mentioned above (or their associated relatives), it is conceivable that horizontal gene transfer and/or gene recombination occurred between ancestor virophages and their viruses, as well as cellular hosts. Such gene replacement traces have been observed in virophages (Sputnik, Mavirus, and OLV) and their hosts (1–3). In addition, significant sequence similarity (E-value<10⁻⁵) was not detected between virophages and any viruses infecting multicellular organisms, which suggested that virophages diverged early and subsequently underwent a strict and unique evolution with their viruses and unicellular eukaryotic hosts.

Table 4.

ORFs and their homologs predicted in YSLVs and ALM

Virophage, ORF	Position		Lengtha		Best blastp hit in GenBank nr database and/or virus data set						NCBI conserved domain (identifier, E-value, alignment length in aa, alignment position [start–end])	Interproscan matches (identifier, E-value[s], alignment position[s] [start–end])
	Start	End	nt	aa	ORF, protein encoded, or mass	Species	Accession no.	E-value	% aa identity	Alignment length in aa (position start–end)
YSLV1
1	1	771	771	256	Hypothetical protein OLV4	Organic Lake virophage	ADX05765	2.01E−79	51	246 (1–246)	AAA domain (pfam13401, 4.46E−04, 124, 57–158)	P loop-containing nucleoside triphosphate hydrolases (SSF52540, 2.0E−6, 49–226)
2	968	768	201	66
3	1387	1049	339	112
4	1475	3775	2,301	766	Putative DNA primase/polymerase	Organic Lake virophage	ADX05784	7.79E−26	23.9	626 (50–675)		P loop-containing nucleoside triphosphate hydrolases (SSF52540, 6.5E−13, 486–663)
5	3959	3825	135	44								SP (1–17), TM (4–22)
6	4037	4429	393	130
7	4563	5129	567	188	VP11	Micromonas pusilla reovirus	YP_654554	5.95E−15	24	146 (4–149)		TM (39–59 69-91)
8	5275	8955	3,681	1,226								ATHOOK (PR00929, 6.8E−6, 6.8E−6, 6.8E−6, 345–355, 380–391, 417–427)
9	9051	9728	678	225	Hypothetical protein MV06	Mavirus	YP_004300284	0.004	34	84 (48–131)	N-terminal catalytic domain of GIY-YIG intron endonuclease I-TevI, I-BmoI, I-BanI, I-BthII, and similar proteins (cd10437, 3.38E−03, 90, 48–129)	Alpha/beta-hydrolases (SSF53474, 1.1E−8, 159–273)
10	9816	10742	927	308	Hypothetical protein OLV11	Organic Lake virophage	ADX05772	2.83E−56	37.3	299 (7–305)
11	11999	10743	1,257	418
12	12201	13709	1,509	502
13	13740	14087	348	115
14	14199	14354	156	51
15	14586	14335	252	83
16	15090	14758	333	110
17	15260	15084	177	58
18	15567	15905	339	112
19	15974	16462	489	162
20	16755	16588	168	55								SP (1–25)
21	16765	16935	171	56
22	20302	16940	3,363	1120	Unnamed protein product	Paenibacillus sp. JDR-2	YP_003010191	4.00E−56	25	834 (114–947)
					Hypothetical protein	Cyanophage NATL2A-133		8.00E−34	31	355 (380–734)
23	20395	20967	573	190	Hypothetical protein OLV7	Organic Lake virophage	ADX05768	8.35E−21	31.6	168 (22–189)
24	20939	21115	177	58								SP (1–31), TM (5–25)
25	22991	21120	1,872	623	Major capsid protein	Organic Lake virophage	ADX05770	1.00E−32	27	579 (1–579)
26	25733	23133	2,601	866	Putative minor capsid protein	Organic Lake virophage	ADX05769	4.00E−15	29	205 (643–847)		Putative isomerase YbhE (SSF101908, 1.6E−10, 119–433)
27	27321	25888	1,434	477	Putative minor capsid protein	Organic Lake virophage	ADX05769	2.00E−15	34	125 (332–456)
28	27741	27427	315	104	Tlr 6Fp protein	Tetrahymena thermophila	AF451864_6	5.28E−10	32.5	80 (25–104)
					Hypothetical protein MAR_433	Marseillevirus	YP_003407157	7.00E−13	35	78 (9–86)
YSLV2
1	1	765	765	254	Hypothetical protein OLV4	Organic Lake virophage	ADX05765	1.30E−48	40.8	242 (3–244)
2	807	1322	516	171	Hypothetical protein PRSM4_062	Prochlorococcus phage P-RSM4	YP_004323191	1.85E−08	40	70 (12–81)
3	1570	1319	252	83
4	1691	2725	1,035	344	Hypothetical protein OLV11	Organic Lake virophage	ADX05772	2.32E−40	33.9	297 (19–315)
5	2839	4758	1,920	639	Putative ATP-dependent RNA helicase	Acanthamoeba polyphaga mimivirus	YP_003987051	2.63E−32	29.8	444 (179–622)	DEAD-like helicases superfamily (smart00487, 4.48E−10, 177, 174–350)	Helicase_C (PF00271, 2.4E−6, 551–606)
											SNF2 family N-terminal domain (pfam00176, 1.19E−09, 238, 182–419)	DEAD-like helicases superfamily (SM00487, 0.0013, 175–378)
											Helicase superfamily C-terminal domain (cd00079, 1.52E−04, 145, 461–605)	P loop-containing nucleoside triphosphate hydrolases (SSF52540, 8.7E−23, 2.3E−20, 137–378, 385–636)
6	4992	7772	2,781	926	C terminus: hypothetical protein 162275902	Organic Lake phycodnavirus 2	ADX06405	6.07E−07	50	66 (834–899)	Methyltransferase domain (pfam13659, 2.52E−07, 111, 151–261)	S-Adenosyl-l-methionine-dependent methyltransferases (SSF53335, 3.3E−10, 156–307)
7	7862	8086	225	74								TM (34–52)
8	8538	8164	375	124
9	9182	8628	555	184
10	9656	12484	2,829	942	Helicase	Acanthamoeba castellanii mamavirus	AEQ60154	4.43E−44	30.5	465 (295–759)	Origin of replication binding protein (pfam02399, 7.47E−08, 148, 353–500)	P loop-containing nucleoside triphosphate hydrolases (SSF52540, 2.5E−8, 276–521)
11	12724	14148	1,425	474							DEAD-like helicases superfamily (smart00487, 2.98E−06, 179, 341–519)
12	14260	14847	588	195	Hypothetical protein OLV7	Organic Lake virophage	ADX05768	1.07E−15	29.7	161 (32–192)		SP (1–30), TM (15–35, 41–56, 77–97)
13	15207	14911	297	98
14	15192	16394	1,203	400	Putative minor capsid protein	Organic Lake virophage	ADX05769	3.00E−19	24.9	393 (1–393)
15	16489	18243	1,755	584	Putative capsid protein V20	Sputnik virophage	YP_002122381	2.15E−37	26.1	554 (10–563)
16	18726	18424	303	100								SP (1–19), TM (4–22)
17	18826	19386	561	186
18	19764	20378	615	204	Hypothetical protein MV08	Mavirus	YP_004300286	0.46	32	96 (88–183)
19	20460	21449	990	329
20	21536	21841	306	101	Hypothetical protein	Paramecium bursaria chlorella virus 1	NP_048469	3.43E−11	33.8	80 (7–86)
21	21902	23116	1215	404	N terminus: hypothetical protein OLV5	Organic Lake virophage	ADX05766	1.24E−08	24.6	197 (25–221)
YSLV3
1	1	765	765	254	Hypothetical protein OLV4	Organic Lake virophage	ADX05765	7.58E−54	40	248 (4–251)
2	950	762	189	62
3	976	1308	333	110
4	1772	1311	462	153								SP (1–18)
5	2528	2010	519	172	Hypothetical protein OLV7	Organic Lake virophage	ADX05768	4.81E−14	30.5	164 (5–168)
6	3442	2606	837	278	Hypothetical protein OLV12	Organic Lake virophage	ADX05773	6.04E−27	55.7	113 (135–247)
7	4020	3523	498	165	Hypothetical protein HMPREF9628_01282	Eubacteriaceae bacterium CM5	ZP_09316646	2.85E−13	32.2	158 (7–164)	Site-specific DNA methylase (COG0338, 1.22E−16, 160, 5–164)
					Putative modification methylase DpnIIA	Clostridium phage phiSM101	YP_699979	3.00E−11	32	160 (5–164)	DNA adenine methylase (TIGR00571, 6.15E−11, 154, 5–158)	S-Adenosyl-l-methionine-dependent methyltransferases (SSF53335, 5.6E−21, 5–165)
8	6254	3954	2,301	766							D12 class N6 adenine-specific DNA methyltransferase (pfam02086, 2.44E−07, 152, 13–164)
9	7249	6317	933	310	Hypothetical protein OLV11	Organic Lake virophage	ADX05772	1.35E−48	39.4	285 (16–300)
10	7715	7311	405	134
11	10417	7820	2,598	865	D5-ATPase-helicase, partial	Moumouvirus ochan	AEY99298	1.27E−33	29.8	426 (287–712)	Phage/plasmid primase, P4 family, C-terminal domain (TIGR01613, 1.40E−22, 293, 495–787)	D5_N (PF08706, 1.4E−11, 390–542)
											D5 N terminus-like (pfam08706, 2.13E−06, 89, 434–522)	PriCT_2 (PF08707, 1.2E−4, 279–345)
12	11103	10600	504	167	Hypothetical protein	Mavirus	YP_004300284	5.79E−02	47.2	51 (24–74)	Phage-associated DNA primase (COG3378, 1.08E−22, 379, 394–772)
13	11975	11307	669	222
14	12033	14423	2,391	796
15	14450	14650	201	66								SP (1–25), TM (5–25)
16	14726	17311	2,586	861
17	17402	17722	321	106
18	18065	19318	1,254	417	Putative minor capsid protein	Organic Lake virophage	ADX05769	4.26E−12	25.2	386 (26–411)
19	19412	21148	1,737	578	Putative capsid protein V20	Sputnik virophage	YP_002122381	4.79E−39	26.8	544 (9–552)
20	21761	22132	372	123
21	22204	23868	1,665	554	C terminus: hypothetical protein OLV10	Organic Lake virophage	ADX05771	0.13	27	133 (421–553)
22	23942	26155	2,214	737
23	26196	27023	828	275	Hypothetical protein OLV5	Organic Lake virophage	ADX05766	1.98E−13	27.6	190 (26–215)
YSLV4
1	1	768	768	255	Hypothetical protein OLV4	Organic Lake virophage	ADX05765	3.10E−67	44	248 (1–248)
2	831	1922	1092	363
3	1982	2317	336	111
4	2341	3300	960	319	Ribonucleoside-diphosphate reductase small subunit	Phaeocystis globosa virus 12T	AET72957	3.39E−136	60.5	317 (3–319)	Ribonucleotide reductase, R2/beta subunit, ferritin-like diiron-binding domain (cd01049, 9.46E−102, 274, 13–286)	Ribonuc_red_sm (PF00268, 1.0E−98, 7–283)
												Ferritin-like (SSF47240, 3.1E−105, 1–303)
5	3337	3489	153	50								SP (1–43), TM (25–45)
6	3496	3687	192	63								SP (1–21), TM (5–25)
7	3730	4248	519	172	Hypothetical protein PRSM4_062	Prochlorococcus phage P-RSM4	YP_004323191	7.07E−09	43.4	76 (12–87)		N6_MTASE (PS00092, −1.0, 76–82)
8	4638	4339	300	99
9	4820	4680	141	46
10	5196	5525	330	109
11	5700	8342	2,643	880	C terminus: D5-like helicase-primase	Marseillevirus	YP_003407183	9.04E−20	28.2	218 (474–691)		VirE (PF05272, 4.4E−5, 554–695)
												PriCT_2 (PF08707, 1.1E−4, 300–375)
12	8402	8551	150	49								TM (14–32)
13	8602	8802	201	66
14	8836	9816	981	326	Hypothetical protein OLV11	Organic Lake virophage	ADX05772	7.68E−59	42.1	296 (11–306)
15	9862	10302	441	146
16	10874	10299	576	191	Hypothetical protein OLV7	Organic Lake virophage	ADX05768	8.06E−27	37.2	147 (37–183)
17	10991	11410	420	139
18	11412	11855	444	147
19	11880	12311	432	143
20	12462	12662	201	66
21	12890	14074	1,185	394	Putative minor capsid protein	Organic Lake virophage	ADX05769	3.20E−40	31.2	359 (27–385)
22	14173	16026	1,854	617	Major capsid protein	Organic Lake virophage	ADX05770	6.10E−48	28	569 (2–570)
23	16143	18170	2,028	675
24	18232	19557	1,326	441
25	21005	19740	1,266	421	C terminus: hypothetical protein OLV12	Organic Lake virophage	ADX05773	8.69E−38	42.2	201 (209–409)		Alpha/beta-hydrolases (SSF53474, 1.3E−9, 238–360)
					N terminus: hypothetical protein OLV12	Organic Lake virophage	ADX05773	2.19E−19	36	194 (14–207)
26	21151	21834	684	227	C terminus: hypothetical protein OLV2	Organic Lake virophage	ADX05763	9.46E−10	41.1	91 (137–227)
27	22466	21915	552	183
28	23048	22686	363	120
29	24066	23677	390	129
30	24923	24360	564	187
31	25230	25613	384	127
32	25909	26466	558	185	N-Acetyltransferase GCN5	Clostridium phytofermentans ISDg	YP_001560895	1.03E−04	27.7	82 (67–148)	Acetyltransferase family (pfam00583, 1.11E−09, 85, 62–146)	Acetyltransf_1 (PF00583, 2.5E−10, 67–146)
33	26677	27198	522	173								Acyl-coenzyme A N-acyltransferases (Nat) (SSF55729, 1.2E−12, 48–168)
34	27327	28304	978	325	Hypothetical protein OLV5	Organic Lake virophage	ADX05766	3.04E−25	28.4	305 (14–318)
ALM
1	706	840	135	44
2	890	2551	1,662	553	Hypothetical protein	Monosiga brevicollis MX1	XP_001743771	5.06E−37	28.9	363 (8–370)	Phage/plasmid primase, P4 family, C-terminal domain (TIGR01613, 1.80E−05, 246, 163–408)	Winged helix DNA-binding domain (SSF46785, 7.9E−6, 366–441)
					Highly derived D5-like helicase-primase	Marseillevirus	YP_003406787	5.51E−16	26.4	283 (144–426)		P loop-containing nucleoside triphosphate hydrolases (SSF52540, 2.2E−6, 126–294)
3	2868	2593	276	91	Putative RVE superfamily integrase	Mavirus	YP_004300280	3.63E−08	39.4	91 (1–91)	Chromo (chromatin organization modifier) domain (pfam00385, 8.54E−03, 21, 56–76)	Chromatin organization modifier domain (SM00298, 8.0E−10, 35–90)
4	3689	3033	657	218	Putative RVE superfamily integrase	Mavirus	YP_004300280	2.21E−36	41.1	192 (1–192)	Integrase core domain (pfam00665, 4.94E−07, 114, 46–159)	Rve (PF00665, 8.4E−16, 48–157) RNase H-like
												(SSF53098, 1.6E−18, 38–209)
5	3773	5533	1,761	586	Putative protein-primed B-family DNA polymerase	Mavirus	YP_004300281	8.21E−88	35	536 (47–582)	DNA polymerase type B, organellar and viral (pfam03175, 1.34E−03, 224, 291–514)	DNA/RNA polymerases (SSF56672, 1.2E−19, 290–569)
6	5578	5850	273	90	Hypothetical protein MV04	Mavirus	YP_004300282	5.9	31	45 (2–46)
7	5884	6315	432	143
8	8027	6366	1,662	553	Putative major capsid protein MV18	Mavirus	YP_004300296	1.18E−76	31.3	522 (3–524)
9	8947	8057	891	296	Hypothetical protein MV17	Mavirus	YP_004300295	1.85E−65	42.1	291 (1–291)
10	9511	8984	528	175	Putative cysteine protease MV16	Mavirus	YP_004300294	1.25E−54	52.2	175 (1–175)
11	10541	9537	1,005	334	Putative FtsK-HerA family ATPase MV15	Mavirus	YP_004300293	7.25E−80	50	266 (69–334)		AAA_17 (PF13207, 1.1E−6, 95–228)
12	11354	10566	789	262	Hypothetical protein MV14	Mavirus	YP_004300292	5.20E−55	43.6	259 (1–259)
13	11406	12266	861	286	N terminus: hypothetical protein MV13	Mavirus	YP_004300291	9.03E−06	29.5	129 (1–129)
14	12160	12930	771	256	Hypothetical protein MV12	Mavirus	YP_004300290	1.88E−69	52.6	196 (57–252)		TM (34–52)
15	12956	13714	759	252
16	13774	14406	633	210	Hypothetical protein MV09	Mavirus	YP_004300287	0.21	29	97 (103–199)
17	14430	15020	591	196	Hypothetical protein MV08	Mavirus	YP_004300286	1.02E−12	48.3	84 (112–195)
18	15794	15060	735	244	Hypothetical protein MV13	Mavirus	YP_004300291	7.52E−13	29.1	198 (29–226)	Lipase (class3) (cd00519, 3.78E−08, 123, 96–218)	Lipase_3 (PF01764, 2.1E−4, 132–162)
19	15806	16093	288	95								Alpha/beta-hydrolases (SSF53474, 5.1E−10, 95–178)
20	16261	17145	885	294
21	17400	17224	177	58
22	17579	17445	135	44

^aaa, amino acids.

Conserved genes of virophages.

Based on a blastp and PSI-BLAST search against NCBI nr databases and a local database comprising all ORFs of eight virophages (five in this study and three published), five genes were found to be present in all eight virophages (Table 5). They were putative FtsK-HerA family DNA packaging ATPase and genes encoding putative DNA helicase/primase (HEL/PRIM), putative cysteine protease (PRSC), putative MCP, and putative minor capsid protein (mCP). These four genes had blastp hits to virophage genes only (E-value<10⁻¹), with the exception of HEL/PRIM (Table 4). Sequence alignment of these four proteins also revealed unambiguous similarity of amino acids (data not shown). Hence, it is reasonable to define them as virophage conserved core genes. The HEL/PRIM homolog was predicted according to either functional domains or sequence similarity, since significant sequence similarity was undetectable among some virophage species (Table 4).

Table 5.

Gene homologues present in virophages

Gene product	ORF(s) (size in aaa) in indicated virophage
	YSLV1	YSLV2	YSLV3	YSLV4	OLV	Sputnik	ALM	Mavirus
Putative FtsK-HerA family ATPase	01 (256)	01 (254)	01 (254)	01 (255)	04 (256)	03 (245)	11 (334)	15 (310)
Putative DNA helicase/primase/polymerase	04 (766)	10 (942)	11 (865)	11 (880)	25 (777)	13 (779)	02 (553)	01 (652)
Putative GIY-YIG endonuclease	09 (225)		12 (167)		24 (129)	14 (114)		06 (165)
Hypothetical protein	10 (308)	04 (344)	09 (310)	14 (326)	11 (298)
Putative cysteine protease	23 (190)	12 (195)	05 (172)	16 (191)	07 (190)	09 (175)	10 (175)	16 (189)
Putative major capsid protein	25 (623)	15 (584)	19 (578)	22 (617)	09 (576)	20 (595)	08 (553)	18 (606)
Putative minor capsid protein	27 (477), 26 (866)	14 (400)	18 (417)	21 (394)	08 (389)	18 (167), 19 (218)	09 (296)	17 (303)
Hypothetical protein	28 (104)	20 (101)	03 (110)	26 (227)	02 (123)
Hypothetical protein		02 (171)		07 (172)
Hypothetical protein		09 (184)					07 (143)
Hypothetical protein		18 (204)					17 (196)	08 (122)
Hypothetical protein		21 (404)	23 (275)	34 (325)	05 (290)	21 (438)
Hypothetical protein			06 (278)	25 (421)	12 (347)
Hypothetical protein			10 (134)	17 (139)
Hypothetical protein			21 (554)		10 (236)		12 (262)	14 (271)
Putative rve superfamily integrase							03 (91), 04 (218)	02 (358)
Putative protein-primed B-family DNA polymerase							05 (586)	03 (617)
Hypothetical protein							06 (90)	04 (112)
Hypothetical protein							13 (286), 18 (244)	13 (712)
Hypothetical protein							14 (256)	12 (211)
Hypothetical protein							16 (210)	09 (190)

^ant, nucleotides; aa, amino acids.

Besides these five conserved genes, the four YSLVs shared two other homologous genes with unknown functions, which were present in the OLV as well, but not in Sputnik, Mavirus, or ALM (Table 5 and Fig. 5). Interestingly, in all four YSLVs, homolog counterparts of the conserved genes of ATPase, PRSC, and mCP always showed the highest sequence similarity to that in OLV (Table 4); their second and third matches were strictly in the order of Sputnik and Mavirus. In most cases, their blast E-values were >10⁻⁵ for Mavirus hits but <10⁻¹⁰ for Sputnik hits. Taken together, these results suggested that the YSLVs were more closely related to OLV than to Sputnik and that they were distantly related to Mavirus.

Fig 5.

Numbers of homologous genes shared among OLV and YSLVs.

The evolutionary relationship between Mavirus and ALM was evident, as they shared 13 homologous genes (Table 4 and Fig. 4). Among them, five were virophage conserved genes, three encoded putative GIY-YIG endonuclease, putative rve (integrase core domain) superfamily integrase, and putative protein-primed B-family DNA polymerase, and five were functionally unknown. Furthermore, three syntenic regions existed between Mavirus and ALM (Fig. 4); however, two of these regions ran in opposite directions in the two virophages (Fig. 4).

Conserved gene clusters.

In this study, a gene cluster (or order) was considered to be several adjacent genes whose arrangement was conserved in some virophages; if present in all eight virophages, it was defined as a conserved gene cluster. As shown in Figures 3 and 4, a conserved gene cluster, comprised of the two conserved genes MCP and mCP, was present in all eight virophages. YSLVs 2, 3, and 4 and OLV shared a gene cluster consisting of the core gene ATPase and an ORF of unknown function. Furthermore, a gene cluster of the conserved PRIM/HEL gene and an ORF with unknown function was detected in YSLVs 3 and 4, and Mavirus and ALM had three gene clusters in common.

Phylogeny and evolution.

Three virophage core genes, encoding ATPase, PRSC, and MCP, were used to reconstruct the phylogenetic tree. As shown in Figure 6, three phylogenetic affiliation groups were observed. YSLVs and OLV seemed to form a group of closely related virophages, and Mavirus and ALM were apparently derived from a common ancestor, whereas Sputnik was an orphaned group. Such phylogenetic clustering of virophages was in agreement with the findings of the physical features of genomic DNA molecules, conserved genes, and gene orders as mentioned above. In addition, the phylogenetic trees of MCP and PRSC suggested that YSLVs were much closer to each other than to OLV (Fig. 6). This observation was consistent with the local tblastx results (search against a local database containing all ORFs of the eight virophages) that the best MCP hits of YSLVs were always themselves. Although it was impossible to shed light on the evolutionary relationship between these four YSLVs based on the current data, YSLVs 3 and 4 appeared to be the closest relatives. They were sister lineages on the MCP tree supported by a 70% bootstrap value (Fig. 6), shared the largest number of homologous genes (10) (Fig. 5), and had the highest number of gene clusters (three) (Fig. 3).

Fig 6.

Unrooted phylogenetic trees of DNA packaging ATPases (A), cysteine proteases (B), and major capsid proteins (C) of virophages. The five new virophages are shown in boldface. The numbers at the branches represent bootstrap values.

Habitat diversity of virophages.

Though they were more closely related to each other than to any other dsDNA viruses known so far, the habitats of these virophages were extremely diverse. Mavirus was from the coastal waters of Texas (1). Its closest relative ALM, however, was discovered in a hypersaline meromictic lake, Ace Lake (68°28′49″S, 78°11′19″E), in Antarctica. This lake is covered with ice for as long as 11 months to an entire year, with an average temperature of approximately 0°C (22). OLV was also found in the neighboring Organic Lake in Antarctica (3). In contrast, YSLVs, close to OLV, were found in a freshwater lake (Yellowstone Lake) with a temperature ranging from 12 to 73°C in Yellowstone National Park, Wyoming (23). Hence, these results indicated that virophages have adapted to habitats with a wide range of temperature variations.

In conclusion, the distinct abundance and global distribution of virophages, including almost all geographical zones as well as a variety of environments (ranging from the deep ocean to inland and iced to hydrothermal lakes), indicated that virophages appear to be widespread and genetically diverse, with at least three major lineages. Moreover, the overall low sequence similarity between the shared homologous genes in virophages and their distant phylogenetic relationships suggested that the genetic diversity of virophages is far beyond what we know thus far.