PH1: An Archaeovirus of Haloarcula hispanica Related to SH1 and HHIV-2

Halovirus PH1 infects Haloarcula hispanica and was isolated from an Australian salt lake. The burst size in single-step growth conditions was 50–100 PFU/cell, but cell density did not decrease until well after the rise (4–6 hr p.i.), indicating that the virus could exit without cell lysis. Virions were round, 51 nm in diameter, displayed a layered capsid structure, and were sensitive to chloroform and lowered salt concentration. The genome is linear dsDNA, 28,064 bp in length, with 337 bp terminal repeats and terminal proteins, and could transfect haloarchaeal species belonging to five different genera. The genome is predicted to carry 49 ORFs, including those for structural proteins, several of which were identified by mass spectroscopy. The close similarity of PH1 to SH1 (74% nucleotide identity) allowed a detailed description and analysis of the differences (divergent regions) between the two genomes, including the detection of repeat-mediated deletions. The relationship of SH1-like and pleolipoviruses to previously described genomic loci of virus and plasmid-related elements (ViPREs) of haloarchaea revealed an extensive level of recombination between the known haloviruses. PH1 is a member of the same virus group as SH1 and HHIV-2, and we propose the name halosphaerovirus to accommodate these viruses.


Introduction
Viruses of Archaea (archaeoviruses [1]) show considerable diversity and encompass novel morphotypes not seen in Bacteria or Eukarya. Relatively few have been examined in detail, partly because of the demanding growth requirements of many extremophilic Archaea (particularly thermophiles), and also because genetic analysis is often technically difficult compared to bacterial systems such as Escherichia coli. Although viruses of thermophilic Archaea show the most innovative capsids and replication strategies [1], the viruses of halophilic Archaea (haloarchaea) are of increasing attention as new isolates are found with unexpected properties. Many of the earliest reported haloviruses, including all those described before 1998, are bacteriophage-like (Caudovirales) with typical head-tail capsids and linear dsDNA genomes. These include groups of related viruses, such as the ΦH-like genus (ΦH, ΦCh1, and BJ1) [2][3][4] and the unassigned virus group comprising of HF1 and HF2 [5]. The first spindle-shaped halovirus, His1, was reported in 1998 [6,7], and the first round virus, SH1, was described in 2003 [8,9], and electron microscopic studies indicate that these morphotypes dominate in natural waters [10][11][12]. Over the last 5 years, there has been a wonderful increase in the number and types of described haloviruses, including further examples of SH1-like viruses (e.g., HHIV-2 [13]) and a range of His2related viruses that are now classified as pleolipoviruses (e.g., HRPV-1 and HHPV-1 [14,15]). One example of the biological novelty displayed by these archaeoviruses is the geometry of the SH1 capsid, which was found to be of a previously undescribed type, = 28 dextro [16].
Halovirus His2 has a 16 kb dsDNA genome with terminal proteins and probably replicates via an encoded proteinprimed DNA polymerase [7]. It is now known to be a pleomorphic virus, distinct from the spindle-shaped His1 [17]. When the genome sequence was first described, it was shown to be related to a cryptic plasmid (pHK2) of Haloferax lucentense [18,19] and to a number of genomic loci found in Archaea several different haloarchaea. Later descriptions of haloviruses HRPV-1, HHPV-1, and several others revealed a spectrum of related viruses with very different genome structures (circular ds-and ssDNA), lengths, and replication strategies [14,15], but they all share a similar set of capsid proteins. The mechanisms underlying the movement of the capsid genes between viruses having such different characteristics and modes of replication (protein primed versus rolling circle) remain unclear. One possibility is that such modular recombinations could be facilitated by previously described genomic loci of virus and plasmid-related elements (ViPREs) [20]. Initially, these did not appear to be simple provirus genomes in varying states of decay, but the description of pleolipoviruses with genomes much smaller than His2 and differing replication strategy can explain many of them as virus integrants. On the other hand, some of these genomic loci have expanded in length as more virus and plasmid sequences become available, and their gene homologs can be recognised in flanking sequences. For these, we retain the ViPRE epithet. From the decades of the study of bacteriophages, it is well documented that related phages frequently recombine (by both legitimate and illegitimate recombination events), giving rise to mosaic genomes with varying evolutionary histories [21,22]. This process can be assisted by the modular arrangement of genes (e.g., capsid formation or replication) commonly found in many virus genomes [23]. The same appears true of haloviruses, such as the large recombination event evident in the comparison of halovirus HF1 and HF2 genomes [5]. More generally, there is good evidence of widespread recombination from the study of archaeal MCM helicase genes (often associated with mobile genetic elements) [24] and in the comparative genomics of archaeal caudoviruses [25].
The group of round haloviruses exemplified by SH1 has been recently expanded by the descriptions of HHIV-2 and SNJ1 [13,26]. Members of the SH1 virus group are related by their capsid proteins but differ in genome length and replication strategy. SH1 and HHIV-2 have linear dsDNA genomes (∼30.5 kb) with terminal proteins, indicative of replication by protein priming, while SNJ1 has a circular dsDNA genome of only 16.3 kb. This diversity of genome types within the same virus group closely parallels that of the pleolipoviruses described above. SH1 has been the most intensively studied, including host range, particle stability, virion structure, genome sequence, transcription mapping, transfection, and the establishment of a method of genetic manipulation [8,13,16,[27][28][29][30][31]. The aim of the current study is to describe a new member of this group, PH1. Its virological characteristics, genome sequence, and major proteins are presented and compared to other members of the SH1 virus group and to genomic loci containing related genes. For convenience, we provide the group name halosphaerovirus to encompass the SH1 virus group, currently consisting of SH1, PH1, HHIV-2, and SNJ1.

Virus Isolation and Host Range.
A water sample from Pink Lake, a hypersaline lake (32 ∘ 00 S and 115 ∘ 30 E) in  western Australia, was screened for haloviruses by plating directly on lawns of Har. hispanica using overlay plates of modified growth medium (MGM) with 12% or 18% salts (w/v). A high titre of similar plaque morphology was observed (1.2 × 10 5 PFU/mL) on the plates with 18% (w/v) salts. A novel halovirus was isolated from one of these plaques and designated PH1 based on the source (Pink Lake) and the isolating host (Har. hispanica).
Plaques were fully developed after two days at 37 ∘ C using overlay plates of 18% (w/v) MGM and were 1-2 mm in diameter, clear, and with ragged edges. At 30 ∘ C, plaques took three days to develop, and, at 25 ∘ C, plaques were hazy and took seven days to develop (data not shown).
The host range of PH1 was identical to that of SH1 [8]; it was unable to plaque on lawns of 12 different species belonging to six different genera of the Halobacteriaceae (Haloarcula, Halobacterium, Haloferax, Halorubrum, Haloterrigena, and Natrialba) but could plaque on Halorubrum strain CSW 2.09.4 (described in [8]), an uncharacterized Australian isolate (Table 1).

Virus Purification and Particle Morphology.
Virus was purified by a slight modification of the method described previously for halovirus SH1 ( [8] and Section 4). PH1 banded at a density of 1.29 g/mL in CsCl gradients and gave a final specific infectivity of ∼5 × 10 11 PFU/A 260 . Negative-stain electron microscopy revealed spherical particles, with an average diameter of ∼51 nm ( Figure 1). The capsid displayed two layers, with a compact core particle of ∼43 nm in diameter. This morphology is similar to that of halovirus SH1, a closely related virus (see below) that was isolated at the same time as PH1, but from a neighbouring salt lake [8].

Single-Burst and Single-
Step Growth of PH1. Around 30% of Har. hispanica cells could be infected by PH1, and single burst experiments [32] indicated an average burst size of 87 PFU/cell (data not shown). This value was supported by single-step growth curves, which gave burst sizes of between 50 and 100 PFU/cell. The number of infectious centres usually increased at 4-6 hr p.i., but cell lysis did not appear to begin Archaea 3 until well after this, usually between 14 and 24 hr p.i. A representative example of a single-step growth curve for PH1 is shown in Figure 2, where the rise begins at ∼6 hr p.i., and visible cell lysis begins at ∼14 hr p.i. This figure shows that a second round of growth occurs at around 18 hr p.i., reflecting the initial infection of only 30% of cells, and it is during this second round of virus release that the cell density decreases most rapidly, reaching a value of about 0.1 A 550 , which is close to the initial cell density. In this example, the average burst size for the first growth step was 95 PFU/cell.

Virus
Stability. The stability of PH1 virus was tested under several conditions (Figures 3(a)-3(d)). When stored in HVD at 4 ∘ C, the infectivity of PH1 remained unaltered for several months (data not shown). A thermal stability curve is presented in Figure 3(a) and shows that PH1 is stable up to 56 ∘ C above in which it rapidly loses titre. Particles were sensitive to a reduced salt environment (Figure 3(b)) and to chloroform ( Figure 3(d)). PH1 was most stable between pH 8 and pH 9 ( Figure 3(c)). In general, the stability of PH1 was similar to that described previously for SH1 [8].

PH1 Structural Proteins and Protein
Complexes. The proteins of purified PH1 virus were separated by SDSpolyacrylamide gel electrophoresis, alongside the proteins of SH1 virus ( Figure 4). Nine PH1 protein bands were detected, with molecular weights from 7 to 185 kDa (Figure 4(a)), and these were designated with the prefix VP and a number corresponding to the homologous protein of SH1 [8,27] (see later). This nomenclature is consistent with that of the recently described HHIV-2 virus, a member of the same virus group [13]. The protein profile of PH1 was very similar to that of SH1 [8] and with similar relative masses of the protein bands. One notable difference was that PH1 proteins VP9 and VP10 ran closely together (calculated MWs are 16.5 and 16.7 kDa, resp.) compared to their SH1 homologs (16.5 and 16.9 kDa, resp.). It is probable, given the strong sequence similarity to SH1 (see later), that at least six additional PH1 structural proteins were unable to be visualized using our staining techniques. The major protein bands of PH1 (asterisked in Figure 4) were excised from gels, digested with trypsin, and analysed by MALDI-TOF MS, and the results are summarized in Table 2. From these data, the structural proteins VP 1-4, 7, 9, 10, and 12 were found to be specified by ORFs 12,24,28,21,20,27,26 and 19, respectively (see later).
2.6. Characteristics of the PH1 Genome. Nucleic acid was extracted from purified PH1 virus preparations, treated with proteinase K and incubated with various nucleases to determine the characteristics of the genome ( Table 3). The PH1 genome was sensitive to dsDNA endo-and exonucleases but not to ssDNA nuclease (mung bean) or RNase A. This indicated that the PH1 genome is linear dsDNA, with free (i.e., not covalently closed) termini. Restriction endonuclease digestions of the PH1 genome gave a length of approximately 29 kb (data not shown).
The presence of terminal bound proteins was examined using a silica binding assay [33]. Nonproteinase K-treated PH1 DNA was restricted with AseI, and the four resulting fragments passed through GF/C filters under conditions where proteins bind firmly to the glass. As shown in Figure 5, the two internal AseI fragments (2.3 and 7.6 kb) passed through the filter, but the right terminal fragment (17. Temperature ( ∘ C)     staining to be detected. Further evidence was obtained by a nuclease protection assay, as previously used for SH1 DNA [31]. As shown in Table 3, exonuclease III (a 3 exonuclease) was able to digest nonproteinase K-treated PH1 DNA, but T7 exonuclease (a 5 exonuclease) and Bal31 (5 and 3 exonucleases) were unable to digest PH1 DNA unless it had been previously treated with proteinase K. This indicated that both 5 termini of the genome had protein attached.

2.7.
Sequence of the PH1 Genome. The complete PH1 genome sequence was determined (accession KC252997) using a combination of cloned fragments, PCR, primer walking on virus DNA, and 454 whole genome sequencing (see Section 4). It was found to be 28,072 bp in length, 67.6% G + C, with inverted terminal repeat sequences (ITRs) of 337 bp. Using GLIMMER [34], manual BLAST searching at the GenBank database, and comparison to related viruses, the PH1 genome sequence was predicted to contain 49 ORFs (Table 4). Most ORFs were closely spaced or overlapping, giving a gene density of 1.74 genes/kb (average of 572 nt/gene). The cumulative AT-skew plot of the PH1 genome shown at the bottom of Figure 6 shows inflection points (circled) that are consistent with the changes in transcription direction  Figure 6: Genome alignments of haloviruses PH1, SH1, and HHIV-2, along with two related genomic loci. (a) Genomic loci of Hap. paucihalophilus and Hbf. lacisalsi that contain genes related to halosphaeroviruses SH1, PH1, and HHIV-2. The names and GenBank accessions for these contigs are given on the far right, and ORFs are coloured and labeled to indicate the relationships of these ORFs to those of the viruses below. The locus tag numbers for the first and last ORFs shown in each locus are given nearby their respective ORFs. In addition, grey coloured ORFs represent sequences that do not match any of the haloviruses, and green coloured ORFs represent protein sequences that are closely related to ORFs found within or very close to previously described virus/plasmid loci, so called ViPREs [20]. The scale bars shown above each contig show the position of the described region within the respective contig. (b) The three virus genomes are labeled at the left, with scale markers below (in kb) and the total length indicated at the far right. At the bottom is a cumulative AT-skew plot of the PH1 genome (http://molbiol-tools.ca/Jie Zheng/), with inflection points circled. The grey shaded bands between the genome diagrams indicate significant nucleotide similarity (using ACT [63]). Annotated ORFs are represented by arrows, with colours indicating structural proteins (red or brown), nonstructural proteins (yellow or orange) or the packaging ATPase (blue). The names of structural protein ORFs are indicated either within the arrow (e.g., VP1) or in text nearby. The numbered, orange coloured ORFs of HHIV-2 are homologous to ORFs found in the genomic loci (probably proviruses or provirus remnants) pictured in (a).
indicated by the annotated ORFs, as has been seen in other haloviruses [7]. There are no GATC motifs in the genome, and the inverse sequence, CTAG, is strongly underrepresented with just 2 motifs (68 expected). When compared to the related viruses SH1 and HHIV-2 ( Figure 6), the PH1 genome is seen to be of similar length and is much more closely related to SH1 than to HHIV-2 (74% and 54% nt identity, resp.). Like PH1, the genomes of SH1 and HHIV-2 lack GATC motifs, and CTAG is either absent (SH1) or underrepresented (HHIV-2). The ITRs of PH1 and SH1 share 78.5% nucleotide identity, but the PH1 ITR is longer than that of SH1 (337 and 309 nt, resp.) partly because SH1 has replaced an 18 bp sequence (gtcgtgcggtttcggcgg) found at the internal end of its left-hand ITR with a sequence at the corresponding position of its right-hand ITR that shows little inverted sequence similarity. In the PH1 ITR, this sequence is retained at both ends. Whether this difference represents a recombination event or a mistake in replication of the ITRs is unclear. An alignment of the ITRs of all three viruses identified a number of highly conserved regions (labeled 1-9 in Supplementary Figure 2, see Supplementary Material available online at http://dx.doi.org/10.1155/2013/456318). A 16 bp sequence at the termini (region 1) is conserved, consistent with the conservation seen at the termini of linear Streptomyces plasmids [35]. A 16 bp GC-rich sequence around the middle of the ITR (region 3) is also conserved. Shorter motifs of either C-rich (region 8) or AT-rich sequence (regions 7 and 9) are found near the internal end of the ITRs.
The grey shading between the schematic virus genomes in Figure 6(b) indicates regions of high nucleotide similarity, revealing that the differences between PH1 and SH1 are not evenly distributed but vary considerably along the length of the aligned genomes. A comparison between SH1 and HHIV-2 has been published recently [13], and since the PH1 and SH1 genomes are so similar, we will focus the following description on the differences between PH1 and SH1.
a Extracted nucleic acid from purified PH1 virus was treated with various nucleases and then analysed by gel electrophoresis to detect whether the nuclease digested (+) or did not digest (−) the genome. Control nucleic acids used to confirm activity of the nucleases were b DNA, c a DNA oligonucleotide, and d yeast transfer RNA.
Within corresponding ORFs, there are blocks of coding sequence with low (or no) similarity. These may represent in situ divergence or short recombination events (e.g., indels). There are also cases where an entire ORF in one virus has no corresponding homolog in the other because of an insertion/deletion event (indel). Lastly, there are replacements, where the sequences within the corresponding regions show low or no similarity but are flanked by sequence with high similarity. The largest visible differences seen in Figure 6 are due to sequence changes within or near the long genes encoding capsid proteins VP1 and VP18. A summary of the main sequence differences between SH1 and PH1 and their locations is given in Table 5, where these divergent regions (DV) are numbered from DV1 to DV18.
The sequences of the two viruses are close enough that in many cases the mechanism for the observed differences can be inferred. For example, DV2 is likely the result of a deletion event that has removed the SH1 ORF10 gene homolog from the PH1 genome. At either end of the SH1 ORF are almost perfect direct repeats (cggcctgac/cggcatgac) that would allow repeat-mediated deletion to occur, removing the intervening sequence. Small direct repeats leading to deletions have been described previously in the comparative analysis of Haloquadratum walsbyi genomes [20]. DV5 appears to be an indel, where SH1 ORFs 14-16 are absent in PH1, and an inverted repeat (AGCCATG) found at each end of the SH1 divergent region may be significant in the history of this change. The proteins specified by SH1 ORFs 14-16 are presumably dispensable for PH1 (or their functions supplied by other proteins), but a homolog of SH1 ORF14 is found also in HHIV-2 (ORF 6), and a homolog of SH1 ORF15 occurs in halovirus His1 (ORF13). Replacement regions are also common and can occur within ORFs (e.g., DV3, in capsid protein VP1) or provide additional or alternative genes (e.g., DV12 and DV13). Several divergent regions occur within the gene for capsid protein VP2, a hot spot for change because of the repetitive nature of the coding sequence, which specifies a protein with runs of glycines and many heptapeptide repeats, as described previously for SH1 VP2 [27]. DV18 is a large replacement that significantly alters the predicted length of the minor capsid protein VP18 in the two viruses (865/519 aa for SH1/PH1, resp.) and also provides SH1 with 3 ORFs that are different or absent in PH1: ORFs 52, 53, and 54. For example, a homolog of SH1 ORF52 is not present in PH1 but is present in HHIV-2 (putative protein 40). While ORF48 of PH1 shows no aa sequence homology to SH1 ORFs 53 and 54, all of these predicted proteins carry CxxC motifs, suggestive of related functions, such as DNA binding activity [36].
The genes of the PH1 genome are syntenic with those of SH1, and are found in similarly oriented blocks of closely spaced or overlapping ORFs that suggest that transcription is organised into operons. The programme of transcription of the SH1 genome has been reported previously [30], and the sequences in PH1 corresponding to the six promoter regions (P1-P6) determined in SH1 showed that five of these (P1-P5) are well conserved. Only the region corresponding to SH1 P6 was poorly conserved (data not shown), but this promoter is strongly regulated in SH1 and only switches on late in infection [30].

Structural Protein Genes.
The genes coding for the major virus structural proteins of PH1 (VP 1-4, 7, 9 10, and 12) were identified by MALDI-TOF (see above), and the genes for these proteins are found in the same order and approximate positions as in the SH1 genome ( Figure 6(b), red ORFs). Genes coding for minor structural proteins can be deduced by sequence comparison with SH1 so that VP5 and VP6 are likely to be encoded by PH1 ORFs 25 and 29, respectively, and VP13 and VP18 by ORFs 23 and 49, respectively. This would give PH1 a total of 11 structural proteins, the same as SH1. Of these, VP7 is the most conserved between the two viruses (98% aa identity), followed by VP4, VP9, VP3, and VP12 (91%-94% identity), VP5, VP6, and VP13 (82%-88% identity), VP2 (77%), VP1 (65%), and, lastly, the least conserved protein was VP18 (31%).

Genomic Loci Related to PH1 and Other Halosphaeroviruses.
Clusters of halosphaerovirus-related genes are present in the genomes of two recently sequenced haloarchaea, Hap. paucihalophilus and Hbf. lacisalsi (Figure 6(a)). Neither of these regions appears to represent complete virus genome, but they retain a number of genes that share a similar sequence and synteny with the virus genomes depicted below them (Figure 6(a)). Both loci show a mixed pattern of relatedness to SH1/PH1 and HHIV-2, as indicated by ORFs of matching colour and gene name. If these genomic loci represent provirus integrants that have decayed over time, then the relationships they show suggest not only that halosphaeroviruses are a diverse virus group but also that a significant level of recombination occurs between them, leading to mosaic gene combinations. The two ORFs coloured green in the Hap. paucihalophilus locus ( Figure 6) are related to ORFs found within or very close to genomic loci of virus/plasmid genes found in other haloarchaea, suggesting additional links between (pro)viruses [20].    The recently described halovirus SNJ1 carries many ORFs that have homologs adjacent to a previously described ViPRE of Hmc. mukohataei (from here on denoted by ViPRE Hmuk1 ), a locus that contains genes related to His2 (and other pleolipoviruses) as well as to Haloferax plasmid pHK2 (probably a provirus) and to two small plasmids of Hqr. walsbyi (∼6 kb, pL6A and pL6B). The latter relationship provides wider links to other viruses because one of the PL6 genes (Hqrw 6002) has homologs in some pleolipoviruses (HPRV-3 and HGPV-1), while another gene (Hqrw 6005) is related to ORF16 of the spindle-shaped halovirus His1 [15]. After adding the SNJ1 gene homologs to ViPRE Hmuk1 , it is extended significantly, and a close inspection of the flanking regions revealed a tRNA-ala gene at one end and a partial copy of the same tRNA-ala gene (next to a phage integrase gene) at the other end (supplementary Figure 1). Genes beyond these points appear to be related to cellular metabolism. ViPRE Hmuk1 is now 39379 nt in length, flanked by a 59 nt direct repeat (potential att sites), and includes 54 ORFs, many of which are related to known haloviruses, are virus-like (e.g., integrases, methyltransferases, transcriptional regulators, DNA methyltransferase, and DNA glycosylase), are homologs in or near other known ViPREs, or show little similarity to other known proteins. This locus also includes an ORC1/CDC6 homolog (Hmuk 0446), which could provide a replication function. The mixture of very different virus and plasmid genes seen within ViPRE Hmuk1 is remarkable, but it also reveals homologs found in or adjacent to genomic loci of other species, such as the gene homologs of Har. marismortui seen in the left end of ViPRE Hmuk1 (supplementary Figure  1). When these genes are added to the previously described Haloarcula locus (ViPRE Hmuk1 ), it is also extended significantly. It becomes 25,550 bp in length, encompasses genes from Hmar 2382 to Hmar 2404, and is flanked by a full tRNA-ala gene at one end and a partial copy at the other. It carries an ORC1/CDC6 homolog, an integrase, and a phiHlike repressor as well as pleolipovirus homologs.

Transfection of Haloarchaea by PH1 DNA. PH1 DNA was introduced into
Har. hispanica cells using the PEG method [37], and the cells were screened for virus production by plaque assay. Figure 7 shows the increase in transfected cells with input (nonproteinase K-treated) virus DNA. The estimated efficiency was 5.3 × 10 3 PFU per g DNA. PH1 DNA that had been treated with proteinase K or with DNAase I (RNase-free) did not produce plaques (data not shown). PH1 DNA was found to transfect six species of haloarchaea other than Har. hispanica, including members of the genera Haloarcula, Haloferax, Halorubrum, Haloterrigena, and Natrialba (Table 6). For these experiments, the virus production from transfected cells was detected by plaque assay on indicator lawns of Har. hispanica.  [64]. Cells were then screened for infective centres by plaque assay. Data shown is the average of three independent experiments, performed in duplicate. Error bars represent one standard deviation of the mean. If proteasetreated DNA was used, no transfectant plaques were observed.

Discussion
PH1 is very similar in particle morphology, genome structure, and sequence to the previously described halovirus SH1 [8,16]. Like SH1, it has long inverted terminal repeat sequences and terminal proteins, indicative of protein-primed replication. Purified PH1 particles have a relatively low buoyant Archaea density and are chloroform sensitive and show a layered capsid structure, consistent with the likely presence of an internal membrane layer, as has been shown for SH1, HHIV-2, and SNJ1 [13,26]. The particle stability of PH1 to temperature, pH, and reduced salt was similar to SH1 [8]. The structural proteins of PH1 are very similar in sequence and relative abundance to those of SH1, so the two viruses are likely to share the same particle geometry [16]. Viruses that use protein-primed replication, such as bacteriophage Φ29, carry a viral type B DNA polymerase that can interact specifically with the proteins attached to the genomic termini and initiate strand synthesis [38]. Archaeoviruses His1 [7] and Acidianus bottle-shaped virus [39] probably use this mode of replication. However, a polymerase gene cannot be found in the genomes of PH1, SH1, or HHIV-2. DNA polymerases are large enzymes, and the only ORF of SH1 that was long enough to encode such an enzyme and had not previously been assigned as a structural protein was ORF55, which specified an 865 aa protein that contained no conserved domains indicative of polymerases [27]. The recent study of HHIV-2 showed that the corresponding ORF in this virus (gene 42) is a structural protein of the virion, and, by inference, this is likely also for the corresponding proteins of SH1 and PH1 (no homolog is present in SNJ1). Without a polymerase gene, these viruses must use a host enzyme, but searches for a viral type B DNA polymerase in the genome of the host, Har. hispanica, or in the genomes of other sequenced haloarchaea, did not find any matches. This argues strongly for a replication mechanism that is different to that exemplified by Φ29. An attractive alternative is that displayed by Streptomyces linear plasmids, which have 5 -terminal proteins and use a cellular polymerase for replication [40]. The terminal proteins are not used for primary replication but for end patching [41], and it has been shown that if the terminal repeat sequences are removed from these plasmids and the ends ligated, they can replicate as circular plasmids [42]. This provides a testable hypothesis for the replication of halosphaeroviruses and also offers a pathway for switching between linear (e.g., PH1) and circular (e.g., SNJ1) forms. Use of a host polymerase would also fit with the ability of SH1 and PH1 DNA to transfect many different haloarchaeal species [31] as these enzymes and their mode of action are highly conserved.
The growth characteristics of PH1 in Har. hispanica, both by single-cell burst and single-step growth experiments, gave values of 50-100 viruses/cell, significantly less than that of SH1 (200 viruses/cell) [8] and HHIV-2 (180 viruses/cell) [13]. The latter virus is lytic, whereas SH1, PH1, and SNJ1 start producing extracellular virus well before any decrease in cell density, indicating that virus release can occur without lysis. This has been most clearly shown with SH1, where almost 100% of cells can be infected [8]. Infection of Har. hispanica by PH1 was less efficient (∼30% of cells), but the kinetics of virus production in single-step growth curves of SH1 and PH1 are similar, with virus production occurring over several hours rather than at a clearly defined time after infection. The two viruses are also very closely related and infect the same host species, so it is likely that they use the same scheme for cell exit. For both viruses, the cell density eventually decreases in single-step cultures, showing that virus infection does result in cell death. This mode of exit appears to maximise the production of the virus over time, as the host cells survive for an extended period. In natural hypersaline waters, cell numbers are usually high, but growth rates are low [12], and the high incident UV (on often shallow ponds) would damage virus DNA, so reducing the half-life of released virus particles [43]. These factors may have favoured the exit strategy displayed by these viruses, improving their chance of transmission to a new host.
The motif GATC was absent in the genomes of PH1, SH1, and HHIV-2, and the motif CTAG was either absent or greatly underrepresented. All three viruses share the same host, but GATC and CTAG motifs are plentiful in the Har. hispanica genome (accessions CP002921 and CP002923). Avoidance of these motifs in haloviruses His1, His2, HF1, and HF2 has been reported previously [7]. Such purifying selection commonly results from host restriction enzymes, and, in Hfx. volcanii, it is exactly these two motifs that are targeted [44]. One Hfx. volcanii enzyme recognises A-methylated GATC sites [45], and another recognises un-methylated CTAG sites (which are protected by methylation in the genome). In the current study, this could explain the negative transfection results for Hfx. volcanii, as the PH1 genome contains two CTAG motifs. By comparison, the SH1 genome contains no CTAG motifs and is able to transfect Hfx. volcanii [31].
Comparison of the PH1 and SH1 genomes allowed a detailed picture of the natural variation occurring between closely related viruses, revealing likely deletion events mediated by small repeats, a process described previously in a study of Hqr. walsbyi and termed repeat-mediated deletion [20]. There are also many replacements, including blocks of sequence that occur within long open reading frames (such as the capsid protein genes, VP1 and VP18). VP3 and VP6 are known to form the large spikes on the external surface of the SH1 virion, and presumably one or both interact with host cell receptors [16]. PH1 and SH1 have identical host ranges, consistent with the high sequence similarity shown between their corresponding VP3 and VP6 proteins.
A previous study of SH1 could not detect transcripts across annotated ORFs 1-3 or 55-56 [30]. ORFs 1 and 56 occur in the terminal inverted repeat sequence, and, in the present study, it was found that there were no ORFs corresponding to these in the PH1 genome. Given the close similarity of the two genomes, the comparative data are in agreement with the transcriptional data, indicating that these SH1 ORFs are not used. SH1 ORFs 2 and 3 are more problematic, as good homologs of these are also present in PH1. The conflict between the transcriptional data and the comparative genomic evidence requires further experimental work to resolve. The status of SH1 ORF55 has recently been confirmed by studies of the related virus, HHIV-2 (discussed above).
The pleolipovirus group has expanded dramatically in the last few years, and it now comprises a diverse group of viruses with different genome types and replication strategies. What is even more remarkable is that a similar expansion can now be seen with SH1-related viruses, which includes viruses with at least two replication modes and genome types (linear and circular). Evidence from haloarchaeal genome sequences Archaea 13 show cases of not only provirus integrants or plasmids (e.g., pKH2 and pHH205) but also genomic loci (ViPREs) that contain cassettes of virus genes from different sources, and at least two cases (described in this study) have likely att sites that indicate circularisation and mobility. While there is a clear relationship between viruses with circular ds-and ssDNA genomes that replicate via the rolling circle method (i.e., the ssDNA form is a replication intermediate); it is more difficult to explain how capsid gene cassettes can move between these viruses and those with linear genomes and terminal proteins. Within the pleolipoviruses, His2 has a linear genome that contains a viral type B DNA polymerase and terminal proteins, while all the other described members have circular genomes that contain a rep homolog and probably replicate via the rolling circle method [14,19]. Similarly among the halosphaeroviruses, PH1, SH1, and HHIV-2 have linear dsDNA with terminal proteins and probably replicate in the same way, but the related SNJ1 virus has a circular dsDNA genome. The clear relationships shown by the capsid genes of viruses within each group, plus the connections shown to haloviruses outside each group (e.g., the DNA polymerases of His1 and His2), all speak of a vigorous means of recombination; one that can readily switch capsid genes between viruses with radically different replication strategies. How is the process most likely to operate? One possibility is suggested by ViPREs, which appear to be mobile collections of capsid and replication genes from different sources. They offer fixed locations for recombination to occur, provide gene cassettes that can be reassorted to produce novel virus genomes, and some can probably recombine out as circular forms. For example, ViPRE Hmuk1 carries genes related to pleolipoviruses, halosphaeroviruses, and other haloviruses. It is yet to be determined if any of the genes carried in these loci are expressed in the cell (such as the genes for capsid proteins) or if ViPREs provide any selective advantage to the host.

Water Sample.
A water sample was collected in 1998 from Pink Lake (32 ∘ 00 S and 115 ∘ 30 E), a hypersaline lake on Rottnest Island, Western Australia, Australia. It was screened in the same year for haloviruses using the methods described in [8]. A single plaque on a Har. hispanica lawn plate was picked and replaque purified. The novel halovirus was designated PH1.

Negative-Stain Electron
Microscopy. The method for negative-stain TEM was adapted from that of V. Tarasov, described in the online resource, The Halohandbook (http:// www.haloarchaea.com/resources/halohandbook/index.html). A 20 L drop of the sample was placed on a clean surface, and the virus particles were allowed to adsorb to a Formvar film 400-mesh copper grid (ProSciTech) for 1.5-2 min. They were then negatively stained with a 20 L drop of 2% (w/v) uranyl acetate for 1-2 min. Excess liquid was absorbed with filter paper, and the grid was allowed to air dry. Grids were examined either on a Philips CM 120 BioTwin transmission electron microscope (Royal Philips Electronics), operating at an accelerating voltage of 120 kV, or on a Siemens Elmiskop 102 transmission electron microscope (Siemens AG), operating at an accelerating voltage of 120 kV.

Large Scale Virus Growth and Purification.
Liquid cultures of PH1 were grown by the infection (MOI, 0.05) of an early exponential Har. hispanica culture in 18% (w/v) MGM. Cultures were incubated aerobically at 37 ∘ C, with agitation, for 3 days. Clearing (i.e., complete cell lysis) of PH1-infected cultures did not occur, and consequently cultures were harvested when the absorbance at 550 nm reached the minimum, and the titre (determined by plaque assay) was at the maximum, usually ∼10 10 -10 11 PFU/mL. Virus was purified using the method described previously for SH1 [8,31], except that after the initial low speed spin (Sorvall GSA; 6,000 rpm, 30 min, 10 ∘ C); virus was concentrated from the infected culture by centrifugation at 26,000 rpm (13 hr, 10 ∘ C) onto a cushion of 30% (w/v) sucrose, in HVD. The pellet was resuspended in a small volume of HVD and loaded onto a preformed linear 5%-70% (w/v) sucrose gradient, followed by isopycnic centrifugation in 1.3 g/mL CsCl (Beckman 70Ti; 60,000 rpm, 20 hr, 10 ∘ C). The white virus band occurred at a density of 1.29 g/mL and was collected and diluted in halovirus diluent (HVD, see Section 4), and the virus pelleted (Beckman SW55; 35,000 rpm, 75 min, 10 ∘ C) and resuspended Archaea in a small volume of HVD and stored at 4 ∘ C. Virus recovery at the major stages of a typical purification is given in supplementary Table 1. The specific infectivity of pure virus solutions was determined as the ratio of the PFU/mL to the absorbance at 260 nm.

PH1 Single-Step Growth
Curve. An early exponential phase culture of Har. hispanica grown in 18% (w/v) MGM was infected with PH1 (MOI, 50). Under these conditions the percentage of infected cells was approximately 30%. After an adsorption period of 1 hr at 37 ∘ C, the cells were washed three times with 18% (w/v) MGM (at room temperature), resuspended in 100 mL 18% (w/v) MGM (these methods ensured the removal of all residual virus), and incubated at 37 ∘ C, with shaking (100 rpm). Samples were removed at hourly intervals for measurements of absorbance at 550 nm and the number of infective centres. Immediately after sampling, titres were determined by plaque assay with an indicator lawn of Har. hispanica. Each experiment was performed in triplicate.

Halovirus Stability.
After various treatments, samples were removed and diluted in HVD, and virus titres were determined by plaque assay on Har. hispanica. Each experiment was performed in triplicate, and representative data are shown. Chloroform sensitivity was examined by the exposure of PH1 lysates in 18% (w/v) MGM to chloroform in a volume ratio of 1 : 4 (chloroform to lysate). Incubation was at room temperature, with constant agitation. At appropriate time points, the mix was allowed to settle, and a sample was removed from the upper layer and diluted in HVD. The effect of a low ionic environment was examined by diluting PH1 lysates (in 18% (w/v) MGM) into double distilled H 2 O in a volume ratio of 1 : 1,000 (lysate to double distilled H 2 O). Incubation was at room temperature, with constant agitation. Samples were removed at various times and diluted in HVD. The pH stability of PH1 was determined by dilution of PH1 lysates in 18% (w/v) MGM in the appropriate pH buffer (HVD buffered with appropriate Tris-HCl) in a volume ratio of 1 : 100 (lysate to buffer). Incubation was at room temperature, with constant agitation. After 30 min, samples were removed and diluted in HVD. Thermal stability of PH1 was examined by a 1 hr incubation of a virus lysate in 18% (w/v) MGM at different temperatures, after which they were brought quickly to room temperature, diluted in HVD, and titrated.

Protein Procedures.
To remove salts, purified virus preparations were mixed with trichloroacetic acid (10% (v/v) final concentration) and incubated on ice for 15 min to allow the proteins to precipitate. After centrifugation (16,000 g, 15 min, room temperature), the precipitate was washed three times in acetone, dried, and resuspended in double distilled H 2 O. Proteins were dissolved in Laemmli sample buffer with 48 mM -mercaptoethanol [50], heated in boiling water for 5 min, then separated on 12% (w/v) NuPAGE Novex Bis-Tris Gels using MES-SDS running buffer, according to the manufacturer's directions (Invitrogen). After electrophoresis, gels were rinsed in double distilled H 2 O and stained with 0.1% (w/v) Brilliant Blue G in 40% (v/v) methanol and 10% (v/v) acetic acid. Gels were destained with several changes of 40% (v/v) methanol and 10% (v/v) acetic acid. Alternatively, gels were stained with GelCode Glycoprotein Staining Kit, Pro-Q Emerald 488 Glycoprotein Gel and Blot Stain Kit, SYPRO Ruby Protein Gel Stain, according to the manufacturer's directions (Invitrogen, Molecular Probes, Pierce Biotechnology).
Protein bands were cut from the gels and sent to the Australian Proteome Analysis Facility (Macquarie University) for trypsin digestion and analysis by matrix assisted laser desorption ionisation time of flight mass spectrometry (MALDI-TOF MS) on an Applied Biosystems 4700 Proteomics Analyser (Applied Biosystems).
4.9. DNA Procedures. Proteinase K-treated and nonproteinase K-treated DNA preparations were made from purified PH1 using the methods described previously for SH1 [8]. DNA was separated on 1% (w/v) agarose gels in Tris-acetate-EDTA electrophoresis buffer and was stained with ethidium bromide (Sigma-Aldrich).
DNA, DNase I (RNase-free), exonuclease III, mung bean nuclease, nuclease BAL-31, T4 DNA polymerase, T4 DNA ligase, T7 exonuclease, and type II restriction endonucleases were purchased from New England Biolabs. RNase A and yeast transfer RNA were purchased from Sigma-Aldrich. Proteinase K was purchased from Promega. Oligonucleotide primers were purchased from Geneworks, Australia.
To clone fragments of the PH1 genome, purified virus DNA was digested either with AccI, Eco0109I, and MseI or SmaI restriction endonucleases, blunted with DNA Polymerase I, Large (Klenow) Fragment where appropriate, and ligated into AccI-, Eco0109I-, or SmaI-digested pBluescript II KS+ using T4 DNA ligase, according to the manufacturer's instructions (New England Biolabs). The DNA was introduced into E. coli XL1-Blue and transformants grown on Luria agar containing 15 g/mL tetracycline and 100 g/mL ampicillin. The resulting clones were sequenced, and the sequences were used to design specific oligonucleotide primers for PCR amplification and/or primer walking using the virus genome (or specific restriction fragments) as templates.
To amplify PH1 nucleic acid, approximately 10 ng virus DNA was combined with 500 nM primers, 100 M of each dNTP, 1 U Deep Vent DNA Polymerase (New England Biolabs), and 1 × ThermoPol buffer (New England Biolabs), in a total reaction volume of 50 L. Template was denatured at 95 ∘ C for 10 min, followed by 30 cycles of denaturation at 95 ∘ C for 30 sec, annealing at 56 ∘ C for 30 sec, extension at 75 ∘ C for 2 min, and a final extension at 75 ∘ C for 10 min. PCR reactions were performed on a PxE0.2 Thermo Cycler (Thermo Electro Corporation). Sequencing reactions using 3.2 pmol primer were performed by the dideoxy chain termination method using ABI PRISM Big Dye Terminator Mix version 3.1 on an ABI 3100 capillary sequencer (Applied Biosystems) by the Applied Genetic Diagnostics Sequencing Service at the Department of Pathology (The University of Melbourne).
To obtain terminal genomic fragments, PH1 DNA was digested by the enzyme Pas1, and the head (∼700 bp) and the tail (∼1300 bp) fragments were purified by the QIAEX II gel extraction kits and treated with 0.5 M piperidine for Archaea 15 2 hr at 37 ∘ C to remove protein residues. The piperidinetreated fragments were sequenced by the Genomics Biotech Company (Taipei, Taiwan), using two primers: TGACCA-ATTAATTAGGCCGGTTCGCC (PH1 head-R) and GTGC-CATACTGCTACAATTCT (PH1 tail-F).
Four hundred fifty-four whole genome sequencing: PH1 DNA samples (∼5 g) were sequenced using parallel pyrosequencing on a Roche 454 Genome Sequencer System at Mission Biotech (Taipei, Taiwan). The largest contig was 27399 with 8803 reads (Newbler version 2.7), with 27387 positions being of Q40 quality. The average depth of coverage for each base was 82.9. The GenBank accession for the entire PH1 sequence is KC252997.