Related giant viruses in distant locations and different habitats: Acanthamoeba polyphaga moumouvirus represents a third lineage of the Mimiviridae that is close to the megavirus lineage.

The 1,021,348 base pair genome sequence of the Acanthamoeba polyphaga moumouvirus, a new member of the Mimiviridae family infecting Acanthamoeba polyphaga, is reported. The moumouvirus represents a third lineage beside mimivirus and megavirus. Thereby, it is a new member of the recently proposed Megavirales order. This giant virus was isolated from a cooling tower water in southeastern France but is most closely related to Megavirus chiliensis, which was isolated from ocean water off the coast of Chile. The moumouvirus is predicted to encode 930 proteins, of which 879 have detectable homologs. Among these predicted proteins, for 702 the closest homolog was detected in Megavirus chiliensis, with the median amino acid sequence identity of 62%. The evolutionary affinity of moumouvirus and megavirus was further supported by phylogenetic tree analysis of conserved genes. The moumouvirus and megavirus genomes share near perfect orthologous gene collinearity in the central part of the genome, with the variations concentrated in the terminal regions. In addition, genomic comparisons of the Mimiviridae reveal substantial gene loss in the moumouvirus lineage. The majority of the remaining moumouvirus proteins are most similar to homologs from other Mimiviridae members, and for 27 genes the closest homolog was found in bacteria. Phylogenetic analysis of these genes supported gene acquisition from diverse bacteria after the separation of the moumouvirus and megavirus lineages. Comparative genome analysis of the three lineages of the Mimiviridae revealed significant mobility of Group I self-splicing introns, with the highest intron content observed in the moumouvirus genome.

The family Mimiviridae consists of giant viruses that together with five previously recognized viral families and the candidate Marseilleviridae family comprise a monophyletic group of viruses known as nucleo-cytoplasmic large DNA viruses (NCLDV) (Iyer et al. 2001, 2006; Yutin and Koonin 2012). Recently, it has been proposed to combine all the NCLDV families into a new virus order tentatively named the Megavirales (Colson et al. 2012). The family Mimiviridae includes by far the largest viral genomes sequenced to date (La Scola et al. 2003; Claverie et al. 2009; Claverie and Abergel 2010). This is the only group of viruses with genomes larger than 1 megabase, which exceeds the genome size of numerous parasitic and symbiotic bacteria. The genomes of three Mimiviridae members have been completely sequenced and characterized in detail: Acanthamoeba polyphaga mimivirus (Raoult et al. 2004), the prototype of the family; A. castellanii mamavirus, which is a close relative, effectively a strain of the mimivirus (Colson et al. 2011); and Megavirus chiliensis that has been recently isolated from a marine environment (Arslan et al. 2011). In addition, 16 virus isolates of the family Mimiviridae have been identified and characterized by proteomic methods and/or partial sequencing (La Scola et al. 2010). Furthermore, marine metagenome analysis has revealed numerous homologs of mimivirus genes indicating that Mimiviridae is an abundant and diverse family of giant viruses whose host range remains unknown but includes organisms from habitats as different as marine water, fresh water, and soil (Monier et al. 2008; Kristensen et al. 2010; Yamada 2011).

Phylogenetic analysis of genes that are conserved in the majority of the NCLDV (Iyer, Aravind, et al. 2001; Iyer, Balaji, et al. 2006; Koonin and Yutin 2010) has shown that another giant virus that has been isolated from the marine microflagellate Cafeteria roenbergensis (CroV) is a distant member of the Mimiviridae (Fischer et al. 2010; Colson et al. 2012). In addition to genes that are shared with the other NCLDV, the giant viruses of the Mimiviridae family possess many genes that have not been previously detected in any viruses, in particular genes encoding components of the translation system such as aminoacyl-tRNA synthetases as well as a variety of metabolic enzymes (Raoult et al. 2004; Colson and Raoult 2010). The comparison of the mimivirus/mamavirus and the megavirus genomes has shown that only 77% of the megavirus proteins have readily detectable homologs in the mimivirus, suggestive of a large pangenome of the Mimiviridae (Arslan et al. 2011). Clearly, additional complete genomes of diverse members of the Mimiviridae are required for the characterization of this pangenome. Here, we describe the genome of another member of the Mimiviridae that we denoted A. polyphaga moumouvirus. The moumouvirus was isolated from water collected in a cooling tower but perhaps unexpectedly is most closely related to the megavirus that was identified in a marine environment.

The moumouvirus was isolated in February 2008 by inoculating A. polyphaga, as previously described, with water from an industrial cooling tower located in the south-east of France (La Scola et al. 2008). Some features of this virus have been briefly described previously (La Scola et al. 2010). Morphologically, the moumouvirus particles resemble the particles of other Mimiviridae (Klose et al. 2010; Arslan et al. 2011). The icosahedral capsid is approximately 420 nm in size and is covered by a dense layer of fibers (fig. 1). In comparison, A. polyphaga mimivirus and Megavirus chiliensis exhibit larger capsids with a diameter of approximately 500 and 520 nm, respectively (Klose et al. 2010; Arslan et al. 2011). In addition, these two mimiviruses harbor fibers that are approximately 125- and 75-nm long, respectively, whereas the size of the moumouvirus fibers is approximately 100 nm. Some of the moumouvirus particles also exhibit, similar to other Mimiviridae members, a distinctive, starfish-like vertex (Klose et al. 2010; Arslan et al. 2011). Finally, viral factories were observed within the A. polyphaga cytoplasm during the replication cycle of the moumouvirus; the morphology of the moumouvirus factories is similar to that observed previously for A. polyphaga mimivirus and Megavirus chiliensis (Suzan-Monti et al. 2007; Arslan et al. 2011).

F

Cryo-electron micrograph of moumouvirus particles. The viral particles have a dense layer of fibers and their morphology resembles the shape of other Mimiviridae members, including a distinctive, starfish like vertex (arrow). Scale bar: 200 nm.

The moumouvirus genome DNA was sequenced using the 454-Roche GS20 device (Roche Diagnostics Corp., Branford, CT) (Raoult et al. 2004; Margulies et al. 2005) and then the AB SOLiD instrument (Life Technologies Corp., Carlsbad, CA). The genome assembly was performed using a combination of Roche 454 paired-end and AB SOLiD sequencing reads (supplementary methods, Supplementary Material online). The moumouvirus genome is 1,021,348 base pairs (bp) in length which is more than 200 kilobase (kb) shorter than the megavirus genome (the current record holder in viral genome size) and more than 100 kb shorter than the mimivirus and mamavirus genomes (the moumouvirus genome sequence was deposited in GenBank under the Accession Number JX962719). Using pulse-field gel electrophoresis, the moumouvirus genome was characterized as a linear DNA molecule of approximately 1 megabase (not shown). Using a combination of prediction tools (supplementary methods and file 1, Supplementary Material online), 930 open reading frames (ORFs) were identified as putative protein-coding genes, with the mean predicted protein size of 290 amino acids (aa). These ORFs are evenly distributed on both DNA strands, with 470 predicted genes located on the “direct” strand and 460 on the “reverse” strand. The mean size of intergenic regions is 130 ± 166 nucleotides, with the predicted protein-coding density of 0.91 genes/kb (as compared with 0.89 genes/kb for the megavirus). In addition, three tRNA genes were predicted using the tRNAscan-SE method (Schattner et al. 2005). The ORFs were analyzed for evolutionary conservation, protein domain content and predicted functions by using PSI-BLAST search (Altschul et al. 1997) of the Refseq database at the NCBI (one iteration by default and up to 3 iterations when initial functional prediction was ambiguous), domain identification by RPS-BLAST search of the Conserved Domain Database (Marchler-Bauer and Bryant 2004), and assignment of proteins to clusters of orthologous NCLDV genes (NCVOGs) (Yutin et al. 2009).

Of the 930 predicted proteins of the moumouvirus, for 879 homologs were detected by protein sequence database search, and for the great majority, the most similar homolog was a megavirus protein (supplementary file 1, Supplementary Material online). For 656 predicted proteins of the moumouvirus, the megavirus homolog was a bidirectional best hit (BBH) (with the expect value cut-off of 10−3); that is, a probable ortholog. The putative moumouvirus–megavirus orthologous protein pairs ranged in identity from 91% to 23%, with a median of 62%. An analogous comparison between moumouvirus and mimivirus yielded 548 putative orthologs, with a median 52% identity, indicating that the moumouvirus is more similar to the megavirus than it is to the mimivirus in terms of both the gene repertoire and sequence conservation. The evolutionary affinity of the moumouvirus and the megavirus was clearly supported by the results of phylogenetic analysis of concatenated conserved NCLDV proteins (fig. 2 and supplementary file 2, Supplementary Material online). Although the trees for individual conserved genes showed topological differences for other branches within Phycodnaviridae and Mimiviridae, the moumouvirus–megavirus clade and its monophyly with the mimivirus–mamavirus clade were invariably recovered (supplementary file 2, Supplementary Material online). In addition, a genomic dot-plot of the moumouvirus against the megavirus reveals near perfect collinearity of orthologous genes in the middle part of the genome (∼650 kb), with rearrangements found only in the peripheral parts of the genomes (fig. 3A). This similarity of genome architectures contrasts the results of the comparison of the moumouvirus and mimivirus genomes that shows shorter, interrupted collinear regions and in addition a large inversion in the central part of the genomes (fig. 3B), similar to that described from the comparison of the megavirus and mimivirus genomes (Arslan et al. 2011). Conservation of the gene order in the middle of the genome with divergence at the genome ends seems to be a general feature of NCLDV evolution that was first noted in poxviruses (Senkevich et al. 1997) and has been more recently pointed out for Chlorella phycodnaviruses (Filee et al. 2007), marseillevirus and lausannevirus (Thomas et al. 2011), mamavirus, mimivirus, and CroV (Boyer et al. 2011; Colson, Gimenez, et al. 2011; Colson et al. 2011).

F

Phylogenetic tree of the Mimiviridae and selected Phycodnaviridae constructed from concatenated alignments of DNA polymerase, A32-like packaging ATPase, and A2-like Transcription Factor. Marseillevirus and lausannevirus were used as an outgroup. The alignment included 1,429 positions that were deemed reliably aligned. The bootstrap values (percentage points) are indicated for each internal branch (Mimiviridae).

F

Genomic dot plots for the moumouvirus and other members of the Mimiviridae. (A) Moumouvirus versus Megavirus chiliensis. (B) Moumouvirus versus Acanthamoeba polyphaga mimivirus. Each point represents a pair of orthologous genes (BBHs in BLASTP searches).

Together, these observations indicate that megavirus and moumouvirus comprise a distinct branch of the Mimiviridae. Given the moderate sequence conservation between the orthologs and differences in the gene repertoire (discussed later), moumouvirus and megavirus clearly are distinct virus species unlike mimivirus and mamavirus that, at >98% mean identity between orthologous proteins and near perfect genomic collinearity, are most appropriately considered strains of the same species (Colson et al. 2011). These findings are in line with the recent demonstration of three evolutionary lineages within the Mimiviridae (Colson et al. 2012).

Although moumouvirus is the sister group of megavirus in the phylogenetic tree of the Mimiviridae (fig. 2), its genome is more than 200 kb smaller than the megavirus genome. Of the 1,120 predicted protein-coding genes of the megavirus (Arslan et al. 2011), for 464 no one-on-one ortholog has been detected in the moumouvirus. Analysis of these megavirus proteins showed that 219 are members of paralogous families common for Mimiviridae members; 139 are ORFans without detectable homologs; 21 apparently were acquired from sources outside the Mimiviridae (mainly from bacteria); and 85 are shared by megavirus and mimivirus/mamavirus but absent in the moumouvirus (supplementary file 3, Supplementary Material online). Thus, these 85 genes that are located in the terminal regions of the genome apparently have been lost in the moumouvirus lineage; an alternative, less parsimonious evolutionary scenario would involve independent acquisition of these genes in the mimivirus and megavirus lineages. Most of the genes that are inferred to have been lost by the moumouvirus are functionally uncharacterized but for some functions could be predicted, in particular in DNA repair (supplementary file 3, Supplementary Material online). Interestingly, one of the lost genes encodes the small polyA polymerase subunit/cap O-methyltransferase, a gene that is shared by megavirus, mamavirus, and poxviruses but is missing in the rest of the NCLDV, suggestive of multiple losses (Colson et al. 2011). The demonstration of extensive gene loss in the moumouvirus echoes the dramatic reduction in the mimivirus genome size after cultivation in germ-free amoeba; notably, the size of the terminal regions that have been eliminated from the mimivirus genome after multiple passages is approximately the same (∼200 kb) as the difference in genome size between megavirus and moumouvirus (Boyer et al. 2011).

In addition to being the largest viral genome sequenced to date, the megavirus is notable for encoding the largest number of translation system components among all viruses including 7 aminoacyl-tRNA synthetases (aaRS) (Arslan et al. 2011). The moumouvirus encodes apparent orthologs of many but not all of these proteins, in particular 5 aaRS (supplementary files 1 and 3, Supplementary Material online).

The majority of the moumouvirus protein-coding genes have apparent conserved orthologs in the megavirus but the remaining genes showed some interesting evolutionary patterns. Two genes encoding metabolic enzymes, cysteine dioxygenase and NAD-dependent epimerase/dehydratase, are shared by moumouvirus and CroV to the exclusion of the other NCLDV. Phylogenetic analysis of both genes demonstrated monophyly of the two giant viruses, along with some uncharacterized environmental sequences (fig. 4A and B). This phylogeny implies the presence of these genes in the common ancestor of the Mimiviridae with at least two subsequent losses (in the mimivirus and megavirus branches) or the less likely evolutionary scenarios involving gene exchange between moumouvirus and CroV or independent acquisition of genes from related sources by the two viruses. Phylogenetic analysis of the moumouvirus genes with closest bacterial homologs supported the origin of these genes from diverse bacteria (two examples are shown in fig. 4C and D), in agreement with the previously noticed extensive gene exchange among symbionts and parasites of amoeba (Ogata et al. 2006; Moreira and Brochier-Armanet 2008; Boyer et al. 2009; Raoult and Boyer 2010).

F

Phylogenetic trees of moumouvirus genes missing in other members of the Mimiviridae. (A) Cysteine dioxygenase, (B) NAD-dependent epimerase/dehydratase, (C) Methyltransferase, and (D) Nudix Hydrolase. The bootstrap values (percentage points) are indicated for each internal branch. Each sequence is denoted by taxa abbreviation and species name. GenBank Identification (GI) numbers for Cafeteria roenbergensis virus are shown on the trees. For other sequences, GI numbers are as follows: Phytophthora sojae, 348690046; Albugo laibachii Nc14, 325183169; Psychroflexus torquis ATCC 700755, 91214785; Dictyostelium fasciculatum, 328865331; Dictyostelium discoideum AX4, 66806929; Aureococcus anophagefferens, 323447704; Gymnochlora stellate, 193875832; Fluviicola taffensis DSM 16823, 327404160; Marivirga tractuosa DSM 4126, 313675758; Cytophaga hutchinsonii ATCC 33406, 110637252; Microscilla marina ATCC 23134, 124004204; Saprospira grandis str. Lewin, 379730028; Kordia algicida OT-1, 163754599; Lacinutrix sp. 5H-3-7-4, 336171271; Niastella koreensis GR20-10, 375148557; wenweeksia hongkongensis DSM 17368, 375013912; Hydra magnipapillata, 221105922; Arthrobacter arilaitensis Re117, 308177223; Actinomyces urogenitalis DSM 15434, 227495863; Myxococcus fulvus HW-1, 338534578; Chelativorans sp. BNC1, 110632790; Spirochaeta thermophila DSM 6192, 307718496; Chlorobium phaeovibrioides DSM 265, 145219574; Flavobacteria bacterium BAL38, 126664257; Neisseria shayeganii 871, 349575093; Dethiosulfovibrio peptidovorans DSM 11002, 288575048; Methylomicrobium alcaliphilum 20Z, 357407184; Desulfovibrio magneticus RS-1, 239906771; Desulfovibrio salexigens DSM 2638, 242280667; Ramlibacter tataouinensis TTB310, 337278313; Opitutus terrae PB90-1, 182415517; Candidatus Chloracidobacterium thermophilum B, 347756407; Mus musculus, 148701036; Homo sapiens, 344217763; Methylibium petroleiphilum PM1, 124266300; Ferrimonas balearica DSM 9799, 308049455; Desulfuromonas acetoxidans DSM 684, 95930587; Blastopirellula marina DSM 3645, 87312225; Stackebrandtia nassauensis DSM 44728, 291297650; Renibacterium salmoninarum ATCC 33209, 163840224; Deinococcus radiodurans R1, 6458004; Bacillus sp.916, 394994508; Bifidobacterium longum NCC2705, 23465838; Frankia sp. EAN1pec, 68197430; Mycobacterium tuberculosis CDC1551, 13879930; Escherichia coli, 50513417; Bacillus cereus, 51975946; Bacillus halodurans C-125, 10176350. Taxa abbreviations: Ba, Actinobacteria; Bb, Bacteroidetes/Chlorobi group; Bd, Deinococcus-Thermus; Bf, Firmicutes; Bi, Acidobacteria; Bo, Planctomycetes; Bp, Proteobacteria; Bs, Spirochaetes; Bv, Chlamydiae/Verrucomicrobia group; Bw, Synergistetes; E8, stramenopiles; Ea, Amoebozoa; Ed, Rhizaria; El, Opisthokonta.

Similarly to other Mimiviridae members, moumouvirus genes were found to contain 8 Group I self-splicing introns and three inteins (supplementary file 4, Supplementary Material online). All sequenced members of the Mimiviridae share an apparent ancestral intron in the gene for the largest subunit of the RNA polymerase (RNAP) and an ancestral intein in the DNA polymerase gene. The moumouvirus contains only a single intron in the major capsid protein gene, similar to the mamavirus, whereas the megavirus and the mimivirus contain two introns in this gene. In the HSP70 chaperone gene, megavirus and moumouvirus share an intron to the exclusion of the other Mimiviridae members but moumouvirus lacks the intron-encoded endonuclease ORF. In addition, the moumouvirus contains short inteins that consist of the cis-acting HINT protease domain alone in the genes encoding the repair ATPase MutS and an uncharacterized protein (supplementary file 4, Supplementary Material online). The positions of other introns in the genomes of the Mimiviridae members vary and several new introns were identified. In particular, unlike other Mimiviridae members, the moumouvirus contains the largest number of introns (5), along with an intein, in the second largest RNAP subunit gene (supplementary file 4, Supplementary Material online). These findings emphasize the dynamic evolution of the introns and inteins in the Mimiviridae.

Analysis of the mimivirus and megavirus genomes has revealed two distinct features of the transcripts, namely the conserved octameric motif AAAATTGA upstream of protein-coding sequences (Suhre et al. 2005) and stable hairpin structures or palindromic sequences at the 3′-ends of transcripts (Byrne et al. 2009). We searched for these two characteristic features in the moumouvirus genome. Overall, AAAATTGA sites were found within 150-nt regions upstream of the predicted start codons in 351 of the 930 predicted moumouvirus genes (37.7%). Among the orthologous genes between moumouvirus and megavirus, the fraction containing this motif was nearly the same. The AAAATTGA motif has been shown to function as an early promoter element in mimivirus and megavirus (Legendre et al. 2010). Together with the previously published comparison of megavirus and mimivirus genes containing this motif (Arslan et al. 2011), our observations on the presence of AAATTGA in upstream regions of moumouvirus genes imply that the expression pattern of orthologous genes is largely conserved among these three members of the Mimiviridae.

We also detected palindromic sites and predicted thermodynamically stable hairpins in the 3′ intergenic regions (150-nts downstream of the stop codon) of 725 of the 930 (78%) predicted protein-coding genes of the moumouvirus. Most of these structures are well-conserved between moumouvirus and megavirus. For example, we aligned and predicted the consensus structure of the hairpin element at the 3′-end of the major capsid gene for mimivirus, megavirus, and moumouvirus (supplementary file 5, Supplementary Material online) for which the presence of this element in the mature transcript has been experimentally validated by RNA sequencing of megavirus mg464 (Arslan et al. 2011). The polyadenylation of the megavirus gene occurs in the predicted conserved hairpin. The position of the experimentally verified polyadenylation sites is conserved between mimivirus and megavirus allowing us to predict the polyadenylation site of the moumouvirus gene (supplementary file 5, Supplementary Material online). These findings are in good agreement with the previous results demonstrating the conservation of these structural elements between mimivirus and megavirus (Arslan et al. 2011), suggesting that these hairpins function as transcription termination signals in moumouvirus.

Conclusions

Analysis of the moumouvirus genome confirms that it represents a third lineage amongst the Mimiviridae, in addition to those represented by mimivirus and megavirus. The moumouvirus genome further expands the pangenome of the Mimiviridae and emphasizes the dynamic evolution of the giant viruses, in particular extensive gene loss. The evolutionary relationship between the moumouvirus isolated from freshwater amoeba and Megavirus chiliensis that was isolated from a marine environment but shown to reproduce in the amoeba host of the other mimiviruses (Arslan et al. 2011) suggests that these giant viruses have a broad host range leading to ecological plasticity.

Supplementary Material

Supplementary methods and files S1–S5 are available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).