A Novel Virus Family, the Rudiviridae: Structure, Virus-Host Interactions and Genome Variability of the Sulfolobus Viruses SIRV1 and SIRV2

Corresponding author: David Prangishvili, Max-Planck-Institut für Biochemie, Am Klopferspitz 18a, 82152 Martinsried, Germany., prangish{at}biochem.mpg.de (E-mail)

Communicating editor: P. BLUM

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The unenveloped, stiff-rod-shaped, linear double-stranded DNA viruses SIRV1 and SIRV2 from Icelandic Sulfolobus isolates form a novel virus family, the Rudiviridae. The sizes of the genomes are 32.3 kbp for SIRV1 and 35.8 kbp for SIRV2. The virions consist of a tube-like superhelix formed by the DNA and a single basic 15.8-kD DNA-binding protein. The tube carries a plug and three tail fibers at each end. One turn of the DNA-protein superhelix measures 4.3 nm and comprises 16.5 turns of B DNA. The linear DNA molecules appear to have covalently closed hairpin ends. The viruses are not lytic and are present in their original hosts in carrier states. Both viruses are quite stable in these carrier states. In several laboratory hosts SIRV2 was invariant, but SIRV1 formed many different variants that completely replaced the wild-type virus. Some of these variants were still variable, whereas others were stable. Up to 10% nucleotide substitution was found between corresponding genome fragments of three variants. Some variants showed deletions. Wild-type SIRV1, but not SIRV2, induces an SOS-like response in Sulfolobus. We propose that wild-type SIRV1 is unable to propagate in some hosts but surmounts this host range barrier by inducing a host response effecting extensive variation of the viral genome.

KNOWLEDGE about viruses of the Archaea, the third domain of life, is still rather limited. Among about two dozen known archaeal viruses, only a few have been studied in detail at the molecular level (reviewed in ZILLIG et al. 1988 ; STOLT and ZILLIG 1994 ).

The morphotypes of known archaeal viruses reflect the division of the domain Archaea into two kingdoms, Euryarchaeota and Crenarchaeota (WOESE et al. 1990 ). All but two viruses of Euryarchaeota (WOOD et al. 1989 ; BATH and DYALL-SMITH 1998 ) are head-and-tail phages assigned to the families Myoviridae or Syphoviridae (reviewed in STOLT and ZILLIG 1994 ). In contrast, all known crenarchaeotal viruses exhibit unique morphotypes and compose three novel virus families: the Lipothrixviridae, enveloped lipid-containing filamentous viruses with linear double-stranded DNA genomes (TTV1, TTV2, TTV3, SIFV, and DAFV); the spindle-shaped Fuselloviridae with circular double-stranded DNA genomes (SSV1, SSV2, and SSV3); and the rod-shaped but uncoated Rudiviridae with linear double-stranded DNA genomes (reviewed in ZILLIG et al. 1996 , ZILLIG et al. 1998 ). One more crenarchaeotal virus, SNDV, with a bearded droplet shape has been proposed to represent a fourth family, the Guttaviridae (H. P. ARNOLD, U. ZIESE and W. ZILLIG, unpublished results). The fusellovirus SSV1 has been studied most intensively (MARTIN et al. 1984 ; REITER et al. 1988 ; PALM et al. 1991 ; SCHLEPER et al. 1992 ).

A preliminary description of the first representative of the Rudiviridae (the family has been approved by the ICTV Committee, 7th Report of the Committee 1999, unpublished data), the virus SIRV ("Sulfolobus islandicus" rod-shaped virus), has been previously published (ZILLIG et al. 1994 ). A second representative of the Rudiviridae has subsequently been found (ZILLIG et al. 1998 ); the original SIRV was renamed SIRV1 and the new virus SIRV2. Here we present the first detailed description of the two viruses: their structure, virus-host interactions, and the extremely high degree of genomic variation of SIRV1. The results not only expand knowledge of viral diversity, but also demonstrate the existence of an unusually efficient mechanism of genomic change in SIRV1 and provide an attractive model for studies on virus evolution.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and cell growth:
Sulfolobus strains KVEM10H1, KVEM10H3, REN2H1, KVE6/3, HVE10/2, and LAL14/1 were isolated from samples taken from solfataric fields in Iceland and colony cloned, as described by ZILLIG et al. 1994 . All of these heterotrophic strains belong to a species provisionally termed Sulfolobus islandicus, which is characterized by a well-defined restriction pattern of the chromosomal DNA, differing from those of S. solfataricus and S. shibatae. Both these species appear, however, to be close relatives of S. islandicus as indicated by DNA-DNA cross-hybridization. A formal description of the novel species awaits the completion of its 16S rRNA sequence. Other strains of Sulfolobus used in this study were S. solfataricus P1, DSM 1616, S. acidocaldarius, DSM 639, and S. shibatae B12, DSM 5389. Cells were grown as described by ZILLIG et al. 1994 . The minimal medium (BROCK et al. 1972 ), used either in liquid form or in Gelrite (Kelco, San Diego) gels, was supplemented with 2 g/liter of tryptone (Difco, Detroit) and adjusted to pH 3.2 with diluted sulfuric acid.

Enzymes:
Terminal deoxynucleotidyl transferase (Pharmacia, Uppsala, Sweden), T4 polynucleotide kinase (Pharmacia), bacteriophage exonuclease (Pharmacia), nuclease BAL 31 (BRL), trypsin (Sigma, Deisenhofen, Germany), proteinase K (Boehringer Mannheim, Mannheim, Germany), and restriction enzymes (MBI Fermentas) were used as recommended by the manufacturers.

Virus purification and plaque assays:
The original hosts of the viruses, S. islandicus strains KVEM10H3 and HVE10/2, were grown to the late stationary phase and, after removal of cells by centrifugation, virus particles were precipitated from cell-free medium by addition of polyethyleneglycol 6000 to 100 g/liter and sodium chloride to 1 M overnight at 4°. The precipitated virus was collected and purified by centrifugation in a cesium chloride density gradient as described by ZILLIG et al. 1994 .

To infect laboratory hosts with viruses, cells were grown in 20 ml of liquid medium to an OD₆₀₀ = 0.25 and were mixed with about 3 x 10⁷ plaque-forming units (pfu) of virus. After incubation for 1 hr at 80° with mild shaking, the mixture was diluted with medium to 500 ml and cells were grown to the late stationary phase. The released virus particles were collected and purified as described above.

Plaque assays were performed as described by ZILLIG et al. 1994 with a soft layer of 1.5 ml of a 0.2% Gelrite gel, containing about 6 x 10⁷ host cells and serial dilutions of the virus suspension, layered over 20 ml of 0.6% Gelrite supporting gel. The plates were incubated for 48 hr at 80°. To analyze a plaque-purified virus, a piece of soft layer, containing a single plaque and the surrounding area of the host lawn, was used to inoculate 25 ml of medium which, after growth to OD₆₀₀ = 0.5, was mixed with 200 ml of a culture of the same host grown to OD₆₀₀ = 0.2. The culture was grown to the late stationary growth phase and, after removal of cells, the virus particles were precipitated and purified.

Plaque-forming efficiency of SIRV1 on REN2H1 lawns and of SIRV2 on LAL14/1 lawns was compared according to the number of pfu produced by the same number of virus particles, as determined in the electron microscope.

Determination of infection time, eclipse, and latent periods:
The time of infection is defined as the time interval required for infection of 50% of the host cells at a multiplicity of infection (m.o.i.) of 3. To determine this time interval, aliquots were taken from infected cultures every 2 min and the cells were collected by centrifugation, resuspended, and tested in a plaque assay.

The eclipse period, the time interval between infection and the appearance of intracellular virus particles, and the latent period, the time interval between infection and release of the first virus particles, were determined as follows. The host cells were grown to OD₆₀₀ = 0.2 and were infected at an m.o.i. of 3. After 25 min the infected cells were diluted 1:15 into preheated growth medium and further incubated at 80°. At regular time intervals aliquots were taken and the cells were collected by centrifugation. The virus titer in the cell-free supernatant as well as inside the cells was determined by a plaque assay. Intracellular virus particles were released by lysis of the host cells in 0.1% Triton X-100 at 50° for 20 min. The titer of the host cells in the experiment was determined by following the OD₆₀₀ as calibrated by counting the cells in a counting chamber (OD₆₀₀ = 0.15 corresponds to 10⁸ cells/ml).

DNA and protein analysis:
Cellular and viral DNA was prepared and purified in a cesium chloride density gradient containing ethidium bromide, as described by ZILLIG et al. 1994 . Restriction digests of the DNA were analyzed on a horizontal 1% agarose gel according to standard procedures (SAMBROOK et al. 1989 ). DNA probes were labeled with [-³²P]ATP (Amersham, Arlington Heights, IL) as described by FEINBERG and VOGELSTEIN 1983 or with digoxigenin, using the DIG labeling and detection system of Boehringer Mannheim. Oligonucleotides were labeled with [-³²P]ATP using T4 polynucleotide kinase (SAMBROOK et al. 1989 ). Southern hybridizations were performed using standard procedures (SAMBROOK et al. 1989 ). The filters were washed in 1x SSC/0.1% SDS at either 58° or 65°. Proteins were analyzed by SDS-PAGE (SCHAGGER and VON JAGOW 1987 ) and stained with Coomassie brilliant blue (Serva).

DNA sequencing:
Double-stranded DNA was cloned into a plasmid pBS19 cut with EcoRI, ClaI, HindIII, or SacII restriction endonucleases and sequenced using the United States Biochemical (Cleveland) Sequenase 2.0 kit, following the manufacturer's instructions. Oligonucleotide primers were complementary either to adjacent vector sequences or to previously sequenced parts of the cloned DNA. To ensure accuracy, both DNA strands were sequenced.

Treatment with nuclease BAL31:
DNA (15 µg) was treated with 2 units of nuclease BAL31 in a buffer containing 20 mM Tris-HCl, pH 8, 12.5 mM MgCl₂, 12.5 mM CaCl₂, 0.6 M NaCl, and 1 mM EDTA (total volume 80 µl) at 37°. After 0, 10, 20, 45, and 70 min of incubation, aliquots were taken. Aliquots were treated with phenol and, after ethanol precipitation, DNA was dissolved and digested with restriction endonucleases.

Electron microscopy and structural analysis of virus particles:
For negative staining, a drop of the virus preparation was placed on a carbon-coated copper grid. The particles were allowed to adsorb to the carbon for 1 min. After that the excess liquid was removed with filter paper. The sample was treated with a solution of 2% uranyl acetate for 30 sec and then air dried.

Specimens were observed in a Philips CM200FEG electron microscope. Images were recorded using a slow-scan CCD camera attached to the microscope. The microscope was operated at 120 kV accelerating voltage and x20,000 total magnification on the CCD camera. The calibrated CCD pixel size referring to the specimen level was 0.48 nm.

Image processing was performed on a Silicon Graphics Indigo workstation using the EM program package (HEGERL 1996 ). Images were filtered so that they included only the strongest features of their Fourier transformations.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Morphology:
The viruses SIRV1 and SIRV2 were produced by the colony-cloned S. islandicus strains KVEM10H3 and HVE10/2, respectively. The two strains were isolated from samples taken from different solfataric fields in Iceland, the Kverkfjöll, and Hveragerdi, which are separated by a distance of 250 km. The viruses were purified to homogeneity from the cell-free supernatant of late exponential cultures of their natural hosts, as described in MATERIALS AND METHODS. The particles of the two viruses had the same density, 1.36 g/ml, as determined by equilibrium centrifugation in a CsCl isopycnic gradient.

Both viruses are stiff rods of about 23 nm in width, but differing in length—SIRV1 is about 830 nm and SIRV2 is about 900 nm long. In electron micrographs of negatively stained virus particles a central cavity was visible (Figure 1). The ends were plugged by stoppers filling the terminal portion of the cavity over a length of about 48 nm. Three short tail fibers each were seen protruding from both ends (Figure 1). Because the virus particles are fragile and were partially broken in the course of purification, the virus body (i.e., virion minus plugs and tail fibers), the plugs, and the bundle of tail fibers were sometimes seen as three morphologically distinct, separate structural components. In fractions containing broken particles the plugs could sometimes be seen protruding out of the hollow virion tubes or separate beside them (Figure 2). The plugs are short rods about 45 nm long and 6 nm wide and are pointed at one end. The darker staining in the middle of the plugs may indicate a narrow central cavity. Three tail fibers connected to each other at their bases could also be seen as a distinct morphological component (Figure 2). The tail fibers are about 28 nm long. It is not clear whether the tail fibers are attached to the plugs or to the virus body, because they were observed to be connected to protruding plugs, as well as to the virus body (data not shown).

View larger version (102K): In this window In a new window Download PPT slide   Figure 1. Electron micrograph of SIRV2 negatively stained with 2% uranyl acetate. Bar, 200 nm.

View larger version (212K): In this window In a new window Download PPT slide   Figure 2. Electron micrograph of broken SIRV2 particles stained with 2% uranyl acetate, showing separate plugs and a bundle of tail fibers (arrow). Bar, 100 nm.

SIRV1 was nearly inactivated by 12 min incubation at 100° and was completely destroyed after autoclaving for 50 min at 120°. Particles of SIRV1 and SIRV2 were not inactivated by treatment with 6 M urea at neutral pH and 25°, 0.1% Triton X-100, absolute ethanol, or octanol-2. This insensitivity to Triton at low concentrations and to the solvents indicates that the viruses have no hydrophobic envelope, which should be destroyed under these conditions. Fractionation experiments in which SIRV2 particles were partially disrupted by incubation with 0.1% SDS at 50° revealed that the helical virus body is formed by a filament (Figure 3A). A central cavity was seen in electron micrographs of disrupted virus particles: cross sections of particles looked like rings with a central hole ~6 nM wide (Figure 3B). Further details of the structure of the virus body were provided by computerized analysis of electron micrographs of virus particles. Figure 3C shows a part of a SIRV2 particle (original image) and the power spectrum of this image. The latter indicates a periodicity of 4.3 nm in the virus particle. If a binary filter is applied to the Fourier transformation of the image, the periodicity can also be seen in the image itself (Figure 3C, filtered image). The results indicate that the virus helix has a period of 4.3 nm. Assuming that the viral DNA is in the B conformation implies that the genome is condensed by a factor of 12.7 with 16.5 B-DNA helix turns per superturn of the virus body. All of these data enabled the construction of a schematic model of the helical body of SIRV2 (Figure 3D). The similarity of the structures of SIRV1 and SIRV2 as observed in the electron microscope suggests that this model is also valid for SIRV1.

View larger version (105K): In this window In a new window Download PPT slide   Figure 3. Filamentous structure of SIRV2. (a) Electron micrograph of SIRV2 after incubation with 0.1% SDS showing fragments connected to each other by a filament (indicated by arrow). (b) Electron micrograph of fragments of SIRV2; one fragment is seen in top view, and the other two are seen in side view. (c) Original and filtered images of the virus body of SIRV2. (d) A schematic model of the helical virus body of SIRV2.

Protein composition:
The protein composition of the viruses was studied by SDS-PAGE. Protein preparations from both viruses yielded only two bands on Coomassie-stained gels: a sharply defined minor band corresponding to a protein with a molecular mass of ~150 kD and a rather broad major zone consisting of many tightly spaced sharp single bands centered at 20 kD (ZILLIG et al. 1994 and data not shown). The amount of the minor protein was <1% of that of the major protein. In spite of the multiple bands of the major protein, no heterogeneity was detected in the N terminus. In the case of SIRV1 the N-terminal sequence was AKGHTKRSYSQRYAKPQAKFNAFS; in the case of SIRV2 the sequence was the same, except that serine was substituted for the lysine at position 6. The protein is strongly basic. The theoretical isoelectric point, determined from the deduced amino acid sequence of the protein (see below), is 11.06. The data imply that the major protein binds to the virus DNA, forming the DNA-protein superhelix of the virus body. The gene for the protein has been sequenced (see below). It encodes a polypeptide of 135 amino acid residues with a calculated molecular mass of 15.8 kD.

DNA structure:
As suggested by Southern hybridization experiments (data not shown), the genomes of SIRV1 and SIRV2 are similar. This is also indicated by the high degree of similarity of the N-terminal sequences of the major proteins from the two viruses (23 amino acids out of 24 were identical; see above).

The DNAs of SIRV1 and SIRV2 could be cleaved by a number of type II restriction endonucleases, indicating that they are double stranded. Because no cccDNA band was detected by ethidium bromide/cesium chloride gradient centrifugation of DNA prepared from purified virus particles, both DNAs are linear. The genomes of the two viruses are of different sizes, as estimated by restriction digestion and agarose gel electrophoresis, and the difference parallels the difference in the length of the virus particles: the shorter SIRV1 has a 32.3-kbp genome and the longer SIRV2 has a 35.8-kbp genome.

The ends of the virus DNAs were found to be protected. It was not possible to label 3' ends with terminal deoxynucleotidyl transferase or 5' ends with T4 polynucleotide kinase or to degrade the DNAs from their 5' ends with exonuclease (data not shown). However, treatment with the nuclease BAL31 led to specific exonucleolytic degradation of the DNAs. In restriction digests of DNAs of SIRV1 variant VIII (see below) and SIRV2, two fragments disappeared upon BAL31 treatment (Figure 4). These fragments should be terminal. This assumption was confirmed by the DNA sequence of SIRV1 variant VIII (H. BLUM, S. MALLOK, K. BARTLAU, H. DOMDAY, D. PRANGISHVILI and W. ZILLIG, unpublished results). Treatment of the viral DNAs with proteolytic enzymes (trypsin, proteinase K) prior to restriction digestion and analysis on an agarose gel did not lead to a change in the restriction cleavage patterns of DNAs of SIRV1 and SIRV2, indicating that the 5' ends of the DNAs are not covalently bound to a protein (data not shown). This assumption was confirmed by the observation that the intensities of the bands representing the terminal fragments were not reduced in patterns obtained after phenol extraction of the restriction digests of virus DNAs (data not shown), in contrast to what has been reported for virus genomes with 5'-covalently attached proteins (reviewed in VARTAPETIAN and BOGDANOV 1987 ).

View larger version (139K): In this window In a new window Download PPT slide   Figure 4. Degradation of termini of different DNAs with BAL31 nuclease followed by digestion with restriction endonucleases. Lanes 1–4, SIRV1 variant VIII DNA (restriction digestion with HindIII); lanes 5–8, SIRV2 DNA (restriction digestion with SalI + BamHI + PstI); lanes 9–12, T7 DNA (restriction digestion with DraI). Times of treatment with exonuclease BAL31 were as follows: lanes 1, 5, and 9, 10 min; lanes 2, 6, and 10, 20 min; lanes 3, 7, and 11, 45 min; lanes 4, 8, and 12, 70 min. Lane M, molecular size marker /BstEII. Fragments disappearing with increased BAL31 treatment are marked with asterisks.

Virus-host interactions:
In restriction digests of total DNAs prepared from exponentially growing cells of the original hosts of SIRV1 and SIRV2, the fragments generated from the virus DNAs could be easily recognized (data not shown). The copy number of the virus DNA was estimated to be in the range of 10–20 copies per host chromosome. As in the virus particles, the virus DNA in the host cells was found in the linear form only: no cccDNA could be isolated by the procedure of BIRNBOIM and DOLY 1979 , and no cccDNA band was detected in the course of fractionation by ethidium bromide/cesium chloride gradient centrifugation of total DNA from cells.

No fragments in addition to those of the virus DNAs were detected by Southern hybridization of labeled virus DNAs to blots of restriction digests of total DNAs from host cells (data not shown). The results indicate that neither SIRV1 nor SIRV2 DNA is integrated into the host chromosome.

Mature virus was released from host cells even in the early exponential growth phase. Virus release continued up to the early stationary growth phase and then ceased. The titer reached a plateau at only 2 x 10⁷ pfu/ml in the case of SIRV1 but 3 x 10⁸ pfu/ml in the case of SIRV2. Virus production could be induced neither by UV irradiation nor by treatment with mitomycin C. It was not accompanied by lysis of host cells.

Growing cells of the original hosts were cured of the virus after several serial transfers into fresh medium and continuous growth. The viral DNAs completely disappeared from cells, as shown by Southern hybridization (data not shown). The cured KVEM10H3 strain could not be reinfected with wild-type SIRV1, but could be infected with SIRV1 variants VIII and XIII (see below) and with SIRV2. The cured HVE10/2 strain could not be reinfected with SIRV2, but could be infected with the SIRV1 variants VIII and XIII.

The S. islandicus strains REN2H1, KVEM10H1, LAL14/1, and KVE6/3 were hosts for both viruses. S. solfataricus P1, S. acidocaldarius, and S. shibatae were not hosts for either virus. In spite of this, SIRV1 is able to efficiently induce production of the Sulfolobus virus SSV1 from S. solfataricus lysogenic for SSV1 and is presumably able to enter the cells of the lysogen (ZILLIG et al. 1994 ). SIRV2 was not capable of such induction (K. STEDMAN, personal communication).

Plaque assays were established for both viruses. Rather clear, sharp-edged plaques appeared within 2–3 days at 80°, due to the inhibition of the growth rate of the host cells. Their diameter depended on the inoculum of host cells in the soft layer. Under standard conditions (see MATERIALS AND METHODS) the diameter was 0.7–3 mm. The plaque-forming efficiency of SIRV1 was about 3 x 10^-3 if we assume that the plaque-forming efficiency of SIRV2 is about 1.

For studying the virus-host interactions, S. islandicus strains REN2H1 and LAL14/1 were used for SIRV1 and SIRV2, respectively. The infection times, eclipse, and latent periods are listed in Table 1. Owing to the fact that the viruses neither lysed nor killed their hosts, a standard one-step growth curve could not be produced. After latent periods of 8 hr for SIRV1 and 6 hr for SIRV2, the titer of free virus continuously increased, reaching plateau values of 7 x 10⁹ pfu/ml for SIRV1 and 4 x 10⁸ pfu/ml for SIRV2. In the case of SIRV2 this corresponds to little more than one free virion per cell although the number of intracellular virus particles reached about 30. As in the case of the original virus-carrying hosts, viral DNAs were not integrated into the chromosomes of the laboratory hosts, again as shown by Southern hybridization (data not shown). Similar to the original hosts, upon continuous growth, infected laboratory hosts were cured of the viruses.

Instability of the SIRV1 genome:
Upon propagation of SIRV1 in REN2H1, the viral genome changed dramatically. The pattern of the EcoRI + ClaI-generated fragments of DNA of the virus produced in this host (Figure 5A, lane 2) was strikingly different from that of the virus produced in the original virus-carrying host KVEM10H3 (Figure 5A, lane 1). Several bands were present only in one or the other pattern. In addition, cleavage of the DNA of the virus produced in REN2H1 yielded substoichiometric bands suggesting a mixture of different genomes. Virus produced by infection of the same host with a single plaque from this mixture yielded a third pattern of DNA fragments (Figure 5A, lane 3). Serial repetition of this procedure (flow scheme in Figure 6) yielded a series of virus variants (variants IV, V, VI, and VII), with DNA fragment patterns partially different from those of the variant used for infection, but identical in some bands (Figure 5A, lanes 4, 5, 6, and 7). These changes appeared reversible in some cases. For example, the largest band in variant III had entirely disappeared in variant IV, but reappeared in variant V. The DNA fragment pattern of variant VII strongly resembled that of variant IV.

View larger version (194K): In this window In a new window Download PPT slide   Figure 5. Agarose gel electrophoresis of EcoRI + ClaI-generated fragments of DNAs from different variants of SIRV1. a, b, and c are stained with ethidium bromide. (a1) Southern hybridization of the blot of gel "a" with 32P-labeled DNA from virus variant VIII; (a2) Southern hybridization of the blot of gel "a" with 32P-labeled oligonucleotide 5'-AAATGGCAAGCCAAATTTAATGCATT-3', encoding a portion of the DNA-binding structural protein. Variants and variant mixtures yielding corresponding fragment patterns are marked with Roman numerals. Lane 1, wild-type SIRV1; lanes 18 and 28, SIRV2.

View larger version (22K): In this window In a new window Download PPT slide   Figure 6. Flow scheme of isolation of variants of SIRV1 by serial multiplication from single plaques (see text for details).

The process of virus variant formation was also studied by following changes of the virus obtained from plaques generated by the original SIRV1 on REN2H1. Two such virus strains (variants VIII) yielded identical DNA fragment patterns differing from those of the original virus, which did not change at all in the virus yielded from two plaques each in 9 subsequent rounds of serial infection (flow scheme in Figure 6, DNA fragment patterns in Figure 5A, lane 8 and 5b, lanes 9 and 10). In the 10th round one of the two plaques yielded a virus identical to variant VIII (Figure 5B, lane 11); the other yielded a new variant IX (Figure 5B, lane 12). Of three plaques of the next round of propagation of virus variant VIII, one yielded variant VIII again (Figure 5B, lane 15), the other two yielded the new virus variants X and XI (Figure 5B, lanes 16 and 17, respectively). All three plaques of the next round of propagation of variant IX yielded variant IX again (Figure 5B, lane 14).

In another experiment, 9 of 10 plaques generated by the wild-type virus yielded the new variant XII, and 1 yielded the variant mixture XIII (Figure 5C, lanes 20 and 21, respectively). Five plaques form one variant XII all yielded variant XII again (Figure 5C, lane 22). In contrast, 5 plaques generated by virus mixture XIII yielded different new variants or variant mixtures XIV, XV, XVI, XVII, and XVIII (Figure 5C, lanes 23, 24, 25, 26, and 27, respectively). Except for many minor bands, the fragment pattern of variant mixture XVIII (Figure 5C, lane 27) did not differ from that of the wild-type virus (Figure 5A, lane 1). Like the wild-type virus, variant XVIII was not stable upon further propagation (data not shown). A flow scheme of the generation of all SIRV1 variants investigated is shown in Figure 6.

The gel shown in Figure 5A was blotted and hybridized with ³²P-labeled DNA from SIRV1 variant VIII. All bands from all variants hybridized, indicating a close relationship of the DNAs from different virus variants (Figure 5a₁). The blot of the gel was also hybridized with the oligonucleotide 5'-AAATGGCAAGCCAAATTTAATGCATT-3' corresponding to a portion of the gene for the major structural protein of SIRV1 (see above). Whereas in DNA from the original virus only one fragment of the restriction digest hybridized to this oligonucleotide (Figure 5a₂, lane 1), in DNA of the variant mixture II four restriction fragments did so with different intensities (Figure 5a₂, lane 2). One or more of these four fragments were also visible in the hybridization patterns of variants or variant mixtures III, IV, V, VI, VII, and VIII (Figure 5a₂, lanes 3, 4, 5, 6, 7, and 8).

Except variant mixture XVIII, the DNA fragment patterns of all variants were different from the stable pattern of the virus produced by the original colony-cloned host KVEM10H3. Hybridization experiments with total DNA prepared from cells of this strain have also revealed the presence in cells of only the wild-type virus genome (data not shown). Thus it appears that the virus variants were formed upon propagation in the laboratory host.

Dramatic changes of the genome of wild-type SIRV1 were observed not only upon propagation in REN2H1, but also in two other hosts, KVEM10H1 and LAL14/1 (data not shown). If not infected, none of these three strains produced virus-like particles or contained extrachromosomal DNA. As determined by DNA-DNA hybridization experiments (data not shown) chromosomal DNAs of the strains did not contain sequences homologous to SIRV1 sequences that could recombine with virus DNA. SIRV1 variant VIII, which was generated and remained mostly stable in REN2H1, changed immediately upon propagation in cured cells of KVEM10H3, the original host of SIRV1 (data not shown).

In contrast to the genome of SIRV1, that of SIRV2 (Figure 5B, lane 18 and 5c, lane 28) did not change upon propagation in the strains LAL14/1 and KVE6/4, or in cured strain KVEM10H3 (data not shown).

The nature of variation:
Insight into the nature of the differences between the variant genomes was obtained by sequencing a 2101-bp-long EcoRI-ClaI fragment from the virus variant mixture II, containing the gene for the putative DNA-binding protein (the smallest of the four bands in lane 2, Figure 5a₂). The fragment was cloned into pBS19 and three arbitrary recombinant clones were sequenced. Unexpectedly, the three sequences were not identical but only similar, differing from each other by nucleotide substitutions. The percent sequence identities between the three possible pairs of these sequences were 87%, 91%, and 94%, respectively. The nucleotide substitutions appeared to be statistically distributed over the whole sequence and their frequency was nearly equal in coding and noncoding regions. However, in coding regions 92% of the nucleotide changes were neutral and there was an obvious preference for changes in the third codon position (72%). Portions of these sequences, encompassing the gene for the DNA-binding protein and adjacent regions, are aligned in Figure 7, which also shows the corresponding amino acid sequences of the protein.

View larger version (84K): In this window In a new window Download PPT slide   Figure 7. An alignment of three nucleotide sequences from genomes of three SIRV1 variants present in variant mixture II, including the putative DNA-binding protein gene and adjacent regions. The gene is also translated into amino acid sequence. Borders of the gene are marked with bars. Nucleotides and amino acids that differ between the sequences are shaded.

It is hard to estimate the mutation rate, mainly because of the difficulty in determining the exact number of replication cycles leading to the production of variant mixture II from the wild-type virus. In this experiment (see MATERIALS AND METHODS), infection of REN2H1 with 3 x 10⁷ pfu of the wild-type SIRV1 resulted in the production of 3.5 x 10¹² pfu of virus variants, requiring about 16 replication cycles. Presuming that about 10% difference in the progeny sequences corresponds to at least 5% difference of these sequences from the original sequence, the mutation rate in the sequenced 2101-bp-long fragment is then in the range of 3 x 10^-3 substitutions per nucleotide per replication cycle. The value could be, however, considerably overestimated, because it does not account for the production of nonviable virus particles, as well as intracellular viral genomes.

A comparison of the nucleotide sequence of the whole genome of virus variant VIII (H. BLUM, S. MALLOK, K. BARTLAU, H. DOMDAY, D. PRANGISHVILI and W. ZILLIG, unpublished results) with the sequences of cloned DNA fragments from the variant mixture II, distributed over the entire length of the genome, about 11 kbp in total, has shown that nucleotide substitutions occur throughout the genome. In addition to nucleotide substitutions, this comparison revealed the existence of deletions and insertions of short sequences (D. PRANGISHVILI, unpublished data).

Certain virus variants showed deletions of considerable portions of the genome. The genomes of variants X and XI lacked a 1.5-kbp-long HindIII fragment present in the genome of variant VIII, from which they were derived (data not shown). Accordingly, the length of particles of virus variants X and XI was 780 nm, 50 nm shorter than that of virus variant VIII.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

SIRV1 and SIRV2 cannot be assigned to any of the previously described virus families. Although morphologically these rod-shaped viruses have pronounced similarities with rod-shaped plant viruses, i.e., tobacco mosaic virus, their genome is a double-stranded linear DNA, rather than single-stranded RNA. In contrast to the flexible filamentous Lipothrixviruses, which also have double-stranded linear DNA genomes, SIRV1 and SIRV2 lack an envelope. Taking the unusual combination of their features into account, SIRV1 and SIRV2 have been assigned to a novel virus family, the Rudiviridae (from Latin rudis, thin rod). Because of its stability in many hosts, SIRV2 is a better candidate for the type species than SIRV1.

The length of the particles of the viruses correlates with the genome size: SIRV1, with a genome of 32.3 kbp, is about 70 nm shorter than SIRV2, with a genome of 35.8 kbp; a deletion of a 1.5-kbp fragment from the SIRV1 genome (variants X and XI), resulted in shortening of the particle by about 50 nm.

No proteins were shown to be covalently attached to the termini of the linear DNAs of SIRV1 or SIRV2. However, the DNAs have no free 3' or 5' ends, as demonstrated by a failure to label these by terminal transferase and T4 polynucleotide kinase and by the inability of exonuclease to degrade the DNAs. Both DNAs are, however, degraded from their termini by exonuclease BAL31, which can initiate digestion in single-stranded regions (ZHOU and GREY 1990 ). Sensitivity to nuclease BAL31 was used to identify covalently closed hairpin ends in the DNAs of Tetrahymena (YAO and YAO 1981 ) and Trypanosoma brucei (WILLIAMS et al. 1982 ; BLACKBURN and CHALLONER 1984 ), in linear plasmids of Borellia burgdorferi (BARBOUR and GARON 1987 ), and in the DNA genomes of Chlorella viruses (ROHOZINSKI et al. 1989 ). Taken together, these data indicate the existence of covalently closed hairpin ends in the genomes of SIRV1 and SIRV2.

In their original hosts, KVEM10H3 and HVE10/2, respectively, SIRV1 and SIRV2 were present in poorly maintained carrier states. Although the copy numbers of virus genomes were as high as 20 per host chromosome, both KVEM10H3 and HVE10/2 were easily cured of the viruses upon continued growth. The cured strains were insensitive to reinfection by their original viruses, thus favoring their segregation from virus-carrying cells. The genome of SIRV1 was entirely stable while being slowly propagated in the carrier state. SIRV2, which reached a titer of 3 x 10⁸ per ml, could also be multiplied efficiently and without change in several other S. islandicus strains.

Already after one round of propagation in REN2H1, the DNA restriction fragment pattern of the original SIRV1 had been replaced by a complex pattern indicating a mixture of variants (variant mixture II). Some virus variants, obtained from single plaques generated by the wild-type SIRV1 on REN2H1, yielded stoichiometric patterns of DNA fragments, thus documenting single variants. Two such variants, VIII and XII, represented more than 90% of single plaque-cloned variants. Some plaques yielded viruses with patterns that were not stoichiometric and contained minor bands, thus indicating mixtures of virus variants, e.g., V, XIII. These were still unstable upon propagation. In one case, a variant mixture (XIII) yielded the new stable variant XVII from one single plaque and an unstable variant XVIII (with a dominant DNA fragment pattern closely resembling that of the original virus but with many additional minor bands) from another plaque. Thus, genome variation can result either in the loss of variability coinciding with the selection of novel stable virus strains or in the intermittent formation of still unstable variants.

The nucleotide sequences of three random clones of one DNA fragment from variant mixture II harboring the gene of the DNA-binding protein and adjacent regions differed from each other by as much as 7–10%. One of the cloned sequences was identical to a fragment of the completely sequenced genome of the stable variant VIII (H. BLUM, S. MALLOK, K. BARTLAU, H. DOMDAY, D. PRANGISHVILI and W. ZILLIG, unpublished results). Comparison of the sequences shows most of the point mutations in the genes to be neutral, mainly in the third codon positions, whereas in the noncoding regions the mismatches were statistically distributed (see Figure 7). This is a convincing demonstration of the pressure exerted on variant selection by the necessity to conserve an essential function. Even if the roughly determined mutation rate, 3 x 10^-3 substitutions per nucleotide per replication cycle, is overestimated (see RESULTS), SIRV1 would be among the fastest mutating viruses known. The mutation rate far exceeds that of DNA viruses and is in the range of that of the most rapidly evolving RNA viruses, 10^-3–10^-6 substitutions per nucleotide per replication cycle (reviewed in HOLLAND et al. 1992 ).

SIRV1 can apparently infect REN2H1 but not spread unless it varies by extensive and rapid accumulation of point mutations. The failure to spread unchanged in new hosts appears to have activated a transient mutator system that allows the virus to escape what otherwise would have been a dead-end situation. In a preliminary description of SIRV1, at that time termed SIRV (ZILLIG et al. 1994 ), it was shown that the virus induces the production of SSV1 in SSV1 lysogens, probably by initiating an SOS-like response (see also SINGER 1993 ). The invariant SIRV2 did not effect SSV1 induction in such lysogens. The SOS response in Escherichia coli includes activation of the expression of the mutator genes umuC and umuD (reviewed in WALKER 1984 ). The point mutations in the variants of SIRV1 could be generated by corresponding gene products of an SOS-like system of Sulfolobus, either effecting low-fidelity replication or mutative repair. This should lead to the selection of virus variants adapted to the new host, eventually coinciding with the recovery of high fidelity replication, and thus finally creating novel stable virus strains able to spread in the new host. If so, the case described here is a good example of how SOS-induced mutagenesis could function as a population rescue mechanism, in this case, to recruit a new host and, more generally, to cope with otherwise fatal impediments. Such a function of SOS mutagenesis has been suggested (e.g., see ECHOLS and GOODMAN 1991 ).

Whereas the host-dependent generation of point mutations in the genome of the mutator virus SIRV1 has been clearly demonstrated, the mechanism of this mutagenesis remains to be elucidated. As already indicated above, it could be effected by low-fidelity replication, e.g., via inaccurate proofreading, or by the introduction of point mutations via the generation of ambiguous disparate sequences by faulty sequence substitution resembling repair but with the opposite result. Both mechanisms could be responsible for the estimated extremely high rate of mutation (see RESULTS).

Because the genome of SIRV1 is about 3.5 kbp shorter than that of SIRV2, it appeared possible that variation resulted from the loss of a gene or genes required for high-fidelity replication, e.g., of a replication system involving virus and host elements. Stable SIRV1 variants have, however, acquired the ability to propagate as efficiently as SIRV2 without enlargement of their genomes, more specifically, without the integration of host chromosomal DNA. We therefore consider two alternatives: (1) the gene(s) excluding variation is (are) present in the genome of the mutator virus in a latent form, or (2) a dormant mutator function in a stable virus strain may become activated, possibly spontaneously by a switch (see below), in a situation requiring adaptive evolution, e.g., for recruiting a new host.

SIRV1 variants X and XI derived from variant VIII are products of deletions rather than point mutations and thus represent another mechanism of variation, examples of which have also been observed in studies of other genetic elements in Sulfolobus (PRANGISHVILI et al. 1998 ; SHE et al. 1998 ). The observed reversibility of changes in patterns of DNA fragments of variants could result from point mutations as well as recombinative events, e.g., a switch similar to that changing host specificity of E. coli phage Mu. In the absence of sufficient sequence data, however, this remains mere speculation.

The complete genome of variant VIII of SIRV1 has been determined (H. BLUM, S. MALLOK, K. BARTLAU, H. DOMDAY, D. PRANGISHVILI and W. ZILLIG, unpublished results). The sequencing of SIRV1 variant XVII and SIRV2 genomes is underway. The comparison of the sequences should furnish valuable information concerning the astonishing mutation rate of SIRV1.

	FOOTNOTES

¹ Present address: Epidauros AG, Am Neuland 1, 82347 Bernried, Germany.
² Present address: University of Waikato, Private Bag 3105 Hamilton, New Zealand.
³ Present address: Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands.

	ACKNOWLEDGMENTS

We are grateful to Kenneth M. Stedman for critical reading of the manuscript and help in preparation of figures, to Beatrice Hautzel, Tania Reuter, and Sonja-Verena Albers for assistance in cultivation of different virus variants, and to Prof. Friedrich Lottspeich and Josef Kellermann for the N-terminal protein sequencing. D.P. was supported by a Max Planck Fellowship and the work was supported by the European Union in the frame of its Biotech Program project "Extremophiles as Cell Factories" and the Deutsche Forschungsgemeinschaft.

Manuscript received March 15, 1999; Accepted for publication May 12, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BARBOUR, A. G. and C. F. GARON, 1987  Linear plasmids of the bacterium Borrelia burgdorferi have covalently closed ends. Science 237:409-411[Abstract/Free Full Text].

BATH, C. and M. DYALL-SMITH, 1998  His1, an archaeal virus of the Fuselloviridae family that infects Haloarcula hispanica.. J. Virol. 72:9392-9395[Abstract/Free Full Text].

BIRNBOIM, H. C. and J. DOLY, 1979  A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucleic Acids Res. 7:1513-1523[Abstract/Free Full Text].

BLACKBURN, E. H. and P. B. CHALLONER, 1984  Identification of a telomeric DNA sequence in Trypanosoma brucei.. Cell 36:447-457[Medline].

BROCK, T. D., K. M. BROCK, R. T. BELLY, and R. L. WEISS, 1972  Sulfolobus: a new genus of sulfur oxidizing bacteria living at low pH and high temperature. Arch. Microbiol. 84:54-68.

ECHOLS, H. and M. F. GOODMAN, 1991  Fidelity mechanisms in DNA replication. Annu. Rev. Biochem. 60:477-511[Medline].

FEINBERG, A. P. and B. VOGELSTEIN, 1983  A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 132:6-13[Medline].

HEGERL, R., 1996  The EM program package: a platform for image processing in biological electron microscopy. J. Struct. Biol. 116:30-34[Medline].

HOLLAND, J. J., J. C. DE LA TORRE, and D. A. STEINHAUER, 1992  RNA virus populations as quasispecies. Curr. Top. Microbiol. Immunol. 176:1-20[Medline].

MARTIN, A., S. YEATS, D. JANEKOVIC, W.-D. REITER, and W. AICHER et al., 1984  SAV1, a temperate, UV-inducible DNA virus-like particle from the archaebacterium Sulfolobus acidocaldarius isolate B12. EMBO J. 3:2165-2168[Medline].

PALM, P., C. SCHLEPER, B. GRAMP, S. YEATS, and P. MCWILLIAM et al., 1991  Complete nucleotide sequence of the virus SSV1 of the archaebacterium Sulfolobus shibatae.. Virology 185:242-250[Medline].

PRANGISHVILI, D., S.-V. ALBERS, I. HOLZ, H. P. ARNOLD, and K. STEDMAN et al., 1998  Conjugation in Archaea: frequent occurrence of conjugative plasmids in Sulfolobus.. Plasmid 40:190-202[Medline].

REITER, W.-D., W. ZILLIG, and P. PALM, 1988  Archaebacterial viruses. Adv. Virus Res. 34:134-188.

ROHOZINSKI, J., L. E. GIRTON, and J. L. VAN ETTEN, 1989  Chlorella viruses contain linear nonpermutated double-stranded DNA genomes with covalently closed hairpin ends. Virology 168:363-369[Medline].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SCHÄGGER, H. and G. VON JAGOW, 1987  Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa. Anal. Biochem. 166:368-379[Medline].

SCHLEPER, C., K. KUBO, and W. ZILLIG, 1992  The particle SSV1 from the extremely thermophilic archaeon Sulfolobus is a virus: demonstration of infectivity and of transfection with viral DNA. Proc. Natl. Acad. Sci. USA 89:7645-7649[Abstract/Free Full Text].

SHE, Q., H. PHAN, R. GARRETT, S.-V. ALBERS, and K. STEDMAN et al., 1998  Genetic profile of pNOB8 from Sulfolobus: the first conjugative plasmid from an archaeon. Extremophiles 2:417-425[Medline].

SINGER, T., 1993 Kontrolle der Synthese des UV-induzierten Transkripts T_ind des Virus SSV1. Diplomarbeit der Fakultät für Biologie der Ludwig-Maximilians-Universität, München.

STOLT, P., and W. ZILLIG, 1994 Archaebacterial bacteriophages, pp. 50–58 in Encyclopedia of Virology, edited by R. G. WEBSTER and A. GRANOFF. Academic Press, San Diego.

VARTAPETIAN, A. and A. BOGDANOV, 1987  Proteins covalently linked to viral genomes. Prog. Nucleic Acid Res. Mol. Biol. 34:209-251[Medline].

WALKER, G. C., 1984  Mutagenesis and inducible responses to Deoxyribonucleic acid damage in Escherichia coli.. Microbiol. Rev. 48:60-93[Free Full Text].

WILLIAMS, R. O., J. R. YOUNG, and P. A. O. MAJIWA, 1982  Genomic environment of T. brucei VSG genes: presence of minichromosomes. Nature 299:417-421[Medline].

WOESE, C. R., O. KANDLER, and M. L. WHEELIS, 1990  Towards a natural system of organisms: proposal for the domains Archaea, Bacteria and Eukarya.. Proc. Natl. Acad. Sci. USA 87:4576-4579[Abstract/Free Full Text].

WOOD, A. G., W. B. WHITMAN, and J. KONISKY, 1989  Isolation and characterization of an archaebacterial virus-like particle from Methanococcus voltae.. J. Bacteriol. 171:93-98[Abstract/Free Full Text].

YAO, M.-C. and C.-H. YAO, 1981  Repeated hexanucleotide C-C-C-C-A-A is present near free ends of macronuclear DNA of Tetrahymena.. Proc. Natl. Acad. Sci. USA 78:7436-7439[Abstract/Free Full Text].

ZHOU, X. G. and H. B. GREY, JR., 1990  Mechanism of exonuclease action of Bal31 nuclease. Biochim. Biophys. Acta 1049:83-92[Medline].

ZILLIG, W., W.-D. REITER, P. PALM, F. GROPP, H. NEUMANN et al., 1988 Viruses of Archaebacteria, pp. 517–568 in The Bacteriophages, Vol. 1, edited by R. CALENDER. Plenum Press, New York.

ZILLIG, W., A. KLETZIN, C. SCHLEPER, I. HOLZ, and D. JANEKOVIC et al., 1994  Screening for Sulfolobales, their plasmids, and their viruses in Icelandic solfataras. Syst. Appl. Microbiol. 16:609-628.

ZILLIG, W., D. PRANGISHVILI, C. SCHLEPER, M. ELFERINK, and S.-V. ALBERS et al., 1996  Viruses, plasmids and other genetic elements of thermophilic and hyperthermophilic archaea. FEMS Microbiol. Rev. 18:225-236[Medline].

ZILLIG, W., H. P. ARNOLD, I. HOLZ, D. PRANGISHVILI, and A. SCHWEIER et al., 1998  Genetic elements in the extremely thermophilic archaeaon Sulfolobus.. Extremophiles 2:131-140[Medline].

This article has been cited by other articles:

				A. C. Ortmann, S. K. Brumfield, J. Walther, K. McInnerney, S. J. J. Brouns, H. J. G. van de Werken, B. Bothner, T. Douglas, J. van de Oost, and M. J. Young Transcriptome Analysis of Infection of the Archaeon Sulfolobus solfataricus with Sulfolobus Turreted Icosahedral Virus J. Virol., May 15, 2008; 82(10): 4874 - 4883. [Abstract] [Full Text] [PDF]

				G. Vestergaard, R. Aramayo, T. Basta, M. Haring, X. Peng, K. Brugger, L. Chen, R. Rachel, N. Boisset, R. A. Garrett, et al. Structure of the Acidianus Filamentous Virus 3 and Comparative Genomics of Related Archaeal Lipothrixviruses J. Virol., January 1, 2008; 82(1): 371 - 381. [Abstract] [Full Text] [PDF]

				J. C. Snyder, B. Wiedenheft, M. Lavin, F. F. Roberto, J. Spuhler, A. C. Ortmann, T. Douglas, and M. Young Virus movement maintains local virus population diversity PNAS, November 27, 2007; 104(48): 19102 - 19107. [Abstract] [Full Text] [PDF]

				C. Geslin, M. Gaillard, D. Flament, K. Rouault, M. Le Romancer, D. Prieur, and G. Erauso Analysis of the First Genome of a Hyperthermophilic Marine Virus-Like Particle, PAV1, Isolated from Pyrococcus abyssi J. Bacteriol., June 15, 2007; 189(12): 4510 - 4519. [Abstract] [Full Text] [PDF]

				G. Erauso, K. M. Stedman, H. J. G. van de Werken, W. Zillig, and J. van der Oost Two novel conjugative plasmids from a single strain of Sulfolobus Microbiology, July 1, 2006; 152(7): 1951 - 1968. [Abstract] [Full Text] [PDF]

				K. Namba, K. Hagiwara, H. Tanaka, Y. Nakaishi, K. T. Chong, E. Yamashita, G. E. Armah, Y. Ono, Y. Ishino, T. Omura, et al. Expression and Molecular Characterization of Spherical Particles Derived from the Genome of the Hyperthermophilic Euryarchaeote Pyrococcus furiosus J. Biochem., August 1, 2005; 138(2): 193 - 199. [Abstract] [Full Text] [PDF]

				G. Rice, L. Tang, K. Stedman, F. Roberto, J. Spuhler, E. Gillitzer, J. E. Johnson, T. Douglas, and M. Young From The Cover: The structure of a thermophilic archaeal virus shows a double-stranded DNA viral capsid type that spans all domains of life PNAS, May 18, 2004; 101(20): 7716 - 7720. [Abstract] [Full Text] [PDF]

				C. Geslin, M. Le Romancer, G. Erauso, M. Gaillard, G. Perrot, and D. Prieur PAV1, the First Virus-Like Particle Isolated from a Hyperthermophilic Euryarchaeote, "Pyrococcus abyssi" J. Bacteriol., July 1, 2003; 185(13): 3888 - 3894. [Abstract] [Full Text] [PDF]

				G. Rice, K. Stedman, J. Snyder, B. Wiedenheft, D. Willits, S. Brumfield, T. McDermott, and M. J. Young Viruses from extreme thermal environments PNAS, October 16, 2001; (2001) 231170198. [Abstract] [Full Text] [PDF]

				W. B. Whitman, F. Pfeifer, P. Blum, and A. Klein What Archaea Have to Tell Biologists Genetics, August 1, 1999; 152(4): 1245 - 1248. [Full Text]

				G. Rice, K. Stedman, J. Snyder, B. Wiedenheft, D. Willits, S. Brumfield, T. McDermott, and M. J. Young Viruses from extreme thermal environments PNAS, November 6, 2001; 98(23): 13341 - 13345. [Abstract] [Full Text] [PDF]