Fowlpox Virus Encodes Nonessential Homologs of Cellular Alpha-SNAP, PC-1, and an Orphan Human Homolog of a Secreted Nematode Protein

Abstract

The genome of fowlpox virus (FWPV), type species of the Avipoxviridae, is considerably rearranged compared with that of vaccinia virus (the prototypic poxvirus and type species of the Orthopoxviridae) and is 30% larger. It is likely that the genome of FWPV contains genes in addition to those found in vaccinia virus, probably involved with its replication and survival in the chicken. A 7,470-bp segment of the FWPV genome has five open reading frames (ORFs), two of which encode ankyrin repeat proteins, many examples of which have been found in poxviruses. The remaining ORFs encode homologs of cellular genes not reported in any other virus. ORF-2 encodes a homolog of the yeast Sec17p and mammalian SNAP proteins, crucial to vesicular transport in the exocytic pathway. ORF-3 encodes a homolog of an orphan human protein, R31240_2, encoded on 19p13.2. ORF-3 is also homologous to three proteins (YLS2, YMV6, and C07B5.5) from the free-living nematode Caenorhabditis elegans and to a 43-kDa antigen from the parasitic nematode Trichinella spiralis. ORF-5 encodes a homolog of the mammalian plasma cell antigen PC-1, a type II glycoprotein with exophosphodiesterase activity. The ORFs are present in the virulent precursor, HP1, of the sequenced attenuated virus (FP9) and are conserved in other strains of FWPV. They were shown, by deletion mutagenesis, to be nonessential to virus replication in tissue culture. RNA encoding the viral homolog of PC-1 is expressed strongly early and late in infection, but RNAs encoding the homologs of SNAP and R31240_2 are expressed weakly and late.

Determination of the complete genome sequences of large DNA viruses (herpesviruses and poxviruses) has revealed the presence of numerous homologs of cellular genes (formally, the viral genes should be described as xenologs, as they are almost certainly related through lateral gene transfer [38], but we will use the more common generic term). Many of the cellular genes are involved in the immune and inflammatory responses, and subsequent study of the products of the viral homologs has demonstrated their role in immunomodulation (reviewed by Smith [56] and by Spriggs [57]). Study of immunomodulation has also revealed surprising functions for some viral genes (which had no known cellular homologs), especially in down-regulation of major histocompatibility complex (MHC) molecules (reviewed by Bonifacino [6]). Thus, various viruses express homologs of receptors for alpha and beta interferon (2, 60), interleukins (IL) and chemokines (1, 27, 28, 58), and tumor necrosis factor (54, 64). Poxviruses, however, express a β-chemokine receptor with no homology to known cellular receptors (55). Virus homologs of the soluble effectors themselves have also been found. Human herpesvirus 8 encodes homologs of IL-6 (42) and of the chemokines MIP-1α and MIP-1β (43). Homologs of IL-10 have been found in herpesviruses (31, 37, 48) and in a poxvirus, orf (24). The products of vaccinia virus genes E3L and K3L interfere with the interferon response (3, 10, 16). The crmA gene of cowpox virus blocks activation of IL-1β by inhibiting IL-1β-converting enzyme (45). Herpesviruses interfere with MHC-mediated presentation of antigens by several mechanisms. For instance, herpes simplex virus ICP47 blocks the TAP transporter (25, 30), while human cytomegalovirus proteins US2 and US11 “dislocate” MHC class I molecules from the endoplasmic reticulum back into the cytosol, where they are degraded (70, 71). The mechanisms of poxvirus down-regulation of MHC expression or presentation have, however, received relatively little attention.

Fowlpox virus (FWPV) is the prototypic member of the avipoxviruses. Able to replicate fully only in avian cells, it has been developed as an avian expression vector but has received considerable attention as a candidate nonreplicating virus vector for use in animals and humans (39, 44). Only about a quarter of its genome, which at 260 kbp (36) is more than a third larger than that of vaccinia virus (190 kbp), has been sequenced. Thus far, no clear candidates for viral immunomodulatory genes have been reported, though homologs of growth factors and serpins have been identified (63). This may in part be due to the dearth of available avian cytokine sequences; avian interferons, for instance, have only recently been cloned (22, 52). The molecular interactions of FWPV with its chicken host have not been well studied, partly because avian molecular immunology lags behind that of mammals in important areas.

Here we report the presence, in an 8.5-kbp block of sequence from the left side of the FWPV genome, of homologs of the cellular genes SNAP (involved in constitutive and regulated vesicle fusion) and PC-1 (a type II membrane glycoprotein with alkaline phosphodiesterase activity, implicated in non-insulin-dependent diabetes mellitus). These homologs have never been observed in any other virus. We show that they are conserved in different FWPV strains but that they are nonessential for virus replication in vitro. Their potential roles in virus-host interactions, including immunomodulation, are discussed. A neighboring gene is a homolog of three spliced genes of unknown function from the free-living nematode Caenorhabditis elegans, of a 43-kDa secreted antigen from the parasitic nematode Trichinella spiralis, and of a human gene on 19p13.2 encoding the orphan protein R31240_2. The virus homolog, FP-CEL1, is nonessential for virus replication in vitro.

MATERIALS AND METHODS

Viruses and cells.

Avipoxviruses were grown on chick embryo fibroblasts (CEFs) in the presence of 2% newborn calf serum. The pathogenic strain HP1 was supplied by A. Mayr and propagated by four passages on chorioallantoic membranes. Partially attenuated strain HP1-200 and attenuated strain HP1-438, derived from virulent HP1 by six passages on CEFs plus two passages on chorioallantoic membranes and 200 or 438 further passages through CEFs, respectively (35), were provided by A. Mayr. A twice-plaque-purified isolate of HP1-438 (FP9) was then passaged six times to constitute a stock. The vaccinal strain Poxine was provided by Duphar. The mild vaccine strain Websters FWPV M and the Chick-N-Pox vaccine strain were obtained from Salsbury Laboratories, Inc. (now Solvay Animal Health), Charles City, Iowa. Canarypox virus (strain 229), pigeonpox virus (strains Peekham and 950), sparrowpox virus (strain 9037), and turkeypox virus were provided by the Central Veterinary Laboratory, Weybridge, United Kingdom.

Genomic cloning and sequencing determinations.

Extraction and molecular cloning of FWPV FP9 DNA as well as manual sequencing of sonicated FWPV DNA in M13 with Sequenase II, ³⁵S-dATP, and dideoxynucleotide chain terminators were performed as described previously (5). The project was completed by manual sequencing of denatured plasmid DNA with specific primers. Sequence data was assembled on a DEC Vax mainframe (VMS) with the Staden sequencing package (59). Database searches were performed on the Vax with the FASTA and BLAST programs, and sequence analyses and comparisons were carried out with the GCG package (Sequence Analysis Software Package 7.2 [1991]; Genetics Computer Group, Madison, Wis.).

Analysis by PCR of gene conservation in avipoxviruses.

PCR was performed on virus DNA, isolated as previously described (8), with the following primer pairs: M29 (5′-GCAGATCTAACCATGGAACAAGAAGCGTATAG-3′) and M30 (5′-CGTCTAGATTCATTGTGTTTGTATATTCTTC-3′) for FP-SNAP; M123 (5′-ATGATATCACCTACGGC-3′) and M106 (5′-CTACATGTTTATAACACAACC-3′) for FP-CEL1; and M39 (5′-CCCCGCATGCATCATGACAGGAAGTAAAAC-3′) and M40 (5′-CCCGGATCCGTACATATACTACATGTAAG-3′) for FP-PC1 (Fig. 1).

FIG. 1.

Genomic arrangement of ORFs showing the positions of primers used to analyze conservation of the ORFs, the location of PCR probes used for Northern blotting, the FWPV genomic sequences used for deletion constructs, and the positions of flanking or internal primers used for screening of deletion mutants. ORFs (hatched or shaded boxes) are shown above the line representing the sequence, as all read from left to right in the deposited sequence. Early termination signals are indicated by asterisks. PCR fragments are indicated by thick lines, and the PCR primers generating them are indicated in boldface type above (for forward) or below (for reverse) the lines. The positions of deleted sequences are indicated by thin dashed lines.

Construction of deletion plasmids for transient dominant selection.

A cassette containing the vaccinia virus p7.5 early/late promoter upstream of the Escherichia coli gpt gene was cloned as a 1.1-kbp SalI/AatII fragment into the SmaI site of pNEB193 (New England Biolabs) in the same orientation as the amp gene to give pGNR. The FP-SNAP gene was amplified by PCR with M47 (5′-CTATTCACGTAGCTGTCG-3′) and M44 (5′-CCCCGCATGCACCAGTCTACTACTTCG-3′). The 2.1-kbp product was digested with PacI and SphI (within M44) and cloned into pGNR to generate pSL3. Deletions were introduced into the FP-SNAP gene by digestion with AflII and BsmI (pSL15) or with BsmI and SnaBI (pSL9) followed by blunt-ending and religation. The FP-CEL1 gene was subcloned from pM564 as a 4-kb BamHI/HindIII fragment into pGNR to make pSL2. Deletion of a 1-kbp SnaBI-to-ClaI fragment from pSL2, to make pSL17, resulted in deletion of most of the FP-CEL1 gene. The FP-PC1 gene was amplified with oligonucleotides M39 and M40. The 2.6-kbp product was cleaved with SphI (in M39) and BamHI (in M40) and then cloned into pGNR to give pPC1.FL. Deletions were made with XbaI-to-XbaI (pPC1.X), SpeI-to-SpeI (pPC1.S), and BstEII-to-XbaI (pPC1.B) fragments. A diagrammatic representation of the deletions is shown in Fig. 1.

Isolation of FWPV deletion mutants by transient dominant selection.

Deletion mutants were isolated by the transient dominant selection method (23), as described previously (8). Cells transfected with plasmids carrying deleted FWPV genes with the gpt gene were infected with FWPV, and then, following three rounds of plaque purification in the presence of mycophenolic acid, recombinant viruses carrying the gpt gene were isolated. The recombinants were subsequently plaque purified in the absence of mycophenolic acid until they lost the gpt gene, as demonstrated by the inability to replicate in the presence of mycophenolic acid.

Screening of FWPV deletion mutants.

Each deletion mutant was screened by PCR with flanking primers (giving PCR products of specific sizes for wild-type and deleted genes) or one flanking primer and one primer internal to the deletion (detecting only wild-type genes [Fig. 1]). The primers used are as follows: FP-SNAP flanking primers M166 (5′-TCTCATAACGAGTCCAG-3′) to M44 (pSL9) or M29 to M30 (pSL15); FP-SNAP internal primers M166 to M30 (pSL9) or M172 (5′-TTAAGAAACGTAAATAACGTTAAAG-3′) to M173 (5′-GGCATTCTATAGATTTTTTAGGATC-3′; pSL15); FP-CEL1 flanking primers M133 (5′-CTAATTCTAGTTGTTAGGG-3′) to M38 (5′-AGTTACTATTCCAGTAATGCG-3′); FP-CEL1 internal primers M133 to M44; FP-PC1 flanking primers M39 to M40; and FP-PC1 internal primers M168 (5′-TCTATAATATGATGTGTG-3′) to M40. A diagrammatic representation of the primers used to screen the deletion mutants is shown in Fig. 1.

Northern blot analysis.

Probes were generated for Northern blotting by amplification of the following fragments (Fig. 1) by PCR: ANK2, M47 to M330 (5′-CCCACCGTCGACTATCTTATACGGAAGAAATCTGG-3′); FP-SNAP, M29 to M30; FP-CEL1, M169 (5′-CGATAAACGTAAATGGAACATCG-3′) to M106 (5′-CTACATGTTTATAACACAACC-3′); ANK3, M25 (5′-CCCACCAAGCTTCCTTTTCTAATAGAGTTATTACG-3′) to M27 (5′-CCCGTAGTCGACCGTTTTACTTCCTGTCATG-3′); and FP-PC1, M41 (5′-GAAGAAAGAATAAATACCGTATTGAGGTGG-3′) to M40 or M39 to M171 (5′-GGTATA GTGTAAAATATATCCACC-3′). All of the probe DNA fragments were purified by agarose gel electrophoresis, and [³²P]dCTP probes were prepared with a random priming kit (Stratagene). CEFs were infected with wild-type or mutant virus at a multiplicity of infection (MOI) of 10. After 4 or 24 h of incubation with or without cytosine arabinoside (present at 40 μg/ml from 1 h preinfection), RNA was extracted with an RNeasy Mini kit (Qiagen) and was stored under alcohol at −70°C until it was used. Denaturing electrophoresis of RNA was carried out with a 1% formaldehyde–agarose gel. The gel was then blotted onto nitrocellulose and hybridized with the above probes (as described by Sambrook et al. [50]).

Growth curves.

Growth curve experiments were performed with wild-type and mutant viruses at MOI of 10 for single-step experiments or 0.001 for multistep experiments, as described previously (8).

Nucleotide sequence accession number.

The FWPV genomic sequence described herein has been given GenBank accession no. AJ006408.

RESULTS

Sequence analysis.

A random M13 clone of FWPV genomic DNA, MFP504, translated in six reading frames and screened against the SWISSPROT database, revealed good homology with murine PC-1 (MUSPC1B [65]). Unusually, this homology was also observed at the nucleotide level. The homology was with a region identified as an active site for the PC-1 nucleotide phosphodiesterase. A labelled prime-cut probe, generated from MFP504, was used to probe a HindIII FWPV genomic library in pAT153. Clone M564, containing a 6.5-kbp fragment, hybridized to the probe. Sequence determination of the clone revealed that it contained only the 5′ end of the gene encoding the PC-1 homolog. A second clone (K03) was isolated and cloned into pUC19 containing a 3.8-kbp SmaI/BglII genomic FWPV fragment which overlapped M564 from the SmaI site within the PC-1 coding sequence. This allowed completion of the sequence encoding the PC-1 homolog.

The sequence reported here is 7,470 bp in length, encompassing five large open reading frames (ORFs) which encode proteins with 406, 287, 375, 341, and 817 residues, with predicted molecular masses of 46.5, 33.4, 43, 39.3, and 94 kDa, respectively (Fig. 1). The first and fourth ORFs (ORF-1 [also called ANK2] and ORF-4/ANK3, respectively) encode proteins containing nine and five ankyrin repeats, respectively. The second ORF (ORF-2/FP-SNAP) encodes a protein with significant homology to yeast Sec17p and to mammalian SNAP proteins involved in vesicular transport. The third ORF (ORF-3/FP-CEL1) encodes a protein with homology to an orphan human gene on chromosome 19p13.2 encoding hypothetical protein R31240_2. ORF-3 is also homologous to three predicted multiply spliced proteins (YLS2, YMV6, and C07B5.5) from the free-living nematode C. elegans and to a secreted 43-kDa antigen from the parasitic nematode T. spiralis. The fifth ORF (ORF-5/FP-PC1) encodes the PC-1 homolog. The properties of ORF-1/ANK2 and ORF-4/ANK3 will be reported elsewhere.

FWPV homolog of cellular SNAP.

ORF-2/FP-SNAP has 40 to 44% amino acid identity over its whole length with mammalian and squid SNAPs and 26% identity with Sec17p from Saccharomyces cerevisiae (Fig. 2). Cellular SNAP (14) plays a crucial role in constitutive and regulated vesicle transport between several compartments within the cell (49). With NSF it appears to bind to the 7S complex formed between SNAP receptors on vesicle and target membranes (v- and t-SNAREs), forming in turn the 20S fusion particle. Three isoforms of SNAP (alpha, beta, and gamma) have been identified; alpha- and gamma-SNAP are found in a wide range of tissues, but beta-SNAP is specific to the brain (69).

FIG. 2.

Alignment of human (humalpha) (g1066084, U39412) and bovine (bovalpha) (g423230, S32367) alpha-SNAP, bovine beta-SNAP (bovbeta) (g423236, S32368), human gamma-SNAP (humgamma) (g1685288, U78107 [shown to residue 297]), and SNAPs from longfin squid (Loligo pealei) (g1078943, S52426), Drosophila melanogaster (drosomel) (g507754, U09374), S. cerevisiae (SaccCer) (Sec17p; g542367, S39837), and FWPV (Fpvsnap). Residues found in 60% or more of the sequences are boxed, and residues showing 85% or more homology to the upper sequence (according to Dayhoff’s PAM250 tables [19]) are shaded. The sequences were aligned with GCG PILEUP (using default parameters) and were displayed with SeqVu (J. Gardner, Garvan Institute of Medical Research, Sydney, Australia).

FWPV encodes a homolog of an orphan protein on human chromosome 19.

ORF-3/FP-CEL1 has 33% amino acid identity over 336 residues with the human hypothetical protein R31240_2, 32% amino acid identity over 232 residues with C. elegans YLS2 (F09G8.2 on chromosome III), 28% identity over 298 residues with YMV6 (K04H4.6 on chromosome III), and 31% identity over 321 residues with C07B5.5 on chromosome X (Fig. 3). It is of particular interest that the region of FP-CEL1 homology spans three to eight coding exons of the cellular homologs (FWPV, like other poxviruses, does not undergo splicing, so there are no introns in the virus homolog). ORF-3/FP-CEL1 has 18% amino acid identity, over 257 residues, with the 43-kDa antigen (EMBL accession no. M95499) from the parasitic nematode T. spiralis (67).

FIG. 3.

Alignment of the predicted human protein R31240_2 (g1905907, AD000092) and its homologs from C. elegans (yls2_caeel [g465851, P34387]; c07b5_5 [g559893, Q17778]; and ymv6_caeel [g465961, P34508 {shown from residue 163}]), T. spiralis (trichina [g345333, A44164]), and FWPV (fpvcel). The derivation and explanations of the alignment are as described in the legend to Fig. 2. Membrane-spanning segments predicted by the TopPred II program (13) are underlined. Signal cleavage sites predicted by the AnalyzeSignalase program (N. Mantei, ETH, Zurich, Switzerland) are indicated by triangles (the suboptimal site predicted for FP-CEL1 is shown by a triangle in parentheses). Well-conserved cysteines are indicated by circles.

The FWPV homolog of cellular PC-1 lacks the somatomedin B domains.

ORF-5/FP-PC1 is predicted to encode a type II membrane glycoprotein which has 38 to 39% amino acid identity, spanning 730 residues, with human and mouse PC-1, also type II membrane glycoproteins (Fig. 4). The FWPV protein shows no homology with these mammalian genes in the N-terminal, cytoplasmic, and transmembrane segments. Two somatomedin B domains, found in the membrane-proximal part of the extracellular segment of mammalian PC-1, are absent from the FWPV homolog as well as from the rice homolog (Fig. 4).

FIG. 4.

Alignment of the predicted FP-PC1 (fppc1) protein with human PC-1 (humpc1 [g129678]), rat B10/gp130/RB13_6 (ratrb13_6 [g1363274]), human autotaxin (humautotax [g1160616]), and rice phosphodiesterase (riceppd [g818849]). Note that human PC-1 is shown with an N-terminal cytoplasmic domain of 24 residues (9), as represented in the databases, rather than the 76 residues subsequently determined (4). The derivation and explanations of the alignment are as described in the legend to Fig. 2. The transmembrane sequences are underlined, and the two somatomedin B repeats in human PC-1, rat B10/gp130/RB13_6, and human autotaxin are indicated by dashed lines with arrowheads. Two predicted N-linked glycosylation sites which are either totally conserved (the site at human PC-1 residue 533, which is also conserved in C. elegans but not S. cerevisiae homologs) or partially conserved (the site at human PC-1 residue 289, which is conserved in mammalian PC-1, FP-PC1, and rice phosphodiesterase) are indicated by solid brackets. A dashed bracket indicates the location of the EF-hand motif. Cysteines conserved in human PC-1, rat B10/gp130/RB13_6, human autotaxin, and FP-PC1 are indicated by closed circles; those conserved in human PC-1, rat B10/gp130/RB13_6, and FP-PC1 are indicated by an open circle, while those conserved in sequences excluding FP-PC1 are indicated by closed triangles. SeqVu was also used to insert pad characters manually to permit local alignment of cysteines at human PC-1 residues 478, 574, 576, and 816.

PC-1 was first recognized on plasma cells (hence plasma cell antigen 1, or Pca-1) in mice as a disulfide-linked homodimer of a membrane glycoprotein (molecular weight, 120,000) with restricted tissue distribution (61). Full-length murine (65, 66) and human (9) cDNA clones were isolated and sequenced. Analysis of the sequence of PC-1 (53) showed that it contains the active site of alkaline phosphodiesterase I (EC 3.1.4.1), obtained by peptide sequencing (15). PC-1 was shown to possess both alkaline phosphodiesterase I (EC 3.1.4.1) and nucleotide pyrophosphatase (EC 3.6.1.9) activities in mice (46, 47) and humans (26), the pyrophosphatase activity being elevated in cultured skin fibroblasts from patients with Lowe’s syndrome.

There is now a growing family of proteins related to PC-1; these proteins have been identified in S. cerevisiae, including YCR026c on chromosome III (7), and in C. elegans, including C27A7.1 (g1805697/Z81041), C27A7.3 (g1805698/Z81041), and C01B10.5 on chromosome III and T03G6.3 on chromosome X (data not shown). Other members of the family include autotaxin, a human protein involved in tumor cell motility (33, 40); PD-Iα, a rat brain phosphodiesterase/pyrophosphatase (41); and RB13_6, a rat neural differentiation and tumor antigen (20).

Conservation of homologs of SNAP, R31240_2, and PC-1 in other FWPV strains and avipoxviruses.

To investigate the possibility that the genes we observed in FP9 were acquired recently during passage of the virus in embryo and tissue culture, PCR analysis was conducted with different passage levels of FP9 precursor (including the pathogenic progenitor HP1-6). The analysis was also extended to investigate the prevalence of the gene in other strains of FWPV (Poxine and Websters FWPV M and Chick-N-Pox) and in other avipoxviruses (pigeon-, canary-, turkey-, and sparrowpox viruses). FP-SNAP, FP-CEL1, and FP-PC1 were all found in HP1-6, HP1-200, Poxine, FWPV M, and Chick-N-Pox (Fig. 5). We were unable to detect these genes in other avipoxviruses, but this does not prove that the genes are absent; the sequences may have diverged to a point at which our primers were no longer functional.

FIG. 5.

FWPV homologs of SNAP, R31240_2, and PC-1 were present before the prolonged tissue culture passage history of plaque-purified isolate FP9. They are also found in other strains of FWPV but are not amplified from other avipoxviruses by PCR. Reactions were performed with the primers described in Materials and Methods; products were then analyzed by agarose gel electrophoresis alongside size marker DNAs and stained with ethidium bromide. Template DNAs were from FWPV virulent precursor HP1; from FWPV intermediate HP1-200; from FWPV attenuated plaque-purified FP9; from commercial FWPV vaccines Poxine (PX), Websters FWPV M (FPV-M), and Chick-N-Pox; from canarypox virus strains 89 (Can89) and 229 (Can229); from pigeonpox virus strains Peekham (PPV-P) and 950 (PPV950); from sparrowpox virus 9037 (SPV9037); and from turkeypox virus (TPV). Sizes of the products are shown.

Northern blot analysis shows that the FWPV homolog of PC-1 is strongly expressed as RNA.

Northern blot analysis with FP-SNAP sequences revealed no more than weak expression, either early or late (Fig. 6B). A weak early RNA of about 2.4 kb is similar in size to the strong transcript (Fig. 6A) produced by the upstream gene, ANK2, and may therefore represent readthrough from ANK2. The weak late transcripts of 3 to 7.5 kb are also in a size range similar to that observed for ANK2, but the distribution appears to be toward transcripts larger than observed for ANK2. Northern blot analysis of FP-CEL1 showed late expression of heterogeneous transcripts, between 2 and 4 kb, in the absence of cytosine arabinoside (araC). There appeared to be some weak araC-sensitive expression of these heterogeneous transcripts even at only 4 h postinfection (Fig. 6C). Northern blot analysis with FP-PC1 sequences (Fig. 6E) revealed the presence of an abundant 3-kb early mRNA in infected but not control cells (the ORF is 2.5 kb). The same probe showed strong expression, in the absence of cytosine arabinoside, of late transcripts of 2.8 to 5 kb in length. The size of the early FP-PC1 mRNA transcript was reduced slightly in virus carrying the 170-bp SpeI deletion and was reduced to 2 kb in virus carrying the 1,050-bp XbaI deletion and to 1 kb in virus carrying the 2-kb BstEII-to-XbaI deletion, thus indicating that the transcript does indeed span the FP-PC1 coding sequence (Fig. 6F).

FIG. 6.

Detection of FP-ANK2, FP-SNAP, FP-CEL1, FP-ANK3, and FP-PC1 transcripts in infected cells. The autoradiographs show the results of Northern blot hybridizations of whole-cell RNA samples from CEFs with random-primed probes for FP-ANK2 (A), FP-SNAP (B), FP-CEL1 (C), FP-ANK3 (D), and FP-PC1 (E and F). Samples were from uninfected cells (C) or from cells at 4 h (E, E+, X, S, and B) or 24 h (L and L+) postinfection with FWPV at an MOI of 10. Infected cells were incubated in the absence (E, L, X, S, and B) or presence (E+ and L+) of araC. Cells were infected with wild-type FWPV FP9 or the FP-PC1 XbaI-to-XbaI deletion mutant (panel F, lane X), the FP-PC1 SpeI-to-SpeI deletion mutant (panel F, lane S), or the FP-PC1 BstEII-to-XbaI deletion mutant (panel F, lane B). The arrows indicate the positions of 28S and 18S rRNAs. The sizes of RNA markers are indicated in kilobases.

Isolation of deletion mutants shows that FWPV homologs of SNAP, R31240_2, and PC-1 are nonessential for virus replication in vitro.

Putative FP-SNAP, FP-CEL1, and FP-PC1 deletion mutants obtained by the transdominant selection method were screened by PCR (Fig. 7) with flanking primers (for the presence of a deleted gene) and with one or both primers internal to the deletions (to exclude retention of a wild-type gene). Of six gpt-negative plaques obtained following recombination with pSL15, three carried only the AflII–BsmI-deleted FP-SNAP gene (Fig. 7A and B). Of five gpt-negative plaques screened after transdominant selection with pSL9, one carried only the BsmI–SnaBI-deleted SNAP gene (data not shown). Two of six gpt-negative plaques screened after transdominant selection with pSL17 carried only the SnaBI–ClaI-deleted FP-CEL1 gene (Fig. 7C and D). Following transdominant selection with deleted FP-PC1 genes, two of four gpt-negative plaques carried only the BstEII–XbaI-deleted FP-PC1 gene from pPC1.B (Fig. 7E and F), only one of eight carried only the XbaI-deleted FP-PC1 gene from pPC1.X (Fig. 7E and F), and one of two appeared to carry only the SpeI-deleted FP-PC1 gene from pPC1.S (Fig. 7E). These results, showing isolation of deletion mutants for the FWPV SNAP, CEL1, and PC-1 homologs, indicate that all three viral homologs are nonessential for replication in tissue culture.

FIG. 7.

PCR analysis to screen putative FWPV FP9 mutants for deletions in FP-SNAP (A and B), FP-CEL1 (C and D), and FP-PC1 (E and F). Flanking primers were used to show the presence of deleted genes (B, D, and E), and deletion-internal primers were used to exclude the retention of wild-type genes (A, C, and F). DNA samples were from wild-type FP9 (wt), the appropriate deletion plasmid (Δ), and gpt-negative putative mutants (1 and 2 for FP-SNAP and FP-CEL1; B1 and B2, X1 and X2, and S1 and S2 for BstEII-to-XbaI, XbaI-to-XbaI, and SpeI-to-SpeI deletions of FP-PC1, respectively). In all cases, 1 is wild type and 2 is mutant. Controls were used with no template DNA (−ve) or with a plasmid, pPC1.FL, containing full-length FP-PC1 (FL). The FP-SNAP mutants shown were produced by using pSL15 (AflIII to BsmI), and the FP-CEL1 mutants shown carry the SnaBI-to-ClaI mutation of pSL17. The sizes of the PCR products are shown in base pairs.

Growth curves.

No differences in the replication kinetics of FP-SNAP, FP-CEL1, or FP-PC1 mutants were observed with single (MOI, 10)- or multiple (MOI, 0.001)-step infections in CEFs (data not shown), suggesting that these genes do not play significant roles in replication of the virus in tissue culture. Given the close proximity of adjacent genes in poxviruses, care has to be taken to avoid disrupting the regulatory sequences of neighboring genes. Currently, with no observable phenotype for any of the deletions, this is not a problem; it would require consideration, however, if any phenotype should be detected, for instance, in vivo. With N- and C-terminal deletions for FP-SNAP, it should be possible to determine whether any phenotype is due to deletion of FP-SNAP or to effects on the expression of ANK2 or FP-CEL1, respectively. It is unlikely that the deletion in FP-CEL1 would significantly affect transcription of FP-SNAP, the coding sequence of which ends 260 bp upstream. It is also unlikely to affect expression of ANK3, the ORF for which starts 280 bp downstream. The SpeI deletion in FP-PC1 is a small internal deletion (leading to a frameshift and immediate truncation), which is very unlikely to affect transcription of neighboring genes. The BstEII-XbaI deletion in FP-PC1 is unlikely to affect expression of the upstream gene (ANK3), as the XbaI site is 450 bp downstream of the end of the ANK3 ORF. The BstEII-XbaI and XbaI deletions in FP-PC1 are unlikely to affect expression of the downstream gene, as the XbaI site is 150 bp before the end of the FP-PC1 ORF.

DISCUSSION

The sequence described here is located between 33 and 42 kbp from the left end of the FWPV FP9 genome, flanking the junction between BamHI fragments B and C (36). Mapping data (not shown) is consistent with the 3′ end of ORF-5/FP-PC1 being proximal to the left terminus of the genome and ORF-1/ANK2 being distal. All of the genes in this segment would therefore be transcribed toward the left terminus. This would be consistent with other poxviruses in which genes near the termini are transcribed toward the nearest terminus. None of the sequence described here has a counterpart in any of the other poxviruses for which sequence data has been reported, nor are there any reports of homologs in any other viruses of any of the proteins encoded by the sequence reported here (although many genes encoding ankyrin repeat proteins have been identified in poxviruses).

An AT-rich core consensus sequence has been derived for vaccinia virus early promoters (17). Although no such consensus has been derived for FWPV, vaccinia virus promoters function well in FWPV so it is assumed that similar constraints are applied to FWPV promoters. Similarly, vaccinia virus late promoters contain the motif TAAAT(A/T/G) at the transcriptional initiation site (18). The sequence T₅NT serves as a terminator for early transcription in vaccinia virus (72) and apparently in other poxviruses. No late transcriptional terminator which correlates with the heterogeneous length of late transcripts has been identified. Both the FP-SNAP and FP-CEL1 genes have early terminators within their coding sequences; FP-PC1 does not. Thus, it would be predicted that FP-SNAP and FP-CEL1 are transcribed late, not early, and that FP-PC1 is transcribed early and/or late. Candidate late promoter sequences exist upstream of each of these genes. There are no candidate early promoter sequences in suitable positions upstream of FP-SNAP and FP-CEL1, even allowing for five mismatches from the 16-bp consensus vaccinia virus sequence (AAAAAATGAAAAAAAA). Upstream of FP-PC1, however, there are three candidate early promoter sequences with start sites 13 to 20 bases upstream of the first predicted AUG codon. The core of each of these candidate promoters has five mismatches compared to the vaccinia virus consensus. Taken together, these data indicate that FP-PC1 would be transcribed early (and possibly late) as an mRNA of at least 2.5 kb but that FP-SNAP and FP-CEL1 would be transcribed late. The results of Northern blotting of infected cell RNA, therefore, correspond with this analysis for FP-PC1 (an abundant 3-kb early mRNA and late transcripts of 2.8 to 5 kb), for FP-SNAP (no early transcripts and only weak late transcripts), and for FP-CEL1 (only late transcripts).

Conservation of the genes in different FWPV strains suggests that FP-SNAP, FP-CEL1, and FP-PC1 play important roles in virus survival. That their deletion does not seem to affect replication in vitro indicates that their roles are in virus-host interactions. It is unlikely that the loss of any one such gene has a great effect on virus replication in vivo, so it will be important to know more about possible activities and targets.

Several roles could be postulated for the viral SNAP homolog. It may be required merely to supplement existing cellular supplies of SNAP to allow efficient replication of FWPV. Similarly, it may be required to replace cellular SNAP depleted as a consequence of the shutdown of host cell metabolism upon virus infection. It may, however, be required to interfere with the normal vesicle transport pathways, subverting them to permit virus replication or, alternatively, to inhibit vesicle transport, possibly as a way of preventing antigen presentation by MHC. Clearly the isolation of SNAP deletion mutants indicates that any replacement or subversion function for SNAP in FWPV replication may not be absolute. Moreover, deletion of FP-SNAP did not result in any obvious changes to the profiles of proteins secreted from the infected cell (data not shown).

Potential roles for the viral homolog of PC-1 are less evident than for the SNAP homolog, mainly because the cellular role(s) for PC-1 is less well defined. PC-1 has been implicated in the inhibition of insulin receptor tyrosine kinase in insulin-resistant patients with non-insulin-dependent diabetes mellitus (34). It has also been shown that PC-1 is coordinately up-regulated in activated T cells with two other ectoenzymes involved in nucleotide metabolism, CD38 NAD glycohydrolase and CD73 5′-nucleotidase (21). Autotaxin is able to hydrolyze ATP to ADP or AMP and AMP to adenosine (12). It is, therefore, able to perform the functions of CD39 apyrase, an ATP diphosphohydrolase (E.C. 3.6.1.5 [68]), and CD73 ecto-5′-nucleotidase (E.C. 3.1.3.5 [62]), converting ATP to adenosine, which can be taken up by cells (11). Acting alone or in concert, these enzymes may provide cells with external supplementary sources of nucleosides. The presence in the FWPV genome of thymidine and thymidylate kinases indicates that the intracellular side of the scavenger pathway can be beneficial. That it may be more important for FWPV than for vaccinia virus is perhaps indicated by the observation that insertion into the thymidine kinase locus of attenuated FWPV is difficult to obtain (51). We have, however, been unable to detect any difference between the replication rates of mutant and wild-type viruses in media which do or do not contain ATP (medium 199 or modified Eagle medium, respectively). Other potential roles for the PC-1 homolog must be speculative but may include down-regulation of signal transduction (cf. PC-1 in non-insulin-dependent diabetes mellitus [34]) or interference in interactions of the infected cell with the extracellular matrix or with other cells (via extracellular phosphodiesterase activity). Of some relevance to this is the observation that disruption of the exophosphodiesterase activity of autotaxin, by mutation of the catalytic threonine (residue 210) to alanine or aspartate, inhibits motility stimulation. This has been held to “open up the possibility of extra-cellular enzymatic cascades as a regulatory mechanism” (32).

Potential roles for FP-CEL1, the viral homolog of the unknown C. elegans proteins and of the orphan human protein R31240_2, are even more mysterious. There is, as yet, no indication of potential roles, functions, or activities for the human homolog of ORF-3, R31240_2, or for any of the C. elegans homologs, but the protein from T. spiralis is known to be a secreted antigen. It is somewhat surprising that there is no such clear homolog encoded in the completely sequenced genome of S. cerevisiae, suggesting that it might have developed in metazoans. Given that the T. spiralis protein is secreted and given the indications that the homologs in C. elegans and humans are also secreted (Fig. 3), it will be interesting to see if FP-CEL1 is secreted from infected cells and, if so, whether a ligand or activity can be identified.

The origins of such clear homologs of cellular genes in FWPV are unknown. As the splicing pattern of the animal homologs of FP-CEL1 is well conserved from humans down to nematodes, it seems most likely that the viral gene (which has no introns) was acquired by insertion of cDNA that had been copied from mRNA by reverse transcriptase from endogenous or exogenous retroviruses. It is tempting to speculate that the ability of poxvirus genomes to carry integrated retroviral proviruses, which can then produce infectious retrovirus (as shown recently for reticuloendotheliosis virus in FWPV [29]), could facilitate this process. Whatever the exact mechanism, the relationship of the viral genes to the cellular genes would be one of xenology. Until viruses more closely related to FWPV than vaccinia virus, variola virus, and molluscum contagiosum virus are sequenced, it will not be clear whether these homologs were acquired by FWPV or by a progenitor. More extensive sequence analysis will also be required to establish whether these homologs were acquired within the avipoxvirus lineage after divergence from the mammalian poxviruses or whether preexisting genes were lost by the mammalian viruses.

The discovery of homologs of cellular SNAP and PC-1 genes, not found previously in any other virus, in such a small part of the genome of FWPV is intriguing. It remains to be seen whether such homologs play a role in virus-host relationships unique to the avian host. It also remains to be seen whether the rest of the genome of FWPV and other avipoxviruses will yield other such surprising homologies.

ADDENDUM IN PROOF

Two independent sequences entered into the genome data bases since submission of this article show >95% amino acid identity to the human homolog of FP-Cell, R31240_2. Both of these sequences (G2895897 and G2921839) have been identified as DNase II (E.C. 3.1.22.1). There has also been an independent report of the cloning of human DNase II (T. Yasuda, H. Takeshita, R. Iida, T. Nakajima, O. Hosomi, Y. Nakashima, and K. Kishi, J. Biol. Chem. 273:2610–2616, 1998).