INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Herpesviruses are a large group of complex, double-stranded DNA viruses which infect vertebrates from teleost (bony) fish to humans. They exhibit narrow host specificites, most infecting only a single species in nature, and are thus considered likely to have evolved with their hosts. Comparisons of primary amino acid sequences predicted from complete genome sequences have shown that mammalian herpesviruses are genetically very divergent but nonetheless share a set of about 40 homologous genes, thus providing compelling evidence that these viruses evolved from a single ancestral herpesvirus (reviewed in reference 7). Moreover, genetic comparisons support the division of the family into three subfamilies, Alphaherpesvirinae, Betaherpesvirinae, and Gammaherpesvirinae, as proposed previously from biological criteria (15). The order of genes is largely conserved within each subfamily, whereas members of different subfamilies are more distantly related and exhibit several large-scale genomic rearrangements (4, 9). Viral phylogenies derived from rigorous sequence comparisons generally fit well with host phylogenies deduced from the fossil record, thus supporting the view that mammalian herpesviruses have cospeciated with their hosts, and this has allowed a time frame to be assigned (13, 14). Moreover, limited sequence data also indicate that avian herpesviruses fit readily into the subfamily Alphaherpesvirinae.

Nearly all research on herpesviruses has involved mammalian (and, to a lesser extent, avian) herpesviruses, and little is known about the many herpesviruses which infect cold-blooded vertebrates. The most extensively studied member of the latter group, channel catfish virus (CCV; ictalurid herpesvirus 1), was initially classified as a herpesvirus on the basis of its virion morphology and as a member of the Alphaherpesvirinae on the basis of its biological properties (15). Analysis of the complete genome sequence (6) indicated, however, that CCV has no specific relationship with mammalian herpesviruses at the level of primary amino acid sequence, in that no counterpart of a protein which is encoded only by mammalian herpesviruses, such as a structural protein, was detected in CCV. Thus CCV cannot be accommodated by the current taxonomy. The virus does encode several enzymes which are also specified by mammalian herpesviruses, such as DNA polymerase, dUTPase, and thymidine kinase. The genes encoding these proteins, however, are ubiquitous and could quite possibly have been acquired independently by the mammalian and fish herpesvirus lineages. Moreover, the CCV enzymes are no more closely related to their counterparts in other herpesviruses than to those in other organisms.

These findings may be interpreted in two ways. First, CCV and mammalian herpesviruses arose independently and have convergently acquired similar virion morphologies. Second, they evolved from an ancestral herpesvirus but have diverged so extensively over the 400 million years since their hosts separated that little sequence evidence remains. Several lines of evidence support the latter view, but it is fair to say that the case is not yet overwhelming. The best genetic indication for divergence rests in a single highly conserved protein which is encoded by two exons in the mammalian herpesviruses and three in CCV (open reading frames [ORFs] 62, 69, and 71). This protein apparently has a distant relative in bacteriophage T4 which functions as a subunit of the terminase involved in DNA packaging, but the fact that no cellular counterpart has yet been discovered highlights it as the best candidate for a gene which may have been inherited from a common ancestor rather than acquired via independent capture events. Moreover, despite the lack of conservation of the amino acid sequences of structural proteins, structural and functional congruences have been detected. Thus, the detailed three-dimensional structure of the CCV capsid is strikingly similar to that of herpes simplex virus type 1 (3). Also, local sequence features of the putative scaffold protein involved in CCV capsid formation suggest that it may be autoproteolytically processed via a pathway that is otherwise found only in mammalian herpesviruses (8).

Evidence for a herpesvirus lineage that lies outside the current taxonomic scheme has prompted investigations of its extent. Comparisons of CCV with salmonid herpesviruses appear useful in this respect, since the fossil record indicates that the three main subgroups of euteleosts (salmoniforms, neoteleosts, and ostariophysans, the latter including catfish) diverged around 130 million years ago (1). Salmonid fish are host to several herpesviruses, the principal of which are salmonid herpesviruses 1 and 2 (SalHV-1 and SalHV-2) (reviewed in reference 19). SalHV-1 was isolated on several occasions from a rainbow trout (Oncorhynchus mykiss) hatchery in the state of Washington in association with excessive mortality in young fish (20). The virus causes disease when injected into young rainbow trout maintained at 6 to 9°C but not in other salmonid species. SalHV-2 was isolated from Oncorhynchus masou, a landlocked Japanese form of Pacific salmon (11). It is serologically distinct from and has a wider host range than SalHV-1, causing virulent disease in the young of several Oncorhynchus species, including the rainbow trout. It also exhibits a higher temperature optimum for growth in cell culture than SalHV-1.

Partial sequence data for two genes have previously indicated that SalHV-2 is related to CCV (2). In this report, I describe the genome structure and gene arrangement of SalHV-1 and show that this virus is evolutionarily related to SalHV-2 and CCV. The data indicate that the processes which have resulted in the generation of certain genome structures and large-scale gene rearrangements during mammalian herpesvirus evolution have parallels in fish herpesvirus evolution. They also imply that fish herpesviruses occupy a distinct evolutionary space of an size equivalent to that occupied by mammalian herpesviruses and urge an accommodation in the herpesvirus taxonomy.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Growth of virus. A rainbow trout gonad cell line (RTG-2; ATCC CCL 55) was grown in minimal essential medium containing Earle's salts and Glutamax I (Life Technologies) supplemented with 10% fetal calf serum, nonessential amino acids, penicillin (100 U/ml), and streptomycin (100 µg/ml). Cells were subcultured at a ratio of 1:5 in trays, flasks, or roller bottles, becoming confluent in 2 to 3 weeks at 22 to 24°C.

Preparation of virion DNA. RTG-2 cells in batches of eight roller bottles were infected with SalHV-1 at a multiplicity of infection of approximately 0.01 PFU/cell. Cytopathic effect was complete after incubation at 10 to 12°C for 2 to 3 weeks. The infected cells were gently shaken into the medium and pelleted by centrifugation for 10 min at an RCF (at r_avg) of 2,560 (i.e., 5,000 rpm in a Sorvall GSA rotor). Cell-released virus was pelleted from the supernatant by centrifugation at an RCF (at r_avg) of 14,750 (i.e., 12,000 rpm in a Sorvall GSA rotor) for 2 h at 4°C and resuspended in a small volume of medium. Virions were purified by centrifugation on 5 to 15% (wt/vol) Ficoll gradients (18) and lysed by incubation for 3 h at 37°C in TE (10 mM Tris-HCl [pH 7.5], 1 mM EDTA) containing 0.5% (wt/vol) sodium dodecyl sulfate and 0.5 mg of protease per ml. The DNA was extracted twice with TE-equilibrated phenol and once with 24:1 (vol/vol) chloroform-isoamyl alcohol, ethanol precipitated, resuspended in TE, and stored at 20°C. A roller bottle typically yielded 1 to 2 µg of DNA.

Construction of cosmids and plasmids. A SalHV-1 cosmid library was constructed by using a modified form of Stratagene's Supercos 1 vector (5), employing methods recommended by Stratagene. SalHV-1 DNA was digested partially with BamHI and ligated into the vector which had been digested with XbaI, dephosphorylated, and digested with BamHI. The DNA was packaged into bacteriophage  particles and introduced into an appropriate Escherichia coli host, using Stratagene's Gigapack II system. Recombinant cosmid DNA was prepared from ampicillin-resistant colonies by alkaline lysis of 1-ml cultures, and aliquots of the DNA (usually about 5% of the yield) were analyzed by restriction endonuclease digestion and agarose gel electrophoresis.

Isotopic labeling of DNA. Oligonucleotides were labeled at their 5' ends by treatment with bacteriophage T4 polynucleotide kinase in the presence of [-³²P]ATP by using standard procedures and were purified in TE on small Sephadex G-50 columns. ³²P-labeled probes were obtained from double-stranded DNA by using the Nonaprimer system (Appligene) and denatured prior to hybridization by heating at 100°C for 2 min.

DNA hybridization. Restriction endonuclease fragments separated by agarose gel electrophoresis were denatured and passively transferred to a Hybond-N membrane (Amersham), using standard procedures. DNA was fixed to the membranes by irradiation in a UV Stratalinker 1800 (Stratagene). DNA hybridization was carried out for 3 to 16 h, using Rapid-hyb buffer (Amersham) in bottles rotating in a Hybaid Mark II hybridization oven. Standard hybridization temperatures were 65°C for double-stranded DNA probes and 42°C for oligonucleotide probes, but in some experiments temperatures were increased by up to 10°C for the former and 3°C for the latter. The membranes were then rinsed in washing buffer (20 mM Tris-HCl [pH 7.5], 0.3 M NaCl, 0.03 M trisodium citrate, 0.1% sodium dodecyl sulfate) heated to the temperature of hybridization, washed twice for 15 min in washing buffer at the temperature of hybridization, and finally rinsed in 10% washing buffer at room temperature. The membranes were air dried, reassembled, and autoradiographed.

DNA sequencing. The sequence of SalHV-1 BamHI P was determined by cloning random DNA fragments generated by sonication into M13mp19 and sequencing them by the dideoxynucleotide chain termination technology (10, 16). Autoradiographs were read with a Summagraphics Digitizer, and the database was compiled by using Staden's sequence analysis program (17), having first eliminated M13 clones containing pTZ19U sequences by reading short sections of each clone. The sequence was edited by reference to autoradiographs and analyzed with programs from the Wisconsin package, version 9.0 (Genetics Computer Group, Madison, Wis.). The 4,245-bp sequence was determined an average of 8.2 times per nucleotide.

Nucleotide sequence accession number. The sequence reported has been deposited with the GenBank data library under accession no. AF023673.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Size and structure of the SalHV-1 genome. Figure 1 shows restriction endonuclease profiles of SalHV-1 DNA for four enzymes. The enzyme of primary interest in this study, BamHI, produced 48 fragments ranging in size from 0.33 to 19.5 kbp (Table 1), as determined from this gel and others of higher agarose concentration (not shown). The total size of the genome calculated from the sum of fragment sizes was 174.4 kbp. The BamHI map for the SalHV-1 genome shown in Fig. 2 was obtained by analyzing the restriction profiles of cosmid clones and from associated experiments as described below. The map was derived from a large amount of data, of which it is possible to present only a small proportion here.

View larger version (54K): [in this window] [in a new window]   FIG. 1.   Restriction fragments of SalHV-1 DNA visualized by short-wavelength UV irradiation on a 0.5% agarose gel stained with ethidium bromide. Sizes are indicated in kilobase pairs.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Genome structure of SalHV-1. One of the two genome isomers is shown and is defined as the prototype. BamHI sites are marked, and fragment nomenclature is shown below in two ranks (A to Z and a to v). The inverted repeat (RS) is shown in a wider format than nonreiterated regions (UL and US). The locations of five cosmids are shown as rectangles below the fragment nomenclature, and BamHI sites are indicated. cos17 arose from a genome molecule in which US was inverted and thus appears to proceed from the right end of UL, through RS and into the right end of US, as indicated by the open ends of the rectangles.

View larger version (55K): [in this window] [in a new window]   FIG. 3.   Examples of hybridization data for an inverting region (US) flanked by an inverted repeat (RS) in the SalHV-1 genome. Restriction fragments of SalHV-1 DNA or cosmids were transferred from a 0.6% agarose gel and probed with radiolabeled BamHI K, U, or T at 75°C. Hybridizing BamHI fragments are indicated to the right of each panel, and EcoRI fragments are indicated to the left.

Relationship between SalHV-1 and other herpesviruses. To assess the relationship between SalHV-1 and other herpesviruses, 298 M13 clones derived randomly from the genome were sequenced. In total, they represented 34% of the genome. Each DNA sequence was conceptually translated in all six reading frames by using Pepdata, producing a concatenated amino acid sequence. The amino acid sequences were then compared with protein sequence databases by using Fasta. Databases used included collections of protein sequences from mammalian herpesviruses, from CCV, and from the Swissprot database.

View larger version (33K): [in this window] [in a new window]   FIG. 4.   Amino acid sequence alignments of random SalHV-1 DNA sequences conceptually translated in appropriate reading frames with the CCV ORF 37 protein (residues 534 to 615) (a) and the CCV ORF 54 protein (residues 306 to 399) (b). Identical residues are indicated in the "con" line.

View larger version (57K): [in this window] [in a new window]   FIG. 5.   Examples of oligonucleotides hybridizing to SalHV-1 DNA. BamHI fragments of SalHV-1 or cosmids were transferred from a 0.8% agarose gel and probed with radiolabeled SalHV-1 DNA at 65°C or oligonucleotides at 42°C. Fragments to which oligonucleotides hybridized are shown to the right of each panel, and SalHV-1 BamHI fragments are shown on the left.

View larger version (31K): [in this window] [in a new window]   FIG. 6.   Scale representation of relative gene order in SalHV-1 and CCV. The locations and orientations of the 18 genes in the 30- to 115-kbp region of the 134-kbp CCV genome are shown at the top; ORFs 62, 69, and 71 are exons of a single gene. BamHI fragments in the 20- to 135-kbp region of the SalHV-1 genome are shown at the bottom. Regions in the CCV genome corresponding to SalHV-1 oligonucleotide probes are connected by lines to the centers of the SalHV-1 fragments to which they hybridized. Five gene blocks (A to E) are shown as shaded rectangles below the CCV genome, and corresponding blocks are shown above the SalHV-1 genome (B' indicating an inversion).

View larger version (44K): [in this window] [in a new window]   FIG. 7.   Amino acid sequence alignments of conceptual translation products of random SalHV-1 DNA sequences with their counterparts in the SalHV-2 genome and the CCV gene 46 protein at residues 180 to 246 (a) and residues 759 to 837 (b) and the CCV ORF 62 protein at residues 345 to 401 (c). Residues conserved between SalHV-1 and SalHV-2 are indicated in the "con" line; those conserved between both viruses and CCV are indicated in the "CON" line.

BamHI P sequence. To confirm the presence of rearranged gene blocks, BamHI P was sequenced. Oligonucleotide hybridization experiments showed that this fragment contains portions of ORFs 49 (dUTPase), 57 (DNA polymerase), and 58. The sequence is 4,245 bp long, and the deduced gene arrangement is shown in Fig. 8a.

View larger version (43K): [in this window] [in a new window]   FIG. 8.   (a) Summary of gene order in SalHV-1 BamHI P displayed in the same orientation as the genome layout depicted in Fig. 2. Predicted protein coding regions are shaded, and the proposed intron linking ORFs 57 and 58 is shown as a white rectangle. An AATAAA element potentially involved in polyadenylation is shown as a vertical arrow. (b) Alignment of the putative amino acid sequences of SalHV-1 and CCV dUTPases. Conserved residues are shown in the "con" line. Five recognized dUTPase motifs (I to V) are indicated (12). The CCV protein is shown as commencing at the second ATG codon in the relevant reading frame. (c and d) Locations of potential splice sites linking ORFs 57 and 58 in SalHV-1 (c) and CCV (d). The first 600 bp of SalHV-1 BamHI P and the corresponding region in the CCV genome are shown, proceeding from the 3' end of ORF 57 through the 5' end of ORF 58. Potential exons and introns are marked. Translated sequences are bracketed by stop codons which define the 3' and 5' limits of ORFs 57 and 58, respectively; the two ORFs overlap in CCV. The first ATG codon in ORF 58 is doubly underlined in each sequence. Conserved amino acid residues are singly underlined. Nucleic acid and amino acid residues outside the putative exons are shown in lowercase. AATAAA elements that could signal polyadenylation of transcripts from ORF 57 are underlined. The sequence corresponding to the oligonucleotide used to locate the 3' end of SalHV-1 ORF 57 is underlined by dots.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The mapping data described in this report showed that the 174.4-kbp SalHV-1 genome consists of an inverting region (U_S; 25.6 kbp) flanked by an inverted repeat (R_S; 7.7 kbp) and a noninverting region (U_L; 133.4 kbp) which is not flanked by a detectable repeat. This arrangement is characteristic of mammalian alphaherpesviruses in the Varicellovirus genus, such as pseudorabies virus, equine herpesvirus 1, bovine herpesvirus 1, and varicella-zoster virus. Given the great divergence between SalHV-1 and mammalian herpesviruses implied by the absence of a specific genetic relationship, it appears that this genome structure has evolved independently in the mammalian and fish herpesvirus lineages. Similarly, the genome structure of CCV, which consists of a U_L region flanked by a substantial direct repeat, is the same as that of betaherpesviruses in the Roseolovirus genus (human herpesviruses 6 and 7) and one of the gammaherpesviruses ostensibly in the Rhadinovirus genus (equine herpesvirus 2) and again has apparently evolved independently.

Analysis of randomly sampled DNA sequence data indicated that SalHV-2 is the closest relative of SalHV-1. This conclusion is based on limited regions of two genes, and information on the overall level of sequence similarity and on gene order is not available. SalHV-1 is also clearly related to CCV in at least 18 genes at the level of amino acid sequence. Nevertheless, the distance between SalHV-1 and CCV is rather substantial, qualitatively of the same order as that between members of different mammalian herpesvirus subfamilies. Moreover, the gene orders in the two viruses are different, being related by rearrangement of at least five sequence blocks in U_L. Large-scale gene rearrangement in U_L is characteristic of the mammalian herpesvirus subfamilies and again appears to have occurred independently in fish herpesviruses.

Detailed sequence analysis confirmed that SalHV-1 BamHI P contains the junction between two gene blocks. It also resulted in the speculation that the DNA polymerases of SalHV-1 and CCV may be encoded by spliced mRNAs. This interpretation is not pivotal to this report, but subsequent confirmation from transcript mapping data would lead to the prediction that CCV DNA polymerase is 173 kDa in size, somewhat larger than mammalian herpesvirus DNA polymerases (approximately 140 kDa) which are not expressed via splicing. The role of the putative ORF 58 domain is a matter for more extreme speculation, given that recognizable CCV DNA polymerase motifs are all located in the ORF 57 domain.

Fish herpesviruses form a distinct and largely unrecognized part of the herpesvirus family which we now see as possibly similar in evolutionary breadth to herpesviruses of mammals and birds. The striking parallels observed in genome structure and gene rearrangement indicate that similar evolutionary mechanisms have operated independently in the two lineages. Thus, it is likely that it will be feasible to classify fish herpesviruses at the level of subfamily (e.g., as probably represented by SalHV-1 and CCV) and genus (e.g., as perhaps represented by SalHV-1 and SalHV-2), as has been undertaken for mammalian herpesviruses. Set in this context, it is apparent that the present position of the family name (Herpesviridae) is inappropriate by the criterion of genetic relatedness, since it encompasses only herpesviruses of mammals and birds. This name should be raised by one taxonomic level (to superfamily or order), and novel names should be developed to represent the mammalian/avian and fish herpesvirus lineages at what is currently the family level. It is possible that further accommodations of a more or less radical nature will eventually be necessary to fit herpesviruses of amphibians, reptiles, and invertebrates into this scheme as data become available.