JOURNAL OF BACTERIOLOGY, Oct. 2011, p. 5412–5419 Vol. 193, No. 190021-9193/11/$12.00 doi:10.1128/JB.05101-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Provirus Induction in Hyperthermophilic Archaea: Characterizationof Aeropyrum pernix Spindle-Shaped Virus 1 and

Aeropyrum pernix Ovoid Virus 1�†Tomohiro Mochizuki,1 Yoshihiko Sako,2 and David Prangishvili1*

Unite Biologie Moleculaire du Gene chez les Extremophiles, Department of Microbiology, Institut Pasteur, 75015 Paris, France,1

and Laboratory of Marine Microbiology, Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, Japan2

Received 15 April 2011/Accepted 28 June 2011

By in silico analysis, we have identified two putative proviruses in the genome of the hyperthermophilicarchaeon Aeropyrum pernix, and under special conditions of A. pernix growth, we were able to induce theirreplication. Both viruses were isolated and characterized. Negatively stained virions of one virus appeared aspleomorphic spindle-shaped particles, 180 to 210 nm by 40 to 55 nm, with tails of heterogeneous lengths in therange of 0 to 300 nm. This virus was named Aeropyrum pernix spindle-shaped virus 1 (APSV1). Negativelystained virions of the other virus appeared as slightly irregular oval particles with one pointed end, while incryo-electron micrographs, the virions had a regular oval shape and uniform size (70 by 55 nm). The virus wasnamed Aeropyrum pernix ovoid virus 1 (APOV1). Both viruses have circular, double-stranded DNA genomes of38,049 bp for APSV1 and 13,769 bp for APOV1. Similarities to proteins of other archaeal viruses were limitedto the integrase and Dna1-like protein. We propose to classify APOV1 into the family Guttaviridae.

Sequencing of diverse bacterial and archaeal genomes hasrevealed the presence of integrated putative viral genomes,their fragments, and viral genes in cellular chromosomes (3, 8,14–16, 28, 32). These results served as a basis for speculationson virus-host relationships and the nature of presumed provi-ruses. For example, in silico analysis allowed us to postulateinfection of methanogenic euryarchaea by viruses related tobacterial or crenarchaeal icosahedral viruses (11) or head-and-tail viruses from the order Caudovirales (12). Traces of inte-gration of viral genomes of members of the families Fusello-viridae and Bicaudaviridae were detected in the chromosomesof hyperthermophilic crenarchaea (8, 23, 25, 31, 32). However,until now, there have been no reports on the induction andisolation of putative integrated archaeal viruses indentified insilico. Moreover, it is unclear whether replication of all theseputative proviruses can be induced at all.

To make matters more difficult, discovery of proviruses by insilico analysis is not straightforward and is often overlooked inprimary genome annotations. For screening genome sequencesfor putative proviruses, conserved viral functions/genes areoften used as markers and the surrounding regions are ana-lyzed for their gene content. However, this approach impliesthat the provirus, in addition to the marker gene, carries otherrecognizable genes of viral origin.

In the case of archaeal genomes, integrase genes are oftenused as markers. On the basis of their sequences and functions,archaeal integrases can be classified into three major catego-ries, the Xer type, the pNOB8 type, and the Sulfolobus spindle-

shaped virus (SSV) type. Enzymes of the Xer type have bac-terial homologues and were recently shown to be involved inresolving chromosome dimers during the cell cycles of Pyro-coccus and Sulfolobus (4, 5). The other two enzymes are ty-rosine recombinases, exclusive to archaea. Both are widelydistributed in archaeal integrative extrachromosomal ele-ments, such as viruses or plasmids, and both target tRNAsequences on cellular chromosomes. The two types are distin-guished by the location of the target sequence. For enzymes ofthe pNOB8 type, the tRNA sequence acting as the attB andattP sites are located adjacent to the integrase gene, whereasfor the enzymes of the SSV type, these sequences are locatedwithin the integrase gene itself. Consequently, after integrationinto the host chromosome, the integrase gene of the SSV typeis split, and its two fragments are found at the boundaries ofthe integrated element, defined as intC and intN. All integrasegenes from known crenarchaeal viruses are of the SSV type.This type of gene is also found on some archaeal plasmids (21,34). The pNOB8-type integrase was found to be encoded onarchaeal plasmids (6) and in putative proviruses of the Euryar-chaeota (11).

Here we report the first example of induction of in silico-identified archaeal proviruses and describe two new viruses ofAeropyrum pernix.

MATERIALS AND METHODS

Search for a putative proviral sequence. Putative proviral sequences in thewhole-genome sequence of A. pernix K1 were identified by searching for inte-grase and tRNA genes. The search for integrase genes was performed byBLASTp and tBLASTx analyses, using integrase gene sequences from archaeaand their viruses as queries. The putative proviral region was determined as aregion between two tRNA-related sequences flanking an integrase gene.

Strain and medium. A. pernix strain K1 was obtained from the DSMZ culturecollection (DSM accession no. 11879). For cell growth, the following media wereused: (i) JXTm (20); (ii) 3ST, consisting of 35 g of sea salts (Sigma) containing1 g Na2S2O3 � 5H2O, 1 g yeast extract (Difco), and 1 g tryptone (Difco) per liter(18); and (iii) MBT, consisting of marine broth 2216 (Difco) containing addi-

* Corresponding author. Mailing address: Molecular Biology of theGene in Extremophiles Unit, Institut Pasteur, rue Dr. Roux 25, 75724Paris Cedex 15, France. Phone: 33-(0)144-38-9119. Fax: 33-(0)145-68-8834. E-mail: david.prangishvili@pasteur.fr.

† Supplemental material for this article may be found at http://jb.asm.org/.

� Published ahead of print on 22 July 2011.

5412

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

tionally 1 g Na2S2O3 � 5H2O per liter (19). The pH value was about 7 and did notchange significantly during the course of cell growth.

Culture conditions. Cells were grown at 90°C aerobically. For small-scalecultures, cells were cultured in 18- by 180-mm hermetically sealed screw-cap testtubes filled with 6 ml of medium under atmospheric conditions. For cultures oflarger scale, 250- or 2,000-ml glass flasks (Fisher Scientific) with Steristoppers(Heinz Herenz, Germany) were filled with 100 or 1,000 ml of medium. In allcases, atmospheric air was used as the gas phase. The degree of aeration wascontrolled by the intensity of shaking of the incubator (0 or 120 rpm).

Virus isolation. Cells were removed by centrifuging them at 5,000 � g for 30min, and virions were collected by either ultracentrifugation at 48,000 rpm in4-ml tubes using an SW60 rotor (Beckman) for 12 h or by polyethylene glycol(PEG) precipitation by addition of NaCl and PEG 6000 to final concentrationsof 1 M and 10%, respectively; after incubation at 4°C overnight, viral particleswere collected by centrifugation at 12,000 � g in 50-ml tubes for 30 min at 4°C,and the particles were suspended in 20 mM Tris-acetate, pH 7.0, containing 3%sodium chloride.

Electron microscopy. For negative staining, samples were applied to glow-discharged carbon-coated copper grids, stained with 2% uranyl acetate for 10 s,and observed under a JEM-1200EXII (JEOL, Japan) transmission electronmicroscope. The images were recorded, using MegaView (SIS, Japan) and iTEMsoftware version 5.0 (SIS, Japan). The sample for cryo-electron microscopyobservation was specially prepared by tangential-flow filtration. Briefly, from 2liters cell culture grown in the MBT medium without shaking, cells were removedby centrifugation at 5,000 � g, followed by filtration with a 0.45-�m filter. Thecell-free supernatant was concentrated using Vivaflow-200 (Sartorius StedimBiotech, France) and Amicon Ultra-15 (nominal molecular mass limit, 100 kDa)(Millipore). The sample was concentrated to 200 �l. Four microliters of thesample was spread on glow-discharged Quantifoil R2/2 grids (Quantifoil MicroTools GmbH, Germany) and cryofixed in liquid ethane. The specimens weretransferred to a Gatan 626 DH cryoholder (Gatan) and examined on a JEOL2010F electron microscope (JEOL, Japan) operating at 200 kV. Images wererecorded under low-dose conditions on a Gatan Ultrascan 4000 with Digital-Micrograph (Gatan) version 1.83.842.

DNA isolation. DNA was extracted from virions as described earlier (30). Inbrief, a virus suspension in 50 mM Tris-HCl (pH 7.5) buffer containing 10 mMMgCl2, in a total volume of 500 �l, was treated with DNase I (Roche) and RNase(Invitrogen) at 37°C for 1 h. The DNase was inactivated at 90°C for 10 min.Virions were disrupted by treatment with 50 �g/ml protease K (Eurobio Labo-ratories) in the presence of 20 mM EDTA and 0.5% sodium dodecyl sulfate(final concentration) at 56°C for 2 h, and DNA was extracted by phenol-chloro-form treatment.

DNA analysis. For PCR amplification of fragments of genomes of Aeropyrumpernix spindle-shaped virus 1 (APSV1) and Aeropyrum pernix ovoid virus 1(APOV1) containing attP sites, different sets of primers were used (Table 1).Sequences of genome fragments of APOV2 were obtained with a sequence-independent primer amplification approach known as the sequence-independentsingle-primer amplification (SISPA) method (1, 26). Briefly, the first step wasperformed using the viral DNA at a final concentration of 0.35 �g/ml and theFR26RV-N primer (5� GCC GGA GCT CTG CAG ATA TCN NNN NN 3�)with 2 PCR cycles with the following parameters: denaturation at 90°C for 30 s,annealing at 45°C for 30 s, and elongation at 72°C for 58 s. The second step wasperformed in a 20-�l assay mixture using 2 �l of the PCR product from the firststep and the FR20RV primer (5� GCC GGA GCT CTG CAG ATA TC 3�)under the following conditions: 40 cycles of denaturation at 90°C for 30 s,annealing at 45°C for 30 s, and elongation at 72°C for 60 s. Both PCR steps wereperformed using Ex Taq polymerase (TaKaRa, Japan). The PCR product was gelpurified and cloned into the Escherichia coli cloning vector pCR-II TOPO (In-vitrogen). The plasmid was extracted from E. coli colonies and submitted forsequence analysis.

Theoretical sizes of restriction digestion fragments derived from selected re-

gions of the A. pernix genome were calculated using REBsites (http://tools.neb.com/REBsites/index.php).

Open reading frames (ORFs) on the viral genomes were predicted by usingGLIMMER version 3.02 and GeneMark.hmm (version 2.6r). For each predictedORF, its validity was confirmed manually by searching for a putative ribosome-binding site about 5 nucleotides (nt) upstream of the start codon. An origin ofreplication (ori) was identified by using Z-curve (http://tubic.tju.edu.cn/zcurve/)and by searching for inverted repeats.

Viral genome sequence accession numbers. The accession numbers of thegenome sequences of APSV1 (38,049 bp) and APOV1 (13,769 bp) were depos-ited in the EMBL database under accession numbers HE580238 and HE580237,respectively.

RESULTS

Aeropyrum pernix K1 integrative elements. In the search forputative proviral sequences, we have analyzed the genomesequence of A. pernix K1, a type strain of the genus Aeropyrum.The observation that the presence in archaeal chromosomes ofan integrase from the tyrosine recombinase family is often aresult of integration of a viral genome (31, 32) served as aguideline. By carrying out a homology search against integrasesequences from Archaea and archaeal viruses, we could iden-tify on the genome of A. pernix three homologous genes forhypothetical integrases, APE_0716, APE_0805, and APE_0818(Fig. 1). Two of the three genes, APE_0716 and APE_0818,were flanked by two sequences which are characteristic ofarchaeal integrative elements. Immediately adjacent upstreamof APE_0818 resides a tRNA sequence (tRNA21-Val) of 78bp, and approximately 38 kb downstream, there is a truncated48-bp sequence with 100% identity with the same tRNA21-Val. Similarly, approximately 13 kb downstream of APE_0716,there is tRNA20-Leu (88 bp), and immediately upstream ofthe integrase is the truncated sequence (64 bp) with 100%identity to the same tRNA. No tRNA-like sequences werefound adjacent to APE_0805, which was homologous to aXerC/XerD-type tyrosine recombinase. The scheme of the de-scribed cellular chromosome region, encompassing two puta-tive proviruses, “alpha” and “beta,” is shown in Fig. 1. Thepresence of putative proviruses gave a good reason for at-tempts at their induction.

Observation of APSV1. Different conditions of culturing ofA. pernix were exploited with the objective to induce provirusreplication. The conditions included three different media,JXTm, 3ST, and MBT (see Materials and Methods), withvarious degrees of aeration. The presence of viruses in cell-freeculture supernatants was monitored by transmission electronmicroscopy (TEM) after their concentration by ultracentrifu-gation. Cell growth was followed by measuring optical densityat 600 nm (OD600).

Without shaking, the cells grew more slowly than with ex-cessive aeration by shaking at 120 rpm (see Fig. S1 in supple-mental material). Moreover, in supernatants of these cultures,we observed the appearance of pleomorphic particles resem-bling virions of archaeal spindle-shaped viruses. The titer ofthese particles was highest in the media 3ST and MBT com-pared to JXTm, where it was estimated to be about 107 parti-cles/ml. When cultures were grown with extensive aeration byshaking, the particles were not detected in any of the media.We presumed that the observed particles represented virionsof a virus induced as a result of the growth of A. pernix under

TABLE 1. PCR primers used in this study

Name SequenceNucleotide positions

in A. pernix K1genome

V1-R CGAGTGGGTAGCTGGTTGTT 542740–542721V1-F CCGCATATGCTTCAGCTACA 580145–580164V2-R CACTCGAGAAACCTGGAAGC 478337–478318V2-F GGGGTGAACCCCTCTTGTAT 491689–491708

VOL. 193, 2011 PROVIRUS INDUCTION IN HYPERTHERMOPHILIC ARCHAEA 5413

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

suboptimal conditions. The virus was named Aeropyrum pernixspindle-shaped virus 1 (APSV1).

Morphology of APSV1. Virions of APSV1 are pleomorphicspindle-shaped particles, �180 to 210 nm in length and �40 to55 nm in width (Fig. 2a). Part of the virions in the purifiedpreparation carried a single tail or two tails protruding frompointed ends (Fig. 2b to d). Among the 444 analyzed particles,the tailed particles comprised about 25%. The majority ofthese tailed particles, about 67%, carried a single tail rangingin length from about 20 to 150 nm; a smaller proportion, 33%,was two tailed, with the total particle length reaching about 700nm. Three short filaments were attached to one end of theparticles, tailless as well as single tailed (Fig. 2c, inset). Thevirions have a tendency to attach to each other with theirfilament-carrying ends, sometimes forming rosette-like struc-tures (Fig. 2e). The incubation of a purified viral preparation atdifferent temperatures in the range of 75 to 90°C up to 7 daysdid not affect the ratios of the tailless, single-tailed, and two-tailed virions, either in Tris-HCl buffer at pH 7 or in the culturemedium (3ST).

Determination of the APSV1 genome sequence. Nucleic acidextracted from the purified APSV1 virions was digested with

several type II restriction enzymes, indicating that it was dou-ble-stranded DNA (dsDNA) (Fig. 3a). The pattern of diges-tion fragments showed stoichiometric distribution, suggestingthat DNA originated from a single viral strain (Fig. 3a). Basedon the lengths of the digestion fragments, the size of the viralDNA was estimated to be about 40 kb.

In order to find out whether viral DNA corresponds to anyof the putative proviruses of A. pernix, the sizes of fragmentsobtained by digestion of the viral DNA with BamHI, EcoRI,SalI, and XhoI (Fig. 3a) were compared to the calculated sizesof fragments of analyzed regions of the genome, shown in Fig.1 (Fig. 3b). One of these regions, about 110 kb long, extendedfrom about 55 kb upstream of APE_0818 until about 55 kbdownstream of this gene. By comparing the lengths of theo-retical and actual DNA fragments, the analyzed sequencecould be trimmed down to about 40 kb, exactly to the segmentcorresponding to the putative “beta” provirus.

To verify that the sequence of the APSV1 genome is iden-tical to the putative beta provirus of A. pernix, PCR analysiswas performed. The primers V1F and V1R were designedbased on the sequences bordering the beta provirus, facingoutwards (Fig. 1). Using APSV1 DNA as a template, this

FIG. 1. Scheme of the fragment of the A. pernix K1 genome containing three putative proviruses. Integrase genes (large black arrows), tRNAgenes, and truncated tRNAs (gray) are indicated. The sites and orientations of the primer pairs which were used for sequence analysis are indicatedwith small arrows beneath the diagrams.

FIG. 2. Transmission electron micrographs of APSV1 virions. Scale bars � 100 nm. Samples were negatively stained with 2% uranyl ace-tate.

5414 MOCHIZUKI ET AL. J. BACTERIOL.

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

primer pair produced a DNA fragment of about 700 bp, ap-parently from the excised and circularized genome fragment.The sequence of the first 547 bp was identical to an A. pernixsequence fragment from positions 580187 to 580733. The se-quence of the last 61 bp was identical to an A. pernix sequencefragment from positions 542635 to 542696 and had an overlapof 50 bp with the first part (see Fig. S2 in supplemental mate-rial). The results confirmed that APSV1 has a circular genomederived from the excision and circularization of the provirusbeta region of the A. pernix genome (Fig. 1). They also indi-cated that the viral genome contains only 50 bp from the tRNA21-Val gene as a part of the integrase gene (Fig. 4). Moreover,based on these results and on the available genome sequenceof A. pernix (9), the sequence of the APSV1 genome could beasserted.

Analysis of the APSV1 genome sequence. The genome ofAPSV1 is a circular dsDNA containing 38,049 bp. The GCcontent (56.51%) is in the range of that of the A. pernix chro-mosome (56.3%) but higher than that of the chromosomalregions bordering the provirus. The ORF of 1,566 bp codingfor the integrase is split as a result of integration into the hostchromosome in two parts. The first 118 bp, including the startcodon, is present at the right border of the proviral region, andthe last 1,498 bp, including the termination codon at the leftborder, has an overlap of 50 bp, as illustrated in Fig. 1.

In the genome, 53 ORFs encoding hypothetical proteinswith 38 to 929 amino acids, which start from the ATG, GTG,and TTG codons, could be identified. In most cases, they werepreceded by recognizable putative ribosome-binding sites (seeTable S1 in the supplemental material). The majority of puta-tive genes were present on one DNA strand, and only sixputative genes were on the other. The genome map of APSV1is illustrated in Fig. 4. The numbering of the nucleotides in thegenome sequence starts from the start codon of the first ORFfollowing the integrase gene. The ORFs are numbered corre-spondingly, and the number following the sequential numberrefers to the number of amino acids in the predicted protein.

Of the predicted ORFs, nine showed significant (E value �0.001) sequence similarity with sequences in the public data-bases, and two of them had homologues in the genomes of

other archaeal viruses (Table 2). Of note, the integrase gene ofAPSV1, ORF53-521, showed the highest sequence similarity tointegrase genes in the chromosomes of Sulfolobales (i.e., Sul-folobus islandicus).

The sequences of three consecutive genes, ORF38-469,ORF39-350, and ORF40-223, showed exceptionally high se-quence similarity with the sequences in the public database(Table 2, and see Table S2 in supplemental material). Se-quence analysis allowed us to predict that ORF38-470 is amethylase and that ORF40-223 is a DNA glycosylase, a mem-ber of a family of enzymes involved in DNA mismatch repair.The function of ORF39-350 could not be predicted by homol-ogy searches. These three genes are present in the same ori-entation as that of blocks in the genomes of two other hyper-thermophilic members of the Crenarchaeota, Pyrobaculumaerophilum and Thermoproteus neutrophilus. In each case, thehigh similarity can be detected at both the amino acid andnucleotide sequence levels. ORF38-469 and ORF40-223 havehomologues, although with lower sequence similarity, in ge-nomes of different members of the three domains. ORF38-469has homologues also in various bacterial and eukaryal virusesand one euryarchaeal virus.

The search for the origin of replication was performed byusing Z-curve and by locating inverted-repeat sequences. TheZ� component of the Z-curve shows four characteristic peaksin the APSV1 genome, about 2 kb, 6 kb, 25 kb, and 34 kb awayfrom nucleotide position 1. Inverted repeats were searched,and a region containing a perfect repeat of 18 bp, 5�-CTATCACCTATACTCTAG(N)45CTAGAGTATAGGTGATAG-3�,was found, starting at nucleotide position 34121 and ending atposition 34202. This region between ORF44-94 (negativestrand) and ORF45-81 (positive strand) is a noncoding regionand is likely to represent the origin of replication of the viralgenome (Fig. 4). No homologue of ORF44-94 was found in

FIG. 3. (a) Restriction enzyme digestion pattern of the APSV1genome. Agarose gel electrophoresis of the APSV1 genome digestedwith BamHI, EcoRI, SalI, and XhoI. (b) Theoretical in silico digestionpattern of a fragment of the sequence of the A. pernix K1 genome fromnt 542607 to 580733, digested with BamHI, EcoRI, SalI, and XhoI.

FIG. 4. Genome map of APSV1. The integrase gene is white. oriand attP sites are indicated.

VOL. 193, 2011 PROVIRUS INDUCTION IN HYPERTHERMOPHILIC ARCHAEA 5415

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

databases, and ORF45-81 showed low sequence similarity tothe transcriptional regulator of Brevibacterium (E-value,0.008).

Morphology of APOV1. After continuous culturing of A.pernix in 3ST or MBT medium without extensive aeration, weoccasionally detected in the culture the presence of virus-likeparticles, different in morphology from virions of APSV1 (Fig.5a). Their titer was comparable to that of APSV1 and wasestimated to be about 107 particles/ml. These particles wereoval and slightly pleomorphic and measured about 80 by 60 nm(Fig. 5b). They were often found in clusters and sometimesshowed a concave morphology resembling that of beans (Fig.5b, inset). No surface structures or tails could be observed.Cryo-electron microscopy analysis enabled better understand-ing of particle morphology. Particles that were embedded invitreous ice were of regular ovoid shape and size, measuring 70by 55 nm (Fig. 5c). Globular subunits about 3.5 nm in widthcould be distinguished as building blocks of the surface layer.The structure was clearly different from that of APSV1 virions,and we presumed that the particles represented virions ofanother induced virus of A. pernix, which we termed Aeropyrumpernix ovoid virus 1 (APOV1).

Determination of the APOV1 genome sequence. Unfortu-nately, we were unable to control the replication of the virusesand could not obtain a large-scale culture producing solely thevirions of APOV1. In attempts to identify the origin ofAPOV1’s genome, it was important to discover an A. pernixstrain, YKP1, which was producing only one type of virus,referred to here as Aeropyrum pernix ovoid virus 2 (APOV2),highly similar in morphology to virions of APOV1 (see Fig. S3

in the supplemental material). Virions of APOV2 were col-lected, and DNA was extracted from them. It could be digestedwith several types of type II restriction enzymes, confirmingthat it was dsDNA (data not shown). The sizes of DNA frag-ments did not clearly match the calculated sizes of fragmentsfrom the two potential proviruses shown in Fig. 1. Applying theSISPA method with random primers, we were able to PCRamplify 6 segments of the DNA, with the lengths ranging from310 to 900 bp. Five of them revealed over 93% sequencesimilarity to different regions of the A. pernix K1 genome, butall were within positions 488171 to 491258 (Fig. 1). The resultsstrongly suggested that the genome sequence of APOV2 ishighly similar to the alpha provirus region (Fig. 1). This, in itsturn, suggested that APOV1 represents an induced alpha pro-virus of A. pernix K1.

In order to verify that the sequence of APOV1 DNA isidentical to that of the alpha provirus, PCR analysis was per-formed. The primers, V2F and V2R, were designed based onthe sequences bordering the putative proviral region, facingoutwards (Fig. 1). When APOV1 DNA served as a template,the PCR product of about 400 bp was produced. The sequenceof the first 250 bp was identical to the alpha provirus sequencefrom positions 491733 to 491982, and the sequence of the last189 bp was identical to the alpha provirus sequence frompositions 478149 to 478337 (see Fig. S4 in the supplementalmaterial). The sequences had identical sequences of 65 bp attwo ends the of alpha provirus region, at positions 491918 to491982 and 478149 to 478213 (see Fig. S3 in the supplementalmaterial).

The results confirmed the origin of APOV1 from the alpha

TABLE 2. Significant database matches to APSV1 ORFs

ORF Best match Organism No. of aminoacids aligned

No. of identical/no.of positive amino

acidsE value

4-235 Hypothetical protein Sulfolobus tokodaii 201 69/109 3e�1912-290 DnaA-like protein ASV1 210 64/113 9e�1626-534 Hypothetical protein ASV1 235 68/99 1e�1231-111 Hypothetical protein Methanocaldococcus vulcanius 85 49/64 6e�1638-469 DNA-cytosine methyltransferase Thermoproteus neutrophilus 429 132/216 4e�17539-350 Hypothetical protein Thermoproteus neutrophilus 349 199/255 2e�11140-223 U/G and T/G mismatch-specific glycosylase Pyrobaculum aerophilum 211 127/169 6e�7443-534 Signal recognition particle Saccoglossus kowalevskii 198 45/90 3e�0453-521 Integrase family protein Sulfolobus islandicus 429 132/216 4e�39

FIG. 5. Transmission electron micrographs of virions of APOV1 (a to c) with APSV1 (a). (a) Black arrowheads indicate APOV1, and whitearrowheads indicate APSV1. (a and b) Samples were negatively stained with 2% uranyl acetate. (c) Sample embedded in vitreous ice. Scale bars,1,000 nm (a), 500 nm and 100 nm in the insets (b), and 100 nm (c).

5416 MOCHIZUKI ET AL. J. BACTERIOL.

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

provirus region of the A. pernix chromosome by its excision andcircularization, which also indicated that the circular viral ge-nome contains only 65 bp from the tRNA20-Leu gene as a partof the integrase gene (Fig. 6). Based on these results and onthe available genome sequence of A. pernix (9), the sequence ofthe APOV1 genome could be asserted.

Genome of APOV1. The genome of APOV1 is circulardsDNA containing 13,769 bp. The GC content (56.51%) was inthe range of that of the whole host chromosome (56.3%). Theintegrase gene (ORF21-400) was split as a result of integrationinto the host chromosome in two parts; the first 298 bp, in-cluding the start codon, was located at the right border of theproviral region, and the last 967 bp, including the stop codon,was located on the left border, with an overlap of 65 bp.

Twenty-one ORFs starting from ATG, GTG, and TTGcodons for hypothetical proteins with 56 to 678 amino acidscould be identified on the genome. In most cases, they werepreceded by recognizable putative ribosome-binding sites (seeTable S3 in the supplemental material). The genome map ofAPOV1 is illustrated in Fig. 6. The numbering of the nucleo-tides in the genome sequence starts from the stop codon of thefirst ORF following the integrase. The majority of putativegenes were located on one DNA strand; only seven were on the

other. Five of these were clustered together, from ORF17-201to ORF21-399 (Fig. 6).

Of the 21 predicted ORFs, 5 showed sequence similarity tosequences in the public database (Table 3). In two cases, thesimilarity was with genes of members of the Fuselloviridaeencoding an integrase and a DnaA-like protein. In other cases,the similarities were with putative genes of hyperthermophiliccrenarchaeotes.

We were unable to determine the origin of replication ofAPOV1 by using Z-curve and locating inverted-repeat se-quences, the approach which was successful in the case ofAPSV1 (see above) and Aeropyrum pernix bacilliform virus 1(APBV1) (18).

DISCUSSION

In our attempts to characterize viruses of the genus Aeropy-rum, we searched for putative proviruses in the genome of thetype strain of the genus A. pernix K1, having in mind to inducetheir replication. Earlier in silico analysis reported the pres-ence of two integrase genes and a possible integrative element,APE1, in the genome of A. pernix (31, 32). In this study, wereanalyzed the A. pernix genome and revealed two putativeproviruses, named alpha and beta (Fig. 1), bearing featurestypical of archaeal integrated elements and having an integrasegene flanked by tRNA-related sequences. Although the sizesdo not exactly correspond to those of the putative integrativeelements reported earlier, we presume that the alpha proviruscorresponds to the integrative element APE1 (31).

Traditionally, to induce integrated viruses, cells are treatedwith stress-causing factors, such as UV irradiation or mitomy-cin C (17). Due to the specific growth conditions of A. pernix(90°C and difficulties in culturing on solid media), we avoidedusing such methods and attempted to induce virus replicationby modifying growth conditions. Members of the genus Aero-pyrum are known to be obligate aerobes (29). Therefore, it hasbeen recommended that they be grown under extensive aera-tion, e.g., by vigorous shaking of cultures (9, 36). We hypoth-esized that for such an organism, a decrease in oxygen acces-sibility can be a stress factor. Indeed, limited oxygen supply inthe course of culturing without shaking caused induction ofprovirus replication.

Two virus species with different morphotypes and genomeswere found to be produced by the “stressed” cells of A. pernix.Their titer in the growth cultures reached about 107 particles/ml. One virus, named Aeropyrum pernix spindle-shaped virus 1

FIG. 6. Genome map of APOV1. The integrase gene is white, andthe attP site is indicated.

TABLE 3. Significant database matches to APOV1 ORFs

ORF Best match Organism No. of aminoacids aligned

No. of identical/no.of similar amino

acidsE value

4-81 CRISPR Csa2 family protein Staphylothermus marinus 70 32/42 1e�0510-284 Hypothetical Ape_0536.1 Aeropyrum pernix K1 267 149/181 1e�73

Endo-alpha-mannosidase Chryseobacterium gleum 258 59/112 7e�0714-121 Hypothetical protein Pyrobaculum aerophilum 120 53/67 6e�1319-247 Hypothetical protein SSV1 194 59/93 1e�08

DnaA-like protein SSV7 134 35/58 0.00321-399 Hypothetical protein Archaeoglobus profundus 179 62/94 4e�20

SSV1 integrase homologue Thermococcus kodakaraensis 263 78/123 1e�17

VOL. 193, 2011 PROVIRUS INDUCTION IN HYPERTHERMOPHILIC ARCHAEA 5417

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

(APSV1), had a genome that originated from the beta provirusregion of A. pernix, and the genome of the other, named Aero-pyrum pernix ovoid virus 1 (APOV1), originated from the alphaprovirus region. The possibility that we might observe stochas-tic variation in spontaneous induction could be excluded. First,we could never observe the presence of virus-like particles incell cultures growing under standard conditions. Second, noextrachromosomal elements were detected in cells of A. pernixK1, which were grown under standard conditions (9). Third, novirion proteins were present in the proteome of A. pernix K1grown under standard conditions (36). These data unambigu-ously indicate that we observed virus induction from the zerolevel to a titer of about 107 particles/ml. The results provide thefirst example of induction of in silico-identified putative inte-grated viruses in the Archaea. Unfortunately, factors whichcould selectively trigger the induction of APSV1 or APOV1could not be determined.

Aeropyrum pernix spindle-shaped virus 1. The spindle-shaped morphotype of APSV1 is common for archaeal virusesinfecting diverse hyperthermophilic and hyperhalophilic hosts,including members of the genera Sulfolobus, Acidianus, Pyro-coccus, and Haloarcula (22, 23), which are abundant in ar-chaeon-rich geothermal and hypersaline environments (24, 27,33, 37). The known species have been classified in the familiesFuselloviridae and Bicaudaviridae and in the genus Salterprovi-rus, while some remain unclassified. Morphologically, APSV1is closest to the unclassified virus Sulfolobus tengchongensisspindle-shaped virus 1 (STSV1) (35). The spindle-shaped viri-ons of the two viruses carry fibers which can stick to each otherto form rosette-like structures (Fig. 2d). Virions of both virusescan adopt a tadpole shape, resulting from elongation of thevirion from one of the pointed ends; the protrusions from bothends were seldom able to be observed and resembled those ofthe Acidianus two-tailed virus (ATV) (7). In the last case, thetails protrude from both pointed ends of the spindle-shapedimmature virion after its release from a host cell, specifically attemperatures above 75°C, similar to those of the natural envi-ronment (7). A population of ATV virions often reveals tailpolymorphism, reflecting different stages of extracellular mor-phogenesis. Whether the tail polymorphisms of APSV1 andSTSV1 have similar origins is still unclear, since we were un-able to affect tail lengths in a purified virion preparation.

The genome of APSV1 was identified in two steps. The firststep was a comparison of the lengths of restriction digestionfragments of the virus DNA with those estimated theoreticallyfrom the nucleotide sequence of the provirus region (betaregion in Fig. 1). The second step was aimed at determiningthe positions of the two ends of the identified proviral se-quence in the viral genome and confirming the circularizationof the excised proviral sequence into a 38,049-bp-long viralgenome. As with genomes of all crenarchaeal viruses with aspindle-shaped morphology, the genome of APSV1 is circular.In the proviral state, the integrase gene is split in two parts, atthe position of the 30th amino acid from the N terminus. Thus,the enzyme is a member of the SSV-type integrase family ofarchaeal viruses, which are split as a result of integration intothe host chromosome (31). However, the amino acid sequenceof the APSV1 integrase gene shows limited similarity to thoseof the integrases from archaeal viruses; the sequence similarityis higher with some cellular integrases (Table 2). However,

these presumably cellular enzymes could also have a viral or-igin.

The presence of the cluster of three putative genes, ORF38-469 (for putative methylase), ORF39-350 (unknown function),and ORF40-223 (for a DNA mismatch repair DNA glyco-sylase), in genomes of the crenarchaea P. aerophilum and T.neutrophilus is remarkable. The extremely high similarity evenat the nucleotide sequence level suggests a recent horizontaltransfer of this cluster among hyperthermophilic archaeal spe-cies which may share a habitat. The fact that these genes aretransferred and maintained together suggests related func-tions. Examples of simultaneous transfer of functionally re-lated genes as a single unit are known from restriction-modi-fication systems between Bacteria and Archaea through theirviruses (13).

Although virions of APSV1 strongly resemble in theirmorphology the virions of STSV1, the lack of any significantsimilarity at the genomic level precludes classification of thetwo viruses in the same taxonomical unit. Additional infor-mation, mainly on the tertiary structures of virion proteins,would be required for revealing evolutionary relationshipsbetween these two unclassified viruses. According to therecently suggested scheme for the nomenclature of virusesof Bacteria and Archaea (10), the full name of the viruswould be vA_Ape“Un”_APSV1 (“Un” is an proposed ab-breviation for unclassified family).

Aeropyrum pernix ovoid virus 1. On negative-contrast elec-tron micrographs, the virions of APOV appeared as pleomor-phic ovoid particles, sometimes resembling droplets (Fig. 5).Cryo-electron microscopy, avoiding osmotic artifacts caused bynegative staining, enabled better understanding of the virionmorphology. The virions embedded in amorphous ice wereregular ovals of uniform size, 70 by 55 nm. They bear clearresemblance to virions of the Sulfolobus newzealandicus drop-let-shaped virus (SNDV), a member of the family Guttaviridae(2). SNDV appears as pleomorphic, mostly droplet-shapedparticles on negative-contrast electron micrographs, and itsmorphological similarity to APOV1 virions is more pro-nounced when cryo-electron micrographs are compared. Themajor structural difference between ovoid virions of APOV1and SNDV appears to be an absence of any fibers attached tothe former virion, whereas they are abundant at one end of theSNDV virion. Additionally, SNDV virions are about 1.5 timeslarger than APOV1 virions.

For the analysis of APOV1 DNA, it was helpful to isolatethe A. pernix strain YKP1, which replicated only one type ofvirus morphologically highly similar to APOV1 (see Fig. S3 inthe supplemental material). PCR products produced by ran-dom primers from the DNA of this APOV1-like virus werehighly similar in sequence to fragments from the putative pro-virus alpha region of the A. pernix K1 genome, strongly sug-gesting that APOV1 represents an induced alpha provirus (Fig.1). This suggestion was verified by determining the locationswhere two boundaries of the putative provirus would circular-ize the viral genome, in the course of excision. The resultsallowed us to conclude that APOV1 represents an inducedalpha provirus and to determine that its genome sequenceconstitutes 13,769 bp.

Based on the available sequence data, it is not possible tounequivocally assign APOV1 to any known viral family. How-

5418 MOCHIZUKI ET AL. J. BACTERIOL.

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

ever, morphologically, the virion shows pronounced similarityto the virion of SNDV, the sole member of the family Gutta-viridae. There is no possibility to claim further similaritiesbetween the viruses. The available information on the SNDV isrestricted to the original description of the virion and a super-ficial characterization of the genome, which was reported to bea circular dsDNA containing about 20 kb (2). SNDV has notbeen characterized further since the original description, and itis impossible to do this at present due to the absence of thevirus in strain collections. Considering the similarities of thevirion morphotypes of APOV1 and SNDV, as well as the factthat the genomes of both viruses are circular, double-strandedDNA molecules of comparable sizes, we propose to assignAPOV1 to the family Guttaviridae and, moreover, to considerit the new type species of the family. Such taxonomical classi-fication will enable researchers to populate the Guttaviridaewith an available member, allowing further characterization ofthe family. According to the recently suggested scheme forthe nomenclature of viruses of the Bacteria and Archaea(10), the full name of the virus would be vA_ApeU_APOV1(“U” is the abbreviation for Guttaviridae).

Conclusions. It has been over 15 years since the genomesof the Archaea began to be sequenced, and presently, over100 archaeal genome sequences are available, many ofwhich contain putative proviral sequences. Nonetheless, noattempts have been made to induce the replication of thesecryptic archaeal viruses. The present study is the first toattempt this, demonstrating how easily, using a simple tech-nique, new viruses can be isolated and described from well-known organisms, and may serve as an example for furtheranalogous studies.

ACKNOWLEDGMENTS

We thank Patrick Forterre and Mart Krupovic for helpful discus-sions, Yutaka Kawarabayasi for helping us with viral genome se-quences, and Gerard Pehau-Arnaudet for taking cryo-electron micro-graphs.

T.M. was supported by a grant of the Gouvernement Francais (dos-sier no. 2008661). This work was partially supported by a grant fromthe Agence Nationale de la Recherche, France (Program Blanc).

REFERENCES

1. Allander, T., S. U. Emerson, R. E. Engle, R. H. Purcell, and J. Bukh. 2001.A virus discovery method incorporating DNase treatment and its applicationto the identification of two bovine parvovirus species. Proc. Natl. Acad. Sci.U. S. A. 98:11609–11614.

2. Arnold, H. P., U. Ziese, and W. Zillig. 2000. SNDV, a novel virus of theextremely thermophilic and acidophilic archaeon Sulfolobus. Virology 272:409–416.

3. Cortez, D., P. Forterre, and S. Gribaldo. 2009. A hidden reservoir of inte-grative elements is the major source of recently acquired foreign genes andORFans in archaeal and bacterial genomes. Genome Biol. 10:R65.

4. Cortez, D., et al. 2010. Evidence for a Xer/dif system for chromosomeresolution in archaea. PLoS Genet. 6:e1001166.

5. Duggin, I. G., N. Dubarry, and S. D. Bell. 2011. Replication termination andchromosome dimer resolution in the archaeon Sulfolobus solfataricus.EMBO J. 30:145–153.

6. Greve, B., S. Jensen, K. Brugger, W. Zillig, and R. A. Garrett. 2004. Genomiccomparison of archaeal conjugative plasmids from Sulfolobus. Archaea1:231–239.

7. Haring, M., et al. 2005. Virology: independent virus development outside ahost. Nature 436:1101–1102.

8. Held, N. L., and R. J. Whitaker. 2009. Viral biogeography revealed bysignatures in Sulfolobus islandicus genomes. Environ. Microbiol. 11:457–466.

9. Kawarabayasi, Y., et al. 1999. Complete genome sequence of an aerobic

hyper-thermophilic crenarchaeon, Aeropyrum pernix K1. DNA Res. 6:83–101, 145–152.

10. Kropinski, A. M., D. Prangishvili, and R. Lavigne. 2009. Position paper: thecreation of a rational scheme for the nomenclature of viruses of Bacteria andArchaea. Environ. Microbiol. 11:2775–2777.

11. Krupovic, M., and D. H. Bamford. 2008. Archaeal proviruses TKV4 andMVV extend the PRD1-adenovirus lineage to the phylum Euryarchaeota.Virology 375:292–300.

12. Krupovic, M., P. Forterre, and D. H. Bamford. 2010. Comparative analysis ofthe mosaic genomes of tailed archaeal viruses and proviruses suggests com-mon themes for virion architecture and assembly with tailed viruses ofbacteria. J. Mol. Biol. 397:144–160.

13. Krupovic, M., S. Gribaldo, D. H. Bamford, and P. Forterre. 2010. Theevolutionary history of archaeal MCM helicases: a case study of verticalevolution combined with hitchhiking of mobile genetic elements. Mol. Biol.Evol. 27:2716–2732.

14. Kunst, F., et al. 1997. The complete genome sequence of the gram-positivebacterium Bacillus subtilis. Nature 390:249–256.

15. Leahy, S. C., et al. 2010. The genome sequence of the rumen methanogenMethanobrevibacter ruminantium reveals new possibilities for controllingruminant methane emissions. PLoS One 5:e8926.

16. Makino, S., N. Amano, H. Koike, and M. Suzuki. 1999. Prophages insertedin archaebacterial genomes. Proc. Jpn. Acad. Ser. B Phys. Biol. Sci. 75:166–171.

17. Martin, A., et al. 1984. SAV 1, a temperate u.v.-inducible DNA virus-likeparticle from the archaebacterium Sulfolobus acidocaldarius isolate B12.EMBO J. 3:2165–2168.

18. Mochizuki, T., et al. 2010. Diversity of viruses of the hyperthermophilicarchaeal genus Aeropyrum, and isolation of the Aeropyrum pernix bacilli-form virus 1, APBV1, the first representative of the family Clavaviridae.Virology 402:347–354.

19. Nakagawa, S., K. Takai, K. Horikoshi, and Y. Sako. 2004. Aeropyrum caminisp. nov., a strictly aerobic, hyperthermophilic archaeon from a deep-seahydrothermal vent chimney. Int. J. Syst. Evol. Microbiol. 54:329–335.

20. Nomura, N., et al. 2002. Heterogeneous yet similar introns reside in identicalpositions of the rRNA genes in natural isolates of the archaeon Aeropyrumpernix. Gene 295:43–50.

21. Peng, X. 2008. Evidence for the horizontal transfer of an integrase gene froma fusellovirus to a pRN-like plasmid within a single strain of Sulfolobus andthe implications for plasmid survival. Microbiology 154:383–391.

22. Pina, M., A. Bize, P. Forterre, and D. Prangishvili. 13 May 2011. Thearcheoviruses. FEMS Microbiol. Rev. doi:10.1111/j.1574-6976.2011.00280.x.[Epub ahead of print.]

23. Prangishvili, D., P. Forterre, and R. A. Garrett. 2006. Viruses of the Ar-chaea: a unifying view. Nat. Rev. Microbiol. 4:837–848.

24. Rachel, R., et al. 2002. Remarkable morphological diversity of viruses andvirus-like particles in hot terrestrial environments. Arch. Virol. 147:2419–2429.

25. Redder, P., et al. 2009. Four newly isolated fuselloviruses from extremegeothermal environments reveal unusual morphologies and a possible inter-viral recombination mechanism. Environ. Microbiol. 11:2849–2862.

26. Reyes, G. R., and J. P. Kim. 1991. Sequence-independent, single-primeramplification (SISPA) of complex DNA populations. Mol. Cell. Probes5:473–481.

27. Rice, G., et al. 2001. Viruses from extreme thermal environments. Proc. Natl.Acad. Sci. U. S. A. 98:13341–13345.

28. Roy, P. H., et al. 2010. Complete genome sequence of the multiresistanttaxonomic outlier Pseudomonas aeruginosa PA7. PLoS One 5:e8842.

29. Sako, Y., et al. 1996. Aeropyrum pernix gen. nov., sp. nov., a novel aerobichyperthermophilic archaeon growing at temperatures up to 100 degrees C.Int. J. Syst. Bacteriol. 46:1070–1077.

30. Sambrook, J., and D. W. Russell. 2001. Molecular cloning: a laboratorymanual, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Har-bor, NY.

31. She, Q., K. Brugger, and L. Chen. 2002. Archaeal integrative genetic ele-ments and their impact on genome evolution. Res. Microbiol. 153:325–332.

32. She, Q., X. Peng, W. Zillig, and R. A. Garrett. 2001. Gene capture in archaealchromosomes. Nature 409:478.

33. Sime-Ngando, T., et al. 25 August 2010. Diversity of virus-host systems inhypersaline Lake Retba, Senegal. Environ. Microbiol. doi:10.1111/j.1462-2920.2010.02323.x. [Epub ahead of print.]

34. Soler, N., et al. 2010. Two novel families of plasmids from hyperthermophilicarchaea encoding new families of replication proteins. Nucleic Acids Res.38:5088–5104.

35. Xiang, X., et al. 2005. Sulfolobus tengchongensis spindle-shaped virusSTSV1: virus-host interactions and genomic features. J. Virol. 79:8677–8686.

36. Yamazaki, S., et al. 2006. Proteome analysis of an aerobic hyperthermophiliccrenarchaeon, Aeropyrum pernix K1. Mol. Cell. Proteomics 5:811–823.

37. Zillig, W., et al. 1996. Viruses, plasmids and other genetic elements ofthermophilic and hyperthermophilic Archaea. FEMS Microbiol. Rev. 18:225–236.

VOL. 193, 2011 PROVIRUS INDUCTION IN HYPERTHERMOPHILIC ARCHAEA 5419

on Novem

ber 17, 2018 by guesthttp://jb.asm

.org/D

ownloaded from