Provirophages and transpovirons as the diverse mobilome of giant viruses

Abstract

A distinct class of infectious agents, the virophages that infect giant viruses of the Mimiviridae family, has been recently described. Here we report the simultaneous discovery of a giant virus of Acanthamoeba polyphaga (Lentille virus) that contains an integrated genome of a virophage (Sputnik 2), and a member of a previously unknown class of mobile genetic elements, the transpovirons. The transpovirons are linear DNA elements of ∼7 kb that encompass six to eight protein-coding genes, two of which are homologous to virophage genes. Fluorescence in situ hybridization showed that the free form of the transpoviron replicates within the giant virus factory and accumulates in high copy numbers inside giant virus particles, Sputnik 2 particles, and amoeba cytoplasm. Analysis of deep-sequencing data showed that the virophage and the transpoviron can integrate in nearly any place in the chromosome of the giant virus host and that, although less frequently, the transpoviron can also be linked to the virophage chromosome. In addition, integrated fragments of transpoviron DNA were detected in several giant virus and Sputnik genomes. Analysis of 19 Mimivirus strains revealed three distinct transpovirons associated with three subgroups of Mimiviruses. The virophage, the transpoviron, and the previously identified self-splicing introns and inteins constitute the complex, interconnected mobilome of the giant viruses and are likely to substantially contribute to interviral gene transfer.

Mobile genetic elements (MGEs) that are collectively referred to as the “mobilome” are key players in the genome evolution of prokaryotes (1) and eukaryotes (2, 3) and are considered “genetic engineers” of biological innovation (1). MGEs can be roughly grouped into four major classes: transposable elements (TEs), plasmids, viruses, and self-splicing elements such as group I and II introns and inteins (4). The mobilomes of many bacteria, archaea, and unicellular eukaryotes include all of these elements in a free or integrated form. Given that viruses constitute a part of the mobilome, they are not normally considered to possess mobilomes of their own. However, some large viruses contain retrovirus sequences integrated into their genomes (5, 6), whereas others, including members of the Mimiviridae family, harbor self-splicing introns and/or inteins (7–9). Furthermore, many viruses support the reproduction of satellite viruses (10).

The discovery of the Sputnik virophage in 2008 added a new twist to the existing understanding of the relationships between different mobile elements by demonstrating for the first time that a giant virus could be infected by another, much smaller virus in a manner similar to the viral infection of cells (11). The Sputnik virophage is a small icosahedral virus (74 nm in diameter) that parasitizes on Mamavirus, a member of the Mimiviridae family (12, 13). Sputnik replicates inside Mamavirus or Mimivirus viral factories when the host giant virus is grown in amoebae such as Acanthamoeba castellanii or A. polyphaga (11). An in-depth analysis of the Sputnik proteins has suggested an evolutionary connection between this virophage and a distinct class of TEs (14). The second virophage, the Mavirus (15), was isolated as a parasite of a distinct member of the Mimiviridae family, Cafeteria roenbergensis virus (CroV) (16). At least four Mavirus proteins, including the major capsid protein, are homologous to proteins of Sputnik. In addition, the Mavirus genome encodes a retroviral-type integrase and a protein-primed DNA polymerase B; these proteins are homologous to the respective proteins of Maverick/polinton DNA transposons, which insert into genomes of diverse eukaryotes, suggesting an evolutionary link between the Mavirus and the polintons (15). The third complete virophage genome sequence has been identified in the metagenome of the hypersaline Organic Lake in Antarctica (17). This Organic Lake virophage (OLV) is thought to parasitize on phycoDNAviruses that infect green algae. The OLV genome encodes seven proteins with homologs in Sputnik (17), including two key proteins, the major capsid protein and the DNA-packaging ATPase, that are shared by all three virophages. Thus, the virophages apparently share a common origin, although each underwent multiple gene replacements. The virophages are likely to be common parasites of nucleocytoplasmic large DNA viruses that infect diverse eukaryotes, and show multiple evolutionary connections to other mobile elements (18).

Here we present findings that substantially expand the complexity of the giant virus mobilome through the description of an integrated form of the virophage and of a distinct class of MGEs, the transpovirons.

Results

Isolation of a Distinct Giant Virus and Its (Pro)Virophage.

Sputnik 2 is the fourth virophage described thus far; unlike the previously identified virophages, Sputnik 2 was isolated from a human-associated sample (19). Sputnik 2 and its virus host, Lentille virus, were isolated from contact lens fluid of a patient with keratitis after treatment of the liquid with a mixture of selected antibiotics to prevent bacterial growth. Lentille virus particles were purified from Sputnik 2 as previously described for Mamavirus and Sputnik (11) using heat inactivation (68 °C) for 1 h and desiccation for 48 h. This virus isolate was then cloned and used to reinfect A. polyphaga. Spontaneous release of Sputnik 2 particles was observed in the infected amoebae, indicating that, although not detected by electron microscopy (SI Appendix, Fig. S1), Sputnik 2 or its DNA remained associated with Lentille virus particles.

Fluorescence in situ hybridization (FISH) was performed using Lentille virus and Sputnik 2-specific probes (Fig. 1A and SI Appendix, Fig. S2 for combined immunofluorescence-FISH). At 7 h postinfection (p.i.), a strong DAPI signal was detected in the viral factories (VFs). Both the host virus (red) and the virophage (green) reproduce within the VF (Fig. 1A). In addition, colocalization of the Lentille virus signal and the Sputnik 2 signal (Fig. 1A, white arrow) was observed for several viral particles already released into the cytoplasm of the amoeba, an observation that is compatible with virophage localization inside the host virus particles. We further performed pulsed-field gel electrophoresis (PFGE; Fig. 1B) on native, ApaI- (which does not cut the Sputnik 2 genome), and EagI- (cuts once inside the Sputnik 2 genome) digested Lentille virus genome (Fig. 1B, lanes 4, 7, and 9, respectively), Mimivirus genome (lanes 3, 6, and 8, respectively), and Sputnik 2 genome (Fig. 1B, lanes 5, 12, and 13, respectively). Mimivirus and Lentille virus DNA appeared as large fragments above the 339-kb low-range (LR) marker (Fig. 1B, lanes 3 and 4, white arrows). In contrast, the Sputnik 2 genome presented a multiband profile that apparently contains concatemeric genome units (Fig. 1B, lane 5) produced from the circular genomic DNA. In vitro formation of concatemers by the end-to-end joining of the mature genomes is a well-known process for T7-bacteriophage (20) and, in the case of Sputnik 2, digestion with EagI at the single site located between the V11 and V12 genes (Fig. 1B, lane 13 and Fig. 1D) linearized the genome units. Surprisingly, Southern blot hybridization using the Sputnik 2 genome as a probe identified a strong signal within the Lentille virus genome (Fig. 1C, black arrows), suggesting that the Sputnik 2 genome is integrated into the genome of its viral host.

Fig. 1.

Lentille virus and Sputnik 2. (A) (Inset) In situ hybridization using Sputnik 2 probe (green), Lentille virus probe (red), and DAPI nucleic acid staining (blue) after 7 h p.i.; the merged image shows intense replication of both Lentille virus and the virophage inside the same viral factory of A. polyphaga cells; the arrow indicates colocalization of the virophage and Lentille virus signals. (B and C) PFGE (B) and Southern blot (C) using DIG-labeled Sputnik 2 probes showing the Sputnik genome integrated into that of Lentille virus. Lane 1: λ marker; lane 2: low-range marker; lane 3: native Mimivirus genome; lane 4: native Lentille virus genome; lane 5: native Sputnik 2 genome; lane 6: ApaI-digested Mimivirus genome; lane 7: ApaI-digested Lentille virus genome; lane 8: EagI-digested Mimivirus genome; lane 9: EagI-digested Lentille virus genome; lane 10: λ marker; lane 11: low-range marker; lane 12: ApaI-digested Sputnik 2 genome; lane 13: EagI-digested Sputnik 2 genome. (D) Map of the Sputnik 1 (outer circle) and Sputnik 2 genomes (inner circle). The ORF number is indicated above the genome along with orientations, clockwise in blue and counterclockwise in red. A line indentifies the EagI restriction site, which cuts once into the Sputnik 1 and 2 genomes. Inside: Sputnik 2 genome GC skew (orange/yellow) and GC content.

Virophage Integration in the Giant Viral Host Genome.

Virus DNA isolated from cloned Lentille virus particles was pyrosequenced with the 454/Roche GS FLX Kit using shotgun technology and the 454/Roche GS FLX Titanium Kit using paired-end technology (SI Appendix, Table S1). The results of the de novo assembly of the combined runs are shown in SI Appendix, Table S4 (SI Appendix, Tables S2 and S3 for separate assembly of shotgun and paired-end pyrosequencing data, respectively). The Lentille virus genome was further sequenced using SOLiD technology, which gives higher coverage and allows closing gaps and correcting sequencing errors (SI Appendix, Table S1).

After finishing, the Lentille virus genome consisted of 10 contigs organized in one scaffold with an average coverage of 29× (SI Appendix, Table S5). A contig of 18,332 bp with a higher coverage (86×) represented the Sputnik-related sequence. The DNA from purified Sputnik 2 particles was also sequenced separately on a GS FLX titration plate, generating 1,700,533 bp assembled in a single contig. The curated sequence of Sputnik 2 consists of 18,338 bp (Fig. 1D).

A map of paired-end links joining contigs generated from the 454/Roche paired-end pyrosequencing run was produced (SI Appendix, Fig. S3) to visualize the connections between the different structures. One paired-read link was identified between Sputnik 2 and Lentille virus. This common region corresponds to a near-identical sequence of 395 nt identified at the end of the Sputnik V6 gene and in the Lentille virus and Mamavirus (but not Mimivirus) gene that encodes a collagen-like repeat-containing protein (SI Appendix, Figs. S4 and S5 A and B). PCR products of the sizes that were predicted under the assumption that the Sputnik genome integrated into the Lentille virus genome were obtained on purified Lentille virus particles from the right and left flanks of the putative insertion site by using one primer in Sputnik 2 and the other in the Lentille virus genome (SI Appendix, Fig. S4, Table S6, and SI Materials and Methods). The PCR product generated using the v5F-SpClR primer pairs was sufficiently pure (SI Appendix, Fig. S6) for direct sequencing. Multiple alignment of the v5F-SpClR PCR sequence (SI Appendix, SI Results) with the Sputnik 2 and Lentille virus sequences (SI Appendix, Fig. S7) showed a central common region of 402 nt flanked by a 3′ region similar to the Lentille virus sequence (identity of 55/58 nt) and a 5′ region similar to the Sputnik 2 sequence (identity of 121/122 nt). Results from massive SOLiD genome sequencing, which gives higher sequence coverage than pyrosequencing, confirmed the common region as a hotspot of integration (SI Appendix, Fig. S8A, red circle). However, the presence of a further candidate insertion site detected in the genome of Mamavirus or Lentille virus (SI Appendix, Fig. S5 A and C), along with the homogeneous distribution of mate pairs between Lentille virus and Sputnik 2 (SI Appendix, Fig. S8A), suggests that Sputnik 2 may also insert in multiple, possibly random, sites.

Transpovirons, a Distinct Class of Mobile Genetic Elements.

We also identified a contig of 7,420 bp (SI Appendix, Table S5). This abundant contig had a GC content of 24.7% and contained six ORFs and long terminal inverted repeats of ∼530 bp in length. This contig is an extrachromosomal DNA molecule that is present in an extremely high copy number in Lentille virus particles, as indicated by the sequencing coverage, which was 3- to 14-fold higher than the coverage of the Lentille virus for the pyrosequencing and SOLiD runs (SI Appendix, Table S5). This small linear plasmid resembles a TE and, because it is found inside a virus, we denoted it a “transpoviron.” The paired-end links inside the transpoviron, which uniformly cover the transpoviron sequence without any break, along with the presence of cohesive ends, imply the existence of circular intermediate forms (SI Appendix, Fig. S3). Five pairs of paired-end reads having one end located in various parts of the Lentille virus genome and the other end matching the transpoviron suggested nonspecific integration (SI Appendix, Fig. S3). Massive SOLiD sequencing revealed that the transpoviron can insert randomly into the Lentille virus genome (SI Appendix, Fig. S8B) and also, although less frequently, into the Sputnik genome (SI Appendix, Fig. S8C). The presence of the transpoviron in Lentille virus particles was further confirmed using PFGE (Fig. 2B) and Southern blot analysis (Fig. 2C) with a specific digoxigenin (DIG)-labeled probe (SI Appendix, Table S6). The transpoviron was not detected in the Sputnik 2 genomic preparation by PFGE (Fig. 2B, lane 4) but was identified by Southern blotting (Fig. 2C, lane 4), in accordance with the low transpoviron sequence coverage observed in the Sputnik 2 sequence data (0.5×) and suggesting that only a small fraction of the Sputnik particles encapsidates the transpoviron. The transpoviron was not detected in the uninfected amoeba genomes (either A. polyphaga or A. castellanii) (Fig. 2 B and C, lanes 5 and 6). Following infection of A. polyphaga by Lentille virus, the signal of the transpoviron was extremely intense (Fig. 2 B and C, lane 7), indicating active replication of the transpoviron in the infected amoebae. Subsequent PCR (SI Appendix, Table S6) revealed early production of large amounts of transpoviron DNA starting at 4 h p.i. (SI Appendix, Table S7). An in situ hybridization analysis using a transpoviron-specific probe showed a diffuse signal spread all over the amoeba cytoplasm. At 7 h p.i. the signal was very bright and appeared concentrated in rings surrounding the VFs (Fig. 2A). No transpoviron was detected in A. polyphaga infected by Sputnik 2 alone (Fig. 2 B and C, lane 8). To investigate the expression of the transpoviron during Lentille virus infection, RT-PCR was performed on RNA extracted from cultures of A. polyphaga infected with Lentille virus or Sputnik 2. The mRNA for the predicted transpoviron helicase was detected in the Lentille virus-infected but not Sputnik 2-infected cultures, indicating that expression of at least one transpoviron gene relies on infection of A. polyphaga by Lentille virus (SI Appendix, Fig. S9). The RT-PCR reactions with viral RNAs extracted from Sputnik 2 particles failed to detect transpoviron-specific transcripts. Thus, given that the transpoviron DNA was detected in purified Sputnik particles, we speculate that Sputnik is a transient host that could be a vehicle transferring transpoviron DNA to giant virus hosts.

Fig. 2.

Identification of the transpoviron. (A) (Inset) In situ hybridization using a Lentille virus probe (green), transpoviron probe (red), and DAPI nucleic acid staining (blue) at 7 h p.i. in A. polyphaga cells. (B) PFGE showing the Lentille virus genome (asterisk) and the transpoviron as a second band between the 6.55- and 9.42-kb bands of the LR marker (lane 3) identified with a white arrow. (C) Southern blot using DIG-labeled transpoviron probes. Lane 1: λ marker; lane 2: low-range marker; lane 3: Lentille virus genome; lane 4: Sputnik 2 genome; lane 5: A. polyphaga; lane 6: A. castellanii; lane 7: A. polyphaga infected with Lentille virus and Sputnik 2; lane 8: A. polyphaga infected with Sputnik 2; lane 9: A. polyphaga infected with Mimivirus.

Distinct Transpovirons Are Associated with Different Groups of Mimiviridae.

Our current collection of Mimiviridae includes 19 virus isolates, namely Mimivirus (7)/Mamavirus (11), 16 isolates partially characterized previously (21), and 2 additional, recently isolated viruses, Ochan and Marais (SI Appendix, SI Materials and Methods). Phylogenetic analysis of these isolates based on the multiple alignment of partial sequences of the DNA polymerase gene revealed three lineages, which were denoted A (Mimivirus group), B (Moumouvirus group), and C (Courdo11 virus group) (22). We extended this phylogenetic analysis by including the highly conserved genes for the DNA polymerase, D5-like helicase, and a viral transcription factor (Fig. 3; the corresponding accession numbers are in SI Appendix, Table S8). The recently described Megavirus chilensis (23) falls into group C, and group B now contains the Moumou, Monve, and Ochan viruses.

Fig. 3.

The spread of transpovirons and virophages in giant viruses. The figure shows the pattern of presence–absence of transpovirons and virophages among the three groups of giant viruses isolated in our laboratory (black) and by others (blue). The identified transpovirons and virophages are mapped onto the phylogenetic tree obtained from concatenated alignments of DNA polymerase B, D5-like ATPase/helicase, and a transcription factor of giant viruses. Accession numbers are indicated in SI Appendix, Table S8.

In the sequencing data for Courdo7, Monve, and Mamavirus, contigs with 2- to 10-fold excess coverage were detected. These sequences of 6,594, 6,717, and 7,438 bp, respectively, were identified as transpovirons by similarity comparison with the originally isolated, Lentille virus-associated transpoviron. Thus, transpovirons were detected in Mimiviruses from all three groups (Fig. 3). The available sequences of Megavirus chilensis do not harbor transpovirons, but we cannot rule out that reads corresponding to the transpoviron have been rejected. Thus, the current genomic survey reveals a broad even if sporadic occurrence of transpovirons among the Mimiviruses. In contrast, the virophages so far show a more narrow spread, with Sputnik detected only in association with the Mimiviruses (group A) and the distantly related Mavirus virophage associated with CroV that represents the outgroup to the A, B, and C groups (Fig. 3).

Comparative Genomics of the Transpovirons and the Links Between Components of the Giant Virus Mobilome.

Comparative genomic analysis of the transpovirons showed general conservation of gene organization, with some variations (Fig. 4 and SI Appendix, Figs. S10 and S11). Long terminal inverted repeats (TIRs) were detected only in the Lentille virus and Mamavirus transpovirons, in which they were nearly identical in sequence. The Monve and Courdo7 transpovirons lacked the TIRs, possibly because their genome sequences were incomplete at the ends. The TIRs are extremely AT-rich (>80% AT), and upon closer inspection were found to consist of six imperfect, direct tandem repeats (SI Appendix, Fig. S11). The presence of long TIRs in the transpovirons resembles the genome structure of the mavericks (polintons), the large eukaryotic DNA TEs (24) that notably are related to virophages (15); other DNA TEs have shorter TIRs (if any).

Fig. 4.

Organization of genes in transpoviron genomes. The approximate position and strand orientation of each ORF (>150 nt) is indicated by an arrow. The ORFs with homologs outside the transpovirons are color-coded and their predicted biochemical activities are indicated (TM, protein with a predicted transmembrane helix; C2H2, Zn-finger protein; Super Family I (SFI) Helicase). An ORF that is homologous between different transpovirons is rendered in blue, and completely uncharacterized ORFs (ORFans) are rendered in white.

The transpovirons encompass six to eight predicted protein-coding genes; two genes are conserved in all four sequenced genomes, whereas the rest of the genes are missing in one or more of the transpovirons (Fig. 4 and SI Appendix, Table S9). The largest and most highly conserved protein encoded in all transpovirons consists of a C-terminal superfamily I helicase domain, which contains the seven signature helicase motifs and is predicted to be an active DNA helicase (SI Appendix, Fig. S12), and an uncharacterized N-terminal region. Helicase domains are also found in other MGEs such as eukaryotic helitrons, diverse bacterial and archaeal plasmids, as well as diverse small viruses (25–28). Typically, these MGEs possess small circular genomes (or form circular replicative intermediates in the case of the helitrons) and replicate via the rolling-circle (RC) mechanism (27, 29). The indications of a circular intermediate in transpoviron replication (see above) imply that the transpovirons might use a similar replication strategy. The RC replication mechanism relies on key replication proteins that consist of a helicase domain and an RC replication initiator endonuclease domain that typically comprises the N-terminal part of the respective two-domain protein (28). However, extensive sequence-similarity searches including direct comparisons to Rep proteins failed to detect significant similarity or any of the diagnostic motifs (28) in the N-terminal region of the large transpoviron protein.

The second conserved transpoviron gene encodes a C2H2 Zn-finger protein that is a homolog of the V4 and V14 gene products of Sputnik; a more distant homolog of this protein is also encoded by the Mavirus (SI Appendix, Fig. S13). The transpovirons of Lentille virus, Mamavirus and Monve (but not Courdo7), encode another protein with a homolog in Sputnik, a DNA-binding transposase subunit (SI Appendix, Fig. S14). The remaining predicted proteins of the transpovirons do not have detectable homologs. The mechanism of transpoviron replication remains unknown. Nevertheless, the results of the transpoviron gene analysis are compatible with the expected features of a mobile element that partly relies for replication on its own encoded proteins (in particular the predicted helicase of the transpoviron) and partly on proteins of the host (apparently the Mimiviruses that would supply the DNA polymerase and other components of the replication machinery). The presence, in the transpovirons, of two genes with homologs in Sputnik implies that recombination between the two components of the giant virus mobilome contributes to their evolution. Phylogenetic analysis of the transposase subunits clearly supports the common origin of this gene in the transpovirons and Sputnik (SI Appendix, Fig. S14). In contrast, phylogenetic analysis of the helicases places the predicted transpoviron helicase into a branch with a distinct group of bacterial homologs (SI Appendix, Fig. S15). These findings indicate that, like Sputnik and the Mavirus, the transpovirons have a chimeric origin, with different genes derived from distinct sources including genes from other components of the giant mobilome as well as genes of apparent cellular derivation.

In an attempt to identify potential recombination events between transpovirons and Mimiviruses and/or virophages, we searched all of the available Mimivirus sequences for regions homologous to the transpovirons. A short sequence segment homologous to the Monve transpoviron was identified in the Sputnik genome (SI Appendix, Fig. S5D). Although this observation does not reveal evidence of transpoviron recombination with Mimiviruses, taken together with the findings on Sputnik integration into the Mimivirus genome, it suggests that all components of the triple system Mimivirus–virophage–transpoviron are linked through recombination.

Discussion

The discovery of the Mimivirus and subsequent identification of other giant viruses revealed unexpected complexity of viral genomes that, with over 1,000 protein-coding genes, are more complex than many parasitic and symbiotic bacteria and are comparable to the most compact genomes of free-living bacteria and archaea (7). The present work shows that giant viruses are associated with a commensurately complex mobilome that encompasses three of the four major classes of mobile elements, namely self-splicing elements, transposable elements or linear plasmids (transpovirons), and viruses (virophages that can form provirophages after integration into the host giant virus genome). Different components of the giant virus mobilome share homologous genes, and genomic comparisons point to DNA transfer between the mobilome components and the host virus but also within the mobilome itself. Thus, the giant virus mobilome is a network that potentially could provide routes and vehicles for gene exchange and might make substantial contributions to the shaping of mosaic viral genomes. The giant viruses and their mobilomes together are part of even more expansive, dynamic genetic networks: the amoebae with their diverse bacterial parasites and symbionts and their own viruses (30).

Of special note is the transpoviron, a distinct plasmid that depends on giant viruses for its replication and spread. Substantial analogies can be found between the transpovirons and virus-associated plasmids present in bacteria and archaea. In particular, the well-studied bacteriophage P4 (also known as a “phasmid”) is a plasmid that replicates episomally in the absence of the helper bacteriophage P2 but is encapsidated into virions and thus can infect new bacterial cells in the presence of the helper (31, 32). A similar replication strategy has been described for the archaeal virus plasmid pSSVx that depends on the fuselloviruses SSV1 or SSV2 and appears to have acquired genes from a fusellovirus (33). The discovery of the transpoviron shows that virus-associated plasmids exist in all three domains of cellular life.

It is unlikely that the present study exhausts the diversity of the giant virus mobilome; additional virophages and transpovirons, and perhaps distinct classes of mobile elements, are likely to be discovered. Indeed, the transpoviron had not been detected until the isolation of Lentille virus from a human sample described here. Furthermore, we failed to detect closely related homologs of transpoviron genes in the available databases of environmental sequences, although close homologs of many Mimivirus and Sputnik genes were readily detectable (11). Thus, specific conditions and/or habitats could be required for accumulation of transpovirons and probably other elements comprising the giant virus mobilome. Characterization of such conditions will likely lead to the discovery of additional genetic elements associated with giant viruses and facilitate elucidation of their replication mechanisms and the relationships between different mobilome components.

Materials and Methods

Isolation, Culture, and Purification of Viruses.

Lentille virus and Sputnik 2 were isolated from contact lens fluid of a patient with keratitis (19) as previously described (11).

Virus Cloning.

A supernatant containing Sputnik and Lentille virus was submitted to heat inactivation at 68 °C for 1 h or to desiccation for 48 h. The viral suspension was then diluted in PAS by 10-fold dilution from 10⁻¹ to 10⁻¹⁰. Each dilution was inoculated into four wells of a suspension of fresh amoebae and observed daily for lysis under an inverted microscope. The supernatant of the last dilution producing lysis in 1/4 wells was subcultured onto fresh amoebae to produce Lentille virus. Purified Lentille virus particles were examined with a transmission electron microscope (Morgagni 268D; Philips) as previously described (34).

PFGE and Southern Blotting.

Plug preparation and treatment and PFGE of viral DNA were performed as previously described (7). Restriction enzyme (ApaI and EagI; New England BioLabs) digestion of agarose-embedded DNA was performed using 20 U of the restriction enzyme of interest in the appropriate buffer at 37 °C for 4 h. The solution was removed, refreshed, and incubated overnight. The digested agarose blocks and molecular weight markers (Low Range PFG Marker and Lambda Ladder PFG Marker; New England BioLabs) were equilibrated in 0.5× TBE (50 mM Tris, 50 mM boric acid, 1 mM EDTA). Each agarose block was placed in the well of a 1% PFGE agarose gel (Sigma) in 0.5× TBE. The pulsed-field gel separation was made on a CHEF-DR II apparatus (Bio-Rad), with pulses ranging from 5 to 25 s at a voltage of 5 V/cm for 22 h at 14 °C. Gels were either stained with ethidium bromide and analyzed using a Gel-Doc 2000 system (Bio-Rad) or used to prepare Southern blots. Resolved uncut or digested genomic DNA by pulsed-field gel electrophoresis was treated and transferred onto Hybond N+ (GE Healthcare) with a vacuum blotter (model 785; Bio-Rad) and UV–cross-linked for 2 min. The blots were then hybridized against the digoxigenin-labeled probe as recommended by the manufacturer (DIG-System; Roche Diagnostics), except detection of the hybridized probe was performed using a horseradish peroxidase-conjugated monoclonal mouse anti-digoxigenin antibody (Jackson Immunoresearch; 1:20,000). After washings, blots were revealed by chemiluminescence assays (ECL; GE Healthcare). The resulting signal was detected on Hyperfilm ECL (GE Healthcare) by using an automated film processor (Hyperprocessor; GE Healthcare).

Probe Labeling and Fluorescence in Situ Hybridization.

Slides for in situ hybridization were prepared from a culture of A. polyphaga infected with Lentille virus. At 7 h postinfection, a culture sample (200 μL) was placed in a cytospin chamber, centrifuged at 800 rpm for 10 min in a Shandon Cytospin 4 (Thermo Electron), and fixed for 10 min with a drop of methanol. A negative control consisting of uninfected A. polyphaga culture was also performed simultaneously. Slides were treated for 15 min at 37 °C with 0.1 mg/mL RNase A (Boehringer) in 2× SSC. After two washes in PBS, preparations were dehydrated in an ethanol series and air-dried. Specific DNA probes were generated by PCR (SI Appendix, Table S6 and SI Materials and Methods) and labeled using either the DIG- or Biotin-High Prime DNA Labeling Kit (Roche Applied Sciences) following the manufacturer’s recommendations. Labeled probes were ethanol-precipitated and resuspended in a solution containing 50% diformamide, 2× SSC, 10% (wt/vol) dextran sulfate, 50 mM sodium phosphate, pH 7 (final concentration of 1 ng/μL). Probe solution (10 μL) and cell preparation were denatured simultaneously for 4 min at 80 °C and hybridized in a moist chamber at 37 °C overnight. Slides were washed three times for 5 min at 45 °C with 2× SSC containing 50% formamide, five times for 2 min with 2× SSC at room temperature (RT), and once for 5 min with TNT buffer [100 mM Tris⋅HCl (pH 7.5), 150 mM NaCl, 0.05% Tween 20]. The preparation was then incubated for 30 min at 37 °C with TNB buffer [100 mM Tris⋅HCl (pH 7.5), 150 mM NaCl, 0.5% blocking reagent] and 30 min at 37 °C with 2 µg/mL rhodamine anti-DIG antibody (Roche Applied Sciences) and 2 μg/mL of Alexa Fluor 488 streptavidin (Invitrogen) diluted in TNB buffer. Slides were finally washed three times for 5 min with TNT buffer, dehydrated in an ethanol series, air-dried, and mounted with 10 μL of Prolong antifade. Slides were observed under epifluorescence microscopy with a Leica microscope equipped with a DS1-QM (Nikon) black and white camera. Native images were colored and merged using ImageJ free software (35).

Combined Fluorescence in Situ Hybridization and Immunofluorescence.

FISH using a DIG-labeled Sputnik 2 probe was performed from a culture of A. polyphaga infected with Lentille virus at time 0, hour 2, hour 4, hour 6, and hour 23 postinfection. Slides were prepared as described above for FISH. For immunofluorescence, preparations were incubated for 30 min with rabbit anti-Mimivirus serum previously adsorbed on A. polyphaga and washed with PBS-Tween as previously described (11). Lentille virus particles were detected with goat anti-rabbit Ig-Alexa 546, and DIG-labeled Sputnik 2 probe was detected with a FITC anti-DIG antibody.

Genome Sequencing, Bioinformatic Analysis, and Sequence Accession Numbers.

DNA extracted from a clonal preparation of Lentille virus particles was first pyrosequenced with a 454/Roche GS FLX system using shotgun technology and then pyrosequenced with a 454/Roche GS FLX Titanium system using paired-end technology. Raw data for the two sequencing runs are presented in SI Appendix, Table S1. Pyrosequencing reads from shotgun and paired-end technologies were then assembled with Newbler version 2.5.3 (Roche Applied Science) using default parameters either separately (SI Appendix, Tables S2 and S3 for shotgun and paired-end assemblies) or combined (SI Appendix, Table S4). The paired-end links between reads were visualized with the Gephi open graph visualization platform (freely accessible at http://gephi.org) from the parsing of a New454PairEndStatus Newbler log file and using the Force Atlas 2 layout. To gain higher coverage and better finishing in closing the Lentille virus genome, a third sequencing round was performed, using SOLiD V4 chemistry on one full slide associated with 49 other projects on an Applied Biosystems SOLiD 4 machine. The paired-end library was constructed from 1 μg of purified genomic DNA of Lentille virus. All of these 50 genomic DNA were barcoded, with the module 1–96 barcodes provided by Life Technologies. Libraries were pooled in equimolar ratios and emulsion PCR was performed according to Life Technologies specification. Raw data from the Lentille virus project are presented in SI Appendix, Table S1. Reads were mapped on the contigs previously obtained after assembly of the two pyrosequencing runs with 99% identity and 99% coverage using the CLC Genomics workbench from CLC Bio (SI Appendix, Table S5). Broken (i.e., pairs broken with respective orientation or with a distance above 65 nt) mate pairs matching different reference genomes were extracted and visualized using an in-house script written in R (www.r-project.org) (SI Appendix, SI Materials and Methods).

The ORFs in contigs were identified using GeneMark (36) and Prodigal 2.5 (37) and annotated using BLASTN, BLASTX, and BLASTP (38), PFam (39), and tRNAscan (40). For in-depth annotation, the nonredundant protein sequence database and the database of environmental sequences at the National Center for Biotechnology Information (National Institutes of Health) were searched using PSI-BLAST (41). Additional searches of protein family profile databases were performed using the HHpred program (42). Multiple alignments of protein sequence were constructed using the MUSCLE program (43). Maximum-likelihood phylogenetic trees were constructed using the TREEFINDER program (44) [WAG matrix, fixed rate model (45) G[Optimum]:4, 1,000 replicates, Search Depth 20]; the bootstrap support was calculated as expected-likelihood weight. DNA repeats were identified using Inverted Repeats Finder (46).