Snapshot of virus evolution in hypersalineenvironments from the characterization of amembrane-containing Salisaeta icosahedral phage 1Antti P. Aaltoa,1, David Bittob, Janne J. Ravanttia, Dennis H. Bamforda, Juha T. Huiskonenb, and Hanna M. Oksanena,2

aInstitute of Biotechnology and Department of Biosciences, Biocenter 2, University of Helsinki, FI-00014, Helsinki, Finland; and bOxford Particle ImagingCentre, Division of Structural Biology, Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford OX3 7BN, United Kingdom

Edited* by James L. Van Etten, University of Nebraska, Lincoln, NE, and approved March 20, 2012 (received for review December 12, 2011)

The multitude of archaea and bacteria inhabiting extreme environ-ments has only become evident during the last decades. As virusesapply a significant evolutionary force to their hosts, there is an in-herent value in learning about viruses infecting these extremophiles.In this study, we have focused on one such unique virus–host pairisolated fromahypersaline environment: an icosahedral,membrane-containing double-stranded DNA virus—Salisaeta icosahedral phage1 (SSIP-1) and its halophilic host bacterium Salisaeta sp. SP9-1 closelyrelated to Salisaeta longa. The architectural principles, virion compo-sition, and the proposed functions associatedwith some of the ORFsof the virus are surprisingly similar to those found in viruses belong-ing to the PRD1–adenovirus lineage. The virion structure, determinedby electron cryomicroscopy, reveals that the bulk of theouter proteincapsid is composed of upright standing pseudohexameric capsomersorganized on a T = 49 icosahedral lattice. Our results give a compre-hensive description of a halophilic virus–host system and shed lighton the relatedness of viruses based on their virion architecture.

virus structure | PRD1-like viruses | ORFan

Viruses are the most abundant depositories of nucleic acid-encoded information in the biosphere, and they outnumber

their hosts by at least an order of magnitude. It has been sug-gested that viruses were the first compartmentalized, self-repli-cating entities on Earth, and that the last universal commonancestor was already infected by numerous viruses using a varietyof assembly principles (1–3). As it is impossible to directly ob-serve the early stages of emergence of life, present day virusescan help to delineate the plausible evolutionary pathways thatled from the primordial soup to the amazing diversity in whichlife is manifested today. A prerequisite to this approach is tocarefully sample different ecological niches and attempt tocharacterize the organisms dwelling in them (4).Organisms inhabiting hypersaline environments are called halo-

philic (for “salt loving”). There are many ways to define halophilicorganisms, but often they can be classified as growing optimally at 50g/L of salt or higher or tolerating at least 100 g/L of salt. Extremelyhalophilic organisms can tolerate salt concentrations approachingsaturation (∼360 g/L) (5). Although hypersaline environmentssupport organisms from all domains of life, their inhabitants havebeen poorly studied. Importantly, life in extreme environments hasbeen suggested to reflect conditions on the “early” Earth, encour-aging the study of extremophiles to understand evolution (6).The vast majority of halophilic viruses described to date infect

archaea, and they fall into a number of morphotypes with capsidarchitectures ranging from tailed and tailless icosahedrally sym-metric viruses to pleomorphic viruses lacking well-defined shapes(7, 8). In contrast, very little is known about halophilic viruses thatinfect bacterial hosts (9–12). This bias stems from the fact thatbacteria from hypersaline environments have remained under-studied (13). However, during the last few decades the number ofknown halophilic bacterial species has increased rapidly, due tomore thorough sampling and the introduction of metagenomicapproaches (5). Consequently, to better understand the rules and

interactions that constitute the virus–host ecology and evolution inhypersaline biotopes, bacterial viruses and their hosts have to bebrought into the limelight.In this study, we introduce a halophilic icosahedral virus, Sali-

saeta icosahedral phage 1 (SSIP-1), which exemplifies a taillessbacteriophage thriving in a hypersaline environment. The SSIP-1capsid consists mainly of pseudohexameric capsomers followinga T = 49 triangulation and encloses a membrane, whose lipids areselectively derived from the host. The membrane further enclosesa circular dsDNA genome. Sequencing of the genome and sub-sequent bioinformatics and proteomic analyses of the virion pro-teins revealed several previously unidentified genes, ORFans. Thelife cycle of SSIP-1 was determined to be lytic. However, the dis-covery of putative integrase and repressor/antirepressor genessuggests that there may also be a temperate phase in the life cycle.The discovery of SSIP-1 adds a piece to the puzzle of icosahedraltailless viruses with an inner membrane. These viruses form thePRD1–adenovirus structural lineage whose members share uprightdouble β-barrel major capsid proteins (MCPs) and a similar virionarchitecture (1, 14–19). Such viruses infect cells belonging to bac-teria, archaea, or eukaryotes supporting the view that these virusesare ancient and share a common origin. PRD1-like viruses canclearly be divided into those having an MCP with the canonicaldouble β-barrel fold [e.g., PM2 (20), vaccinia virus (21), PRD1 (15),PBCV-1 (22), STIV (23), and adenovirus (24)] and to those witha more complex capsid arrangement with two MCPs [SH1 (25),P23-77 (26), and HHIV-2 (27)]. So far, the latter group has beenfound infecting only organisms from extreme conditions.

Results and DiscussionSSIP-1 Is a Virulent Bacteriophage Infecting Rod-Shaped Salisaeta sp.SP9-1 from High Salinity. During our search for new prokaryotesand their viruses from extremely halophilic environments (11),a single virus isolate designated SSIP-1 and its host bacteriumSalisaeta sp. SP9-1 (SP for Sedom ponds) were found in a watersample from experimental ponds at Sedom, Israel, where waterfrom the Dead Sea is diluted with water from the Red Sea. Thesemesocosms simulate the environmental effects of regulating thelevels of the Dead Sea with seawater (28).

Author contributions: A.P.A., J.J.R., D.H.B., and H.M.O. designed research; A.P.A. and J.J.R.performed research;A.P.A., J.J.R., andH.M.O. analyzeddata;D.B. and J.T.H. performedelectroncryo-microscopy reconstruction; and A.P.A., D.B., D.H.B., J.T.H., and H.M.O. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.

Data deposition: The sequences reported in this paper have been deposited in the Gen-Bank database (accession no. JN880423) and the Electron Microscopy Data Bank at theEuropean Bioinformatics Institute (accession no. EMD-2061).1Present address: Division of Biological Sciences, University of California, San Diego, LaJolla, CA 92093.

2To whom correspondence should be addressed. E-mail:hanna.oksanen@helsinki.fi.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1120174109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1120174109 PNAS | May 1, 2012 | vol. 109 | no. 18 | 7079–7084

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020

SP9-1 cultures displayed a bright red color and their turbidityincreased more rapidly as the salinity of the growth medium waslowered (Fig. S1A). A partial 16S rRNA gene sequence of SP9-1 ishighly similar (99.6%) to that of Salisaeta longa (11). S. longa isa Gram-negative, rod-shaped, halophilic bacterium (Bacteroidetes)previously isolated from the Sedom ponds (29). Indeed, only a fewdifferences could be detected in the whole-cell protein patterns ofSP9-1 and S. longa (Fig. S1B). Moreover, under the electron mi-croscope, SP9-1 (Fig. 1A) displayed a similar morphology to that ofS. longa (Fig. 1B).The growth phase of the host has an effect on the plating effi-

ciency (EOP): using logarithmic and stationary phase cultures theEOPs were 2 × 109 and 4 × 108 pfu/mL, respectively. Furthermore,SSIP-1 was unable to induce plaque formation in any of the otherhalophilic bacteria tested, including S. longa, Salisaeta sp. SP10-1,and five Salinibacter strains (Table S1). In the single-step growth

experiment, the turbidity of the SSIP-1–infected SP9-1 culturestarted to decline after 15 h postinfection (p.i.) (Fig. 1C). At thesame time infectious viruses began to accumulate in the superna-tant, indicating that in these conditions SSIP-1 lyses its host.Electronmicrographs of thin-sectioned SSIP-1–infected SP9-1 cellssupported these findings by revealing virus-sized particles accu-mulating within ruptured cells at 17 h p.i. (Fig. 1D). SSIP-1 boundspecifically, but rather slowly, onto SP9-1 cells with an approximateadsorption rate of 7.0 × 10−9 mL/min (Fig. 1E). SSIP-1 was unableto bind onto S. longa, suggesting that the receptor seems to bespecific to SP9-1 despite the close relatedness between the twospecies (Fig. 1E and Fig. S1B).Optimal salinitywas critical forSSIP-1 virion and its life cycle.The

infectivity of SSIP-1 decreased dramatically when the virus was ex-posed to lowered salinity (Fig. 1F). The virus particles remainedinfectious in 9% (wt/vol) salt water (SW) (see composition in TableS2) for 3 h. However, for longer periods, higher salinity [18% (wt/vol) SW;Table S2]was needed to retain infectivity.Moreover, SSIP-1 was unable to produce plaques when the salt concentration of thegrowth medium was lowered below 19% (wt/vol) SW (Table S2),although the virion remains infectious in these conditions. TheEOPalso increased with the increasing salinity. These results are inconcert with amodel where low salinity affects the adsorption of thevirus onto the host by impairing receptor binding. It is also feasiblethat the low salinity modifies the biophysical properties of the hostcell membrane or the functions of some of the enzymes critical forvirus entry or exit.

Predicted Functions of SSIP-1 ORFs Suggest Sophisticated Nucleic AcidChemistry and the Ability to Integrate into the Host Genome. TheSSIP-1 genome is a nonmethylated, circular, dsDNA molecule of43,788 bp with a GC content of 57.2% (Fig. 2 and Fig. S1C). Thesecond adenine (underlined) of the unique EcoRV cleavage site(5′-GATATC-3′) was assigned as the first nucleotide. We in-cluded all of the ORFs that encode a protein longer than 40 aa.In the following text, an ORF that is confirmed by proteomics(see below) to encode for a gene product (gp) is referred to asa gene. In total, there are 57 nonoverlapping and tightly packedORFs putatively encoding polypeptides with variable sizes (48–2,320 residues) and having GC contents between 47.5 and 67.4%(Fig. 2 and Table S3). The ORFs and genes are organized inblocks with alternating directions (at least four operons), sug-gesting temporal transcription regulation during different stagesof the virus life cycle (Fig. 2). Thirteen of the hypothetical pro-teins have predicted transmembrane helices (Table S3).Most of the SSIP-1 ORFs (38 of 57) were ORFans i.e., ORFs

that have no detectable sequence similarity to other sequences inthe databases (30, 31). As more and more viral genomes are beingsequenced, the abundance ofORFans seems to be becoming a rulerather than an exception (31). Viruses have traditionally beenregarded as pickpockets of cellular genes, and the presence of viralORFs having no apparent homologs has been thought to be anindication of insufficient sampling of the sequence space. How-ever, it has recently been suggested that ORFans may actuallyrepresent genes that are of “viral origin,” which emphasizes thepossibility that viruses are a major driving force of evolution (3).We obtained highly significant homology scores for 19 ORFs

(Table S3). The predicted functions of many of these ORFswere related to nucleic acid chemistry, e.g., (i) DNA invertase,resolvase, and transposase; (ii) RNA polymerase σ70 subunit;and (iii) AAA ATPase or primase homologs were detected(Table S3). The structural protein gp40 harbors the canonicalWalker A and B motifs and signatures of the P9/A32-specificmotif found in the packaging proteins of tailless membrane-containing icosahedral viruses, e.g., PRD1 (32–34). The giantvirion structural protein gp43 (262 kDa) contains several distinctdomains that reveal similarities to DNA methyltransferase, typeI site-specific DNase, and the chromosome partitioning protein

Fig. 1. SSIP-1 is a lytic bacteriophage infecting Salisaeta sp. SP9-1. (A and B)Thin-section electron microscopic image of (A) SP9-1 and (B) Salisaeta longa.White areas are holes causedby thehigh salt concentration. (Scale bar inB, 1 μmfor A and B.) (C) SSIP-1 life cycle. Optical density of SSIP-1 infected [usinga multiplicity of infection (MOI) of 10; open circles] and uninfected (closed cir-cles) SP9-1 cultures was observed for 30 h. At 6 h to 17 h postinfection (p.i.)samples were collected, the bacteria pelleted by centrifugation, and thesupernatants assayed for infective viruses (gray bars). (D) Thin-section electronmicroscopic images of SSIP-1 infected SP9-1 cell at 17 h p.i. containing virusparticles (arrows). (Scale bar, 200 nm.) (E) Binding of SSIP-1 on SP9-1 (closedcircles) and S. longa (open circles). For control, no cells were added (triangles).After the removal of bound viruses and cells, the supernatants were assayed forinfective free viruses (n = 3). Error bars indicate SEM. (F) SSIP-1 was incubated atthe indicated salt water (SW,%wt/vol) buffers for 3 h (closed circles), 1 d (opencircles), and 7 d (triangles), and the infectivity was assayed.

7080 | www.pnas.org/cgi/doi/10.1073/pnas.1120174109 Aalto et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020

ParB (Table S3). These data suggest that SSIP-1 is capable ofparticipating in RNA and DNA metabolism. Gene 43 is con-siderably larger than any of the other ORFs within the genomeand could be an example of horizontal gene transfer. As no DNAor RNA polymerase could be found within the SSIP-1 genome, itis possible that gp43 might be a sophisticated molecular machineimportant for the replication and/or transcription of the genome.A few putative genes seem to have predicted functions in virus

entry or exit (lysozyme g and an endoglucanase-related protein;Table S3). Although SSIP-1 obeys a strictly virulent lifestyle inSP9-1 (Fig. 1C), the SSIP-1 genome contains genes homologousto a phage repressor, an antirepressor, a putative integrase, andsite-specific recombinase (Table S3). It can be envisioned thatthese proteins, together with other unidentified components,enable the virus to integrate into the host genome. To detectpossible host genome integration, we performed colony PCR onuninfected SP9-1 and S. longa cultures using several SSIP-1–specific oligos, but were unable to detect any evidence of thepresence of SSIP-1. It is conceivable that integration of SSIP-1takes place in specific conditions or with another host. An al-ternative scenario is that the recombination machinery of SSIP-1has undergone mutations and is no longer functional.

Analysis of Putative Moron-Like Elements. It seems that horizontalgene transfer is possible between viruses infecting phylogeneti-cally distant hosts that inhabit completely different ecologicalniches as demonstrated, e.g., head–tail viruses ΦCh1 and ΦH(35). We noticed that SSIP-1 ORF 25 is strikingly similar (63.7%similarity; Table S3) to P23-77 ORF 3 [National Center forBiotechnology Information (NCBI) ID: YP_003169710]. P23-77

is a tailless icosahedral, internal membrane-containing dsDNAbacteriophage that infects Thermus thermophilus growing opti-mally above 70 °C (26). SSIP-1 ORF 25 is located (together withORF 24) in an area of the genome that is flanked by ORF-freesequences (Fig. 2). The area is moron-like (a recent geneticaddition) (36, 37), because the GC content of this region is muchlower than its neighboring ORFs (Table S3). Also other SSIP-1ORFs have considerably lower GC contents than their neighbors(e.g., ORFs 2, 16, and 32; Table S3). To emphasize the possibilitythat SSIP-1 could have moron-like regions with their own pro-moter and termination sites enabling autonomous transcription(36), we analyzed the genome for Rho-independent transcriptionterminators, σ70-promoter regions and tandem repeat sequences

Fig. 2. The genome of SSIP-1 is a circular dsDNA molecule with 57 predictedORFs. Inner graph indicates the GC profile of the genome. Predicted ORFs andgenes (1–57) are seen on the outer circle. ORFs in the forward and reversedirections are colored in blue and green, respectively. Gene products that havebeen confirmed to be structural proteins of the virion aremarked in red (Fig. 3Cand Table S7). Unique restriction enzyme cleavage sites are indicated in theoutermost circle. Tandemrepeat sequenceswere located inORF27 (3.5×GAGTG-GAACACCCGCGGAACAGT, 4.6×GGAACAGTCGT), ORF 29 (2.2× CTCCGCCAGCA-GAAGAAAGAG), and ORF 52 (2.4× CGGTGGTGGCGGCGGTAATCCCGGCGGTGG-CTA). (Numbers before the tandem repeat sequences indicate how many timesthe corresponding sequence is repeated.) Red lollipops indicate predicted termi-nator sequences and putative σ70-promoter regions are shown by gray arrows.

Fig. 3. Lipids and structural proteins of SSIP-1. (A) SSIP-1 purified to nearhomogeneity. Graph indicates absorbance (open circles), density (closedcircles), and infectivity (bars) of the CsCl gradient equilibrium centrifugationfractions from the final purification step. Only the virus band-containingfractions are shown. (B) Extracted polar lipids of SSIP-1, Salisaeta sp. SP9-1,Salisaeta longa and Salinibacter ruber were analyzed by TLC followed byiodine vapor staining. Major lipid bands that differ quantitatively betweenSP9-1 and SSIP-1 are indicated by arrows. (C) SSIP-1 was analyzed by SDS-PAGE. “M” is a molecular marker (Fermentas; SM0661). Major protein bandsthat were subjected to N-terminal protein sequences are indicated by anasterisk and the ones subjected to mass spectrometry are indicated bya hashmark. Major capsid proteins (MCPs) are indicated by arrows. Geneproducts (gp) determined by proteomics are given (Table S7 and Fig. 2).

Aalto et al. PNAS | May 1, 2012 | vol. 109 | no. 18 | 7081

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020

that might function as recognition sites for site-specific nucleases orintegrases. Nine potential terminator sequences were predicted,and a high confidence value was assigned to four of these (TableS4). Intriguingly, three of the plausible terminators are locateddownstream of the operons containing ORFs 13, 23, and 29,whereas the fourth lies in between genes 46 and 47 (Fig. 2). Indeed,SSIP-1 ORF 25 is surrounded by putative transcription terminators(Table S4). Also, many of the tandem repeats found within thegenome are located around this area, which might suggest a pre-disposition for recombination events (Fig. 2). In addition, this areacontains the predicted gene coding for the phage antirepressor.Four distinct σ70-promoter sequences were located, two of which lieupstream of putative ORFs 3 and 15 (Table S5). Although some ofthe promoters might be functional, it is probable that SSIP-1mostlyemploys transcription factors that are different from σ70. Most ofthe putative moron-like elements existed as single ORFs in the partof the SSIP-1 genome that contains the genes encoding for non-structural proteins. This suggests that several (recent) sequentialrecombination events have shaped the genome and probably of-fered selective benefit for the virus.

Internal Membrane of SSIP-1 Virion Originates from Halophilic HostLipids. SSIP-1 virions were purified to near homogeneity by ratezonal and equilibrium centrifugations (Table S6) yielding particleswith a specific infectivity of 1.2 × 1012 pfu/mg of protein. The lowdensity of the purified virions (1.35 g/mL inCsCl; Fig. 3A) suggestedthe presence of lipids (38), which was confirmed by lipid extractionand TLC (Fig. 3B). No major differences in the lipid compositionscould be detected between SP9-1, S. longa, or Salinibacter ruber (13)(Fig. 3B). The membranes of S. longa and S. ruber consist mostly ofphosphatidylcholine, phosphatidylethanolamine, cardiolipin, glyco-lipid, and sulfonolipids (39, 40). In addition, phosphatidylglyceroland phosphatidylserine are present in S. ruber but not in S. longa.The sulfonolipids (halocapnines) are characteristic of halophilicmembers of Bacteroidetes and might in part support the halophiliclifestyle of these organisms (39). The lipid pattern of SSIP-1 wasqualitatively similar to that of SP9-1, indicating that the virusmembrane is derived from the host (Fig. 3B). However, SSIP-1seems to display a certain degree of selectivity in its lipid in-corporation (Fig. 3B). This phenomenon has been observed also inbacteriophage Bam35, in which the transmembrane protein com-plexes modulate the viral internal membrane curvature and thick-ness, providing a possible mechanism for lipid selectivity duringvirion assembly (41). Such amechanismmay also function in SSIP-1.The genes coding for the structural proteins of SSIP-1 (Fig. 3C

and Table S7) were mostly located in the same genomic region(Fig. 2). The two most intense protein species (∼25 kDa) in thevirion corresponded to the proteins gp45 and gp46. The calcu-lated masses and the abundance suggest that they represent theMCPs (Figs. 2 and 3C). Six of the structural proteins (gp1, gp34,gp37, gp40, gp41, and gp43) gave a significant BLAST score toother sequences (Table S3). However, no homologs were detectedfor the MCPs gp45 and gp46, nor are these proteins homologousto each other. It is possible that in this case, sequence-basedmethods are unsuitable for detecting homologous proteins, asonly their folds, but not the primary sequence, seem to be con-served. However, when the crystal structures are solved they maywell reveal similarities to other viruses (16, 17).

Structure of the SSIP-1 Virion Reveals a T = 49 Capsid Arrangement.We used electron cryomicroscopy and 3D image reconstructionto determine the structure of the SSIP-1 virion (Fig. 4). High

A

B

C

E F

D

D

M

C

***

Fig. 4. Icosahedral reconstruction of the SSIP-1 virion. (A) Electronmicrograph ofSSIP-1 virions vitrified at 9% (wt/vol) salt water (SW) buffer (see composition inTable S2). Three spikes are indicatedwith black arrowheads (Scale bar, 50 nm.) (B)Central slice throughan icosahedral reconstruction. Inset showsa radially averageddensity profile. DNA (D), membrane (M), capsid (C), and spikes (S) are indicated.Twofold, threefold, andfivefold axes of icosahedral symmetry are indicated by anellipse, a triangle, andapentagon, respectively. Three concentric layers ofDNAareindicated with asterisks. Lipid bilayer is interrupted by transmembrane densitiesat the threefold axes of symmetry (triangle). (C) Radially colored isosurface rep-resentation of the reconstruction with an arbitrary handedness is rendered at 2σabove the mean density. Color bar shows radial coloring. Inset shows a modellatticeexemplifyingtheT=49 icosahedral triangulation.Geometricalarrangement

of the capsomers is given by the relationship T = h2 + hk + k2, where h and kdefine the lattice point. Here h = 7, k = 0. The two frontmost fivefold verticesare in red. (D–F) Six times magnified close-ups of the reconstruction takenalong the (D) twofold, (E) threefold, and (F) fivefold axes of symmetry.

7082 | www.pnas.org/cgi/doi/10.1073/pnas.1120174109 Aalto et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020

salinity hampers electron cryomicroscopy by reducing contrast.However, samples plunge-frozen immediately after diluting to9% SW (see composition in Table S2) had enough contrast forimaging. Furthermore, the structural integrity of the virions wasretained (Figs. 1F and 4A).Icosahedral reconstruction calculated using 2,747 particles

yielded a structure of the virion at 12.5-Å resolution (Fig. 4 B–Fand Fig. S2A). The dimensions of the viral capsid are 106 nm(vertex to vertex), 97 nm (facet to facet), and 95 nm (edge toedge). The capsid exhibits icosahedral symmetry and the cap-somers are organized on a T = 49 icosahedral lattice (h = 7, k =0; Fig. 4C). A thin fiber-like density (at least 10 nm in length) wasoccasionally seen extending from icosahedral vertices (Fig. 4A).However, these were less pronounced in the icosahedral re-construction than in the original images, possibly due to flexi-bility. Six-coordinated capsomers (8 per asymmetric unit, 480 intotal) make up most of the capsid shell. Only the 12 icosahedralvertexes are occupied by 5-coordinated capsomers, termed“pentons.” Interestingly, there are two types of 6-coordinatedcapsomers, “capped” and “uncapped.” Most of the capsomers (7per asymmetric unit, 420 in total) display a trimeric cap featureon top of a base with sixfold or pseudosixfold symmetry (Fig. 4D–E). Fitting of the canonical double β-barrel capsid protein ofbacteriophage PM2 (20) to the SSIP-1 icosahedral lattice (Fig.S2B) reveals a tight fit supporting the presence of a hexagonalbuilding block in the base of these capsomers as seen previouslyin PRD1-like viruses (15). The cap feature is lacking in the 5capsomers surrounding the penton, termed “peripentonal” cap-somers (1 per asymmetric unit, 60 in total) (Fig. 4F). Instead, sixlobes forming most of the density are evident. Furthermore, theshape of these peripentonal capsomers appears asymmetric orthreefold rather than sixfold.The capsid encloses a lipid core, which is 80 nm in diameter

(Fig. 4B). The lipid core harbors the dsDNA genome, engulfedby the membrane. Several weak densities, especially under theicosahedral vertices and possibly corresponding to peripheral orintegral membrane proteins, are sandwiched between the lipidbilayer and the capsid. Notable exceptions are the icosahedraltwofold positions, at which such densities are absent. At thesepositions, the capsomers, one at each side of the twofold axis ofsymmetry, are kinked toward the membrane and may interactdirectly with it (Fig. 4 B and D). The membrane is ∼5 nm thickand interrupted at some locations by densities, possibly corre-sponding to transmembrane regions of lipid core proteins. Threeconcentric shells of DNA with an average spacing of 2.0 nm weredetected under the membrane (Fig. 4B). The average packingdensity of the DNA was calculated to be 0.29 bp/nm3.

Evolutionary Perspectives. Several icosahedral, tailless dsDNAviruses have been proposed to belong to the same structural viruslineage with PRD1 (for a recent review, ref. 14). Instead of theirrespective host or genome type, the classification is based on thefold of their MCPs and virion architecture. Common to allPRD1-like viruses is their icosahedral capsid composed of cap-somers with pseudohexameric bases. In addition, the icosahedralvertices are occupied by penton proteins and extended vertexstructures. The capsid covers a lipid bilayer enclosing the dsDNAgenome. The only exception is adenovirus, where the membraneis absent. The six-coordinated capsomers consist of six verticalβ-barrels, which in PRD1 are in the form of a trimer of doubleβ-barrels (15). However, recent findings suggest that the lineageof PRD1-like viruses is divided into two subgroups, those withone MCP, such as PRD1, and those with two (P23-77, SH1, andHHIV-2) (25–27). Interestingly, thermophilic P23-77 infectsbacteria, whereas extremely halophilic SH1 and HHIV-2 infectarchaea. The SSIP-1 structure (Fig. 4 B and C) shares all thosecommon features with PRD1-type viruses. Additionally, SSIP-1harbors two MCPs akin to P23-77, SH1, and HHIV-2, although

the T = 49 capsid arrangement is specific to SSIP-1. On the basisof these similarities, SSIP-1 is a putative member of the PRD1-like viruses, but belonging to the extremophilic subgroup withtwo MCPs.Over 1031 viruses reside in the biosphere exceeding the num-

ber of their hosts by an order of magnitude. They exert a massiveselective pressure on their hosts on a global scale (42–44), whichmeans that strong evolutionary forces shape the viral and hostgenomes. However, the known virion structures are based ona limited number of architectural principles (1, 14, 16–19) andconsequently the enormous sequence diversity is reduced toa limited structure space. How many different viral structurallineages might there be? In our recent global search of some 50independently acquired unique viruses, we discovered only 1unique structural type not previously described (11). Obviouslysampling of more viral structures is the only way forward to testour hypothesis. Here, we have taken a small step forward tobring more order to the viral universe.

Materials and MethodsViruses, Bacteria, Media, and Growth Conditions. SSIP-1 (11) and bacteria(Table S1) were grown aerobically at 37 °C in modified growth media (MGM)with varying salt concentrations (45). SSIP-1 grown on Salisaeta sp. SP9-1 waspurified by rate zonal and equilibrium centrifugation. For details, see SIMaterials and Methods, Viruses, Bacteria, Media, and Growth Conditionsand SI Materials and Methods, Propagation and Purification of SSIP-1.

Life Cycle of SSIP-1. Logarithmically growing Salisaeta sp. SP9-1 (1.2 × 108 cfu/mL) collected by centrifugation (5,112 × g, 15 min, 20 °C) was infected byresuspending the cells in the same volume of SSIP-1 virus stock (22 °C) toobtain a multiplicity of infection (MOI) of ∼10. Four hours postinfection thecells were collected (see above) and resuspended in fresh media to removeunbound viruses. A noninfected control culture was treated similarly. Par-allel with turbidity measurements (OD550), samples were taken for trans-mission electron microscopy and free viruses were assayed at 1-h intervals.To determine the number of free viruses, the cells were removed by cen-trifugation (13,800 × g, 5 min, 22 °C), and the viruses in the supernatantwere determined by plaque assay. For a detailed description of the EManalysis and adsorption assay, see SI Materials and Methods, Thin-SectionElectron Microscopy and SI Materials and Methods, Adsorption Assays.

Protein and Lipid Analysis. The proteins were resolved by SDS polyacrylamideelectrophoresis (SDS-PAGE) using polyacrylamide or modified polyacryla-mide-tricine-SDS gels (46, 47), and the structural proteins were either iden-tified by N-terminal sequencing performed by degradative Edman chemistryor mass spectrometry (Protein Chemistry Core Laboratory, Institute of Bio-technology, University of Helsinki). For details, see SI Materials and Methods,Protein Identification. Extracted lipids were analyzed by TLC as described inSI Materials and Methods, Lipid Analysis.

Sequencing and Annotation of SSIP-1 Genome. The genome of SSIP-1 wassequenced by conventional Sanger sequencing (LGC Genomics), and analyzedusing Geneious software (48). For a detailed description, see SI Materials andMethods, Sequencing and Annotation of SSIP-1 Genome.

Colony PCR. The possible integration of SSIP-1 into bacterial genomes wasstudied by a modified colony PCRmethod using SP9-1 and S. longa cultures atdifferent growth phases as templates. Purified SSIP-1 genome and mQ-H2Owere used as positive and negative controls, respectively. See SI Materialsand Methods, SSIP-1 Colony PCR for primer sequences.

Electron Cryomicroscopy and Icosahedral Reconstruction of SSIP-1 Virions.Micrographs of plunge-frozen samples in 9% (wt/vol) SW were recordedusing a 300-kV transmission electronmicroscope (F30 Polara; FEI), operated at59,000× nominal magnification and at liquid nitrogen temperature in low-dose conditions on a CCD camera (Ultrascan 4000; Gatan) leading to a cali-brated pixel size of 2.0 Å. A low-resolution initial model of the virion wascalculated from class averages in IMAGIC (Image Science) and the re-construction was refined in EMAN (49), using 3,564 virion images. The res-olution was determined using Fourier shell correlation at 0.5 threshold. Fora detailed description, see SI Materials and Methods, Electron Cryomicro-scopy and Icosahedral Reconstruction of SSIP-1 Virions.

Aalto et al. PNAS | May 1, 2012 | vol. 109 | no. 18 | 7083

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020

ACKNOWLEDGMENTS. The authors thank S. Korhonen and S. Stormanfor technical assistance, B. Fartmann for sequencing, N. Kalkkinen andG. Rönnholm for their help with the protein identification, and A. Oren forproviding the halophilic strains. This work was supported by the Academy ofFinland Centre of Excellence Program in Virus Research Grant 11296841 (toD.H.B.), Academy Professor (Academy of Finland) Funding Grants 256197

and 256518 (to D.H.B.), the Academy of Finland Grants 130750 and 218080(to J.T.H) and 127665 (to H.M.O.), and the Wellcome Trust Core Award Grant(090532/Z/09/Z to the Wellcome Trust Centre for Human Genetics). We alsothank University of Helsinki for the support to the Instruct Associate Centrefor Virus Production and Purification (The European Strategy Forum on Re-search Infrastructures) used in this study.

1. Bamford DH (2003) Do viruses form lineages across different domains of life? ResMicrobiol 154:231–236.

2. Jalasvuori M, Bamford JK (2008) Structural co-evolution of viruses and cells in theprimordial world. Orig Life Evol Biosph 38:165–181.

3. Forterre P, Prangishvili D (2009) The origin of viruses. Res Microbiol 160:466–472.4. Holmes EC (2011) What does virus evolution tell us about virus origins? J Virol 85:

5247–5251.5. Oren A (2008) Microbial life at high salt concentrations: Phylogenetic and metabolic

diversity. Saline Syst 4:2.6. Martin W, Baross J, Kelley D, Russell MJ (2008) Hydrothermal vents and the origin of

life. Nat Rev Microbiol 6:805–814.7. Pina M, Bize A, Forterre P, Prangishvili D (2011) The archeoviruses. FEMSMicrobiol Rev

35:1035–1054.8. Roine E, Oksanen HM (2011) Viruses from the hypersaline environment. Halophiles

and Hypersaline Environments, eds Ventosa A, Oren A, Ma Y (Springer, Berlin), pp153–172.

9. Kauri T, Ackermann HW, Goel U, Kushner DJ (1991) A bacteriophage of a moderatelyhalophilic bacterium. Arch Microbiol 156:435–438.

10. Kukkaro P, Bamford DH (2009) Virus-host interactions in environments with a widerange of ionic strengths. Environ Microbiol Reports 1:71–77.

11. Atanasova NS, Roine E, Oren A, Bamford DH, Oksanen HM (2012) Global network ofspecific virus-host interactions in hypersaline environments. Environ Microbiol 14:426–440.

12. Calvo C, de la Paz AG, Bejar V, Quesada E, Ramos-Cormenzana A (1988) Isolation andcharacterization of phage F9-11 from a lysogenic Deleya halophila strain. Curr Mi-crobiol 17:49–53.

13. Antón J, et al. (2002) Salinibacter ruber gen. nov., sp. nov., a novel, extremely halo-philic member of the Bacteria from saltern crystallizer ponds. Int J Syst Evol Microbiol52:485–491.

14. Abrescia NGA, Bamford DH, Grimes JM, Stuart DI (2012) Structure unifies the viraluniverse. Annu Rev Biochem 81, in press.

15. Benson SD, Bamford JK, Bamford DH, Burnett RM (1999) Viral evolution revealed bybacteriophage PRD1 and human adenovirus coat protein structures. Cell 98:825–833.

16. Bamford DH, Grimes JM, Stuart DI (2005) What does structure tell us about virusevolution? Curr Opin Struct Biol 15:655–663.

17. Krupovic M, Bamford DH (2008) Virus evolution: How far does the double beta-barrelviral lineage extend? Nat Rev Microbiol 6:941–948.

18. Krupovic M, Bamford DH (2011) Double-stranded DNA viruses: 20 families and onlyfive different architectural principles for virion assembly. Curr Opin Virol 1:118–124.

19. Abrescia NGA, et al. (2010) What does it take to make a virus: The concept of the viral“self”. Emerging Topics in Physical Virology, eds Stockley PG, Twarock R (ImperialCollege Press, London), pp 35–58.

20. Abrescia NG, et al. (2008) Insights into virus evolution and membrane biogenesis fromthe structure of the marine lipid-containing bacteriophage PM2.Mol Cell 31:749–761.

21. Bahar MW, Graham SC, Stuart DI, Grimes JM (2011) Insights into the evolution ofa complex virus from the crystal structure of vaccinia virus D13. Structure 19:1011–1020.

22. Nandhagopal N, et al. (2002) The structure and evolution of the major capsid proteinof a large, lipid-containing DNA virus. Proc Natl Acad Sci USA 99:14758–14763.

23. Rice G, et al. (2004) The structure of a thermophilic archaeal virus shows a double-stranded DNA viral capsid type that spans all domains of life. Proc Natl Acad Sci USA101:7716–7720.

24. Rux JJ, Kuser PR, Burnett RM (2003) Structural and phylogenetic analysis of adeno-virus hexons by use of high-resolution x-ray crystallographic, molecular modeling, andsequence-based methods. J Virol 77:9553–9566.

25. Jäälinoja HT, et al. (2008) Structure and host-cell interaction of SH1, a membrane-containing, halophilic euryarchaeal virus. Proc Natl Acad Sci USA 105:8008–8013.

26. Jaatinen ST, Happonen LJ, Laurinmäki P, Butcher SJ, Bamford DH (2008) Biochemicaland structural characterisation of membrane-containing icosahedral dsDNA bacter-iophages infecting thermophilic Thermus thermophilus. Virology 379:10–19.

27. Jaakkola ST, et al. (2012) Closely related archaeal Haloarcula hispanica icosahedralviruses HHIV-2 and SH1 have nonhomologous genes encoding host recognitionfunctions. J Virol, 10.1128/JVI.06666-11.

28. Oren A, et al. (2004) Biological effects of dilution of Dead Sea brine with seawater:Implications for the planning of the Red Sea - Dead Sea “Peace Conduit”. J Mar Syst46:121–131.

29. Vaisman N, Oren A (2009) Salisaeta longa gen. nov., sp. nov., a red, halophilic memberof the Bacteroidetes. Int J Syst Evol Microbiol 59:2571–2574.

30. Fischer D, Eisenberg D (1999) Finding families for genomic ORFans. Bioinformatics 15:759–762.

31. Comeau AM, et al. (2008) Exploring the prokaryotic virosphere. Res Microbiol 159:306–313.

32. Gorbalenya AE, Koonin EV (1989) Viral proteins containing the purine NTP-bindingsequence pattern. Nucleic Acids Res 17:8413–8440.

33. Strömsten NJ, Bamford DH, Bamford JKH (2005) In vitro DNA packaging of PRD1: Acommon mechanism for internal-membrane viruses. J Mol Biol 348:617–629.

34. Walker JE, Saraste M, Runswick MJ, Gay NJ (1982) Distantly related sequences in thealpha- and beta-subunits of ATP synthase, myosin, kinases and other ATP-requiringenzymes and a common nucleotide binding fold. EMBO J 1:945–951.

35. Klein R, et al. (2002) Natrialba magadii virus phiCh1: First complete nucleotide se-quence and functional organization of a virus infecting a haloalkaliphilic archaeon.Mol Microbiol 45:851–863.

36. Hendrix RW, Lawrence JG, Hatfull GF, Casjens S (2000) The origins and ongoingevolution of viruses. Trends Microbiol 8:504–508.

37. Juhala RJ, et al. (2000) Genomic sequences of bacteriophages HK97 and HK022:Pervasive genetic mosaicism in the lambdoid bacteriophages. J Mol Biol 299:27–51.

38. Porter K, et al. (2005) SH1: A novel, spherical halovirus isolated from an Australianhypersaline lake. Virology 335:22–33.

39. Baronio M, Lattanzio VM, Vaisman N, Oren A, Corcelli A (2010) The acylhalocapninesof halophilic bacteria: Structural details of unusual sulfonate sphingoids. J Lipid Res51:1878–1885.

40. Lattanzio VM, Baronio M, Oren A, Russell NJ, Corcelli A (2009) Characterization ofpolar membrane lipids of the extremely halophilic bacterium Salinibacter ruber andpossible role of cardiolipin. Biochim Biophys Acta 1791:25–31.

41. Laurinmäki PA, Huiskonen JT, Bamford DH, Butcher SJ (2005) Membrane proteinsmodulate the bilayer curvature in the bacterial virus Bam35. Structure 13:1819–1828.

42. Suttle CA (2007) Marine viruses—major players in the global ecosystem. Nat RevMicrobiol 5:801–812.

43. Bergh O, Børsheim KY, Bratbak G, Heldal M (1989) High abundance of viruses foundin aquatic environments. Nature 340:467–468.

44. Srinivasiah S, et al. (2008) Phages across the biosphere: Contrasts of viruses in soil andaquatic environments. Res Microbiol 159:349–357.

45. Nuttall SD, Dyall-Smith ML (1993) HF1 and HF2: Novel bacteriophages of halophilicarchaea. Virology 197:678–684.

46. Schägger H, von Jagow G (1987) Tricine-sodium dodecyl sulfate-polyacrylamide gelelectrophoresis for the separation of proteins in the range from 1 to 100 kDa. AnalBiochem 166:368–379.

47. Olkkonen VM, Bamford DH (1989) Quantitation of the adsorption and penetrationstages of bacteriophage ϕ 6 infection. Virology 171:229–238.

48. Drummond AJ, et al. (2011) Geneious v5.4.49. Ludtke SJ, Baldwin PR, Chiu W (1999) EMAN: Semiautomated software for high-res-

olution single-particle reconstructions. J Struct Biol 128:82–97.

7084 | www.pnas.org/cgi/doi/10.1073/pnas.1120174109 Aalto et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

6, 2

020