Complete Genome Sequence of Avian Paramyxovirus (APMV) Serotype 5 Completes the Analysis of Nine APMV Serotypes and Reveals the Longest APMV Genome

Background Avian paramyxoviruses (APMV) consist of nine known serotypes. The genomes of representatives of all APMV serotypes except APMV type 5 have recently been fully sequenced. Here, we report the complete genome sequence of the APMV-5 prototype strain budgerigar/Kunitachi/74. Methodology/Principal Findings APMV-5 Kunitachi virus is unusual in that it lacks a virion hemagglutinin and does not grow in the allantoic cavity of embryonated chicken eggs. However, the virus grew in the amniotic cavity of embryonated chicken eggs and in twelve different established cell lines and two primary cell cultures. The genome is 17,262 nucleotides (nt) long, which is the longest among members of genus Avulavirus, and encodes six non-overlapping genes in the order of 3′N-P/V/W-M-F-HN-L-5′ with intergenic regions of 4–57 nt. The genome length follows the ‘rule of six’ and contains a 55-nt leader sequence at the 3′end and a 552 nt trailer sequence at the 5′ end. The phosphoprotein (P) gene contains a conserved RNA editing site and is predicted to encode P, V, and W proteins. The cleavage site of the F protein (G-K-R-K-K-R↓F) conforms to the cleavage site motif of the ubiquitous cellular protease furin. Consistent with this, exogenous protease was not required for virus replication in vitro. However, the intracerebral pathogenicity index of APMV-5 strain Kunitachi in one-day-old chicks was found to be zero, indicating that the virus is avirulent for chickens despite the presence of a polybasic F cleavage site. Conclusions/Significance Phylogenetic analysis of the sequences of the APVM-5 genome and proteins versus those of the other APMV serotypes showed that APMV-5 is more closely related to APMV-6 than to the other APMVs. Furthermore, these comparisons provided evidence of extensive genome-wide divergence that supports the classification of the APMVs into nine separate serotypes. The structure of the F cleavage site does not appear to be a reliable indicator of virulence among APMV serotypes 2–9. The availability of sequence information for all known APMV serotypes will facilitate studies in epidemiology and vaccinology.


Introduction
Paramyxoviruses constitute a large family that includes important respiratory and systemic pathogens of humans and animals [1]. Some members of the family are responsible for major diseases, others cause inapparent infections, and the pathogenic potential of others remains unknown. Family Paramyxoviridae is divided into two subfamilies: Paramyxovirinae and Pneumovirinae.  [2].
Members of the family Paramyxoviridae are pleomorphic, enveloped viruses containing a negative-sense, single-stranded RNA genome. The viral genomes vary in length between 13 to 19 kb and contain 6-10 tandemly linked genes that encode at least 7 and as many as 12 different proteins [1]. For members of subfamily Paramyxovirinae, efficient RNA replication requires that the nucleotide (nt) length of the genome is an even multiple of six, known as 'rule of six', reflecting the precise packaging of the polynucleotide in the nucleocapsid [3,4]. The 39 and 59 ends of the genome contain short extragenic sequences known as the 'leader' and 'trailer', regions, respectively. The viral RNA polymerase begins transcription at the 39 end and proceeds downstream in a sequential manner generating individual mRNAs by a start-stop mechanism guided by gene-start (GS) and gene-end (GE) signals that flank each gene. Non-coding intergenic sequences (IGS) are present between gene boundaries and are not copied into mRNAs. During RNA replication, the GS and GE signals are ignored and a complementary copy of the genome (the antigenome) is synthesized, which serves as the template for synthesis of progeny genome.
All members of family Paramyxoviridae encode a nucleoprotein (N), a phosphoprotein (P), a matrix protein (M), a fusion glycoprotein (F), an attachment glycoprotein that in the case of the APMVs is a hemagglutinin-neuraminidase (HN), and a large polymerase protein (L) [1]. The N protein binds the entire length of the viral genomic and antigenomic RNAs to form a stable nucleocapsid that is the template for transcription and RNA replication. The P protein associates with monomeric N protein destined for nucleocapsid assembly and serves as a polymerase cofactor essential for RNA synthesis [5]. The M protein lines the inner membrane of the viral envelope and plays a major role in virion morphogenesis [6,7]. The F protein mediates viral penetration and syncytium formation. F is synthesized as an inactive precursor F0 that is activated by cleavage by host protease into F1 and F2 subunits. HN initiates infection by binding to sialic acid-containing host-cell surface receptors. HN protein also possesses neuraminidase activity that prevents self-aggregation by progeny viruses. The large protein (L) is the viral RNA dependent RNA polymerase. In addition, most members of subfamily Paramyxovirinae encode two additional accessory proteins, V and W (or I, in case of genus Rubulavirus) from the P gene by RNA editing during transcription. RNA editing involves the cotranscriptional insertion of one or more non-templated G residues into the nascent P mRNA by the viral polymerase at a conserved RNA editing motif located midway along the P gene. This results in translational frameshifts that access alternative open reading frames downstream, generating the V and W/I proteins. These are proteins whose N-terminal domain is identical to that of P, but each of which have a unique C-terminal domain encoded by the alternative reading frame. The V protein has a conserved cysteine rich zinc finger binding motif in its unique C-terminal domain that is necessary for the activity of V in regulating RNA synthesis and antagonizing the host interferon system [8,9].
Most of the paramyxoviruses that are isolated from avian hosts are classified in genus Avulavirus of subfamily Paramyxovirinae: the only exceptions are the avian metapneumoviruses, which are classified in genus Metapneumovirus of subfamily Pneumovirinae due to substantial differences in genome organization and sequence relatedness [1]. Genus Avulavirus is composed of nine recognized serotypes (APMV-1 through APMV-9), a classification that is based on Hemagglutination Inhibition (HI) and Neuraminidase Inhibition (NI) assays [10]. APMV-1 is composed of the many naturally-occurring strains of Newcastle disease virus (NDV) and is the only well-characterized serotype, owing to the high morbidity, mortality, and economic loss caused by highly virulent strains. APMV-2 and -3 also have been reported to cause significant disease in poultry [11,12], whereas the pathogenic potential of APMV-4 to -9 is generally unknown [13]. Recently, APMV-3 infections in Neophema species of birds was shown to cause 70% mortality [14]. Infections from APMV-4, -8 and -9 appear to be restricted to ducks and geese. APMV-6 and -7 infections in turkeys cause drops in egg production and induce respiratory disease. Currently, our knowledge on the APMV serotypes is expanding with the availability of complete genome sequences of prototypes of all APMV serotypes except APMV-5 [15,16,17,18,19,20,21,22].
APMV-5 has not been well-characterized but appears to differ from all the other APMV serotypes in two major attributes: (1) the inability to grow in allantoic cavity of embryonated chicken eggs, and, (2) the failure to cause hemagglutination with chicken RBC [23]. APMV-5 was first isolated from an epizootic outbreak involving budgerigars (Melopsittacus undulatus) at Kunitachi, Tokyo, Japan in 1974 [23]. This virus was serologically and antigenically distinct from the other previously known APMV serotypes and serves as the prototype virus for this serotype. APVM-5 causes a disease in budgerigars that is characterized by depression, dyspnoea, diarrhea, torticollis and very high mortality. A second outbreak occurred among budgerigars in Brisbane in Queensland, Australia. Experimental infection of APMV-5 isolated from this outbreak caused no signs of disease in young and adult chickens and pigeons, but caused acute fatal enteritis among immature budgerigars [24]. The third known outbreak of APMV-5 disease occurred in budgerigars from UK in 1993 [25]. The clinical signs consisted mainly of vomiting and diarrhea followed by death. Of all the APMVs examined to date, only APMV-1 and APMV-5 have been associated with 100% mortality [23].
As an initial step towards understanding the molecular biology and biological characteristics of APMV-5, we have determined the complete genome sequence of APMV-5 prototype strain APMV-5/ budgerigar/Kunitachi/74 (GeneBank accession no. GU206351). An understanding of the molecular and biological characteristics of APMV-5 is important for characterizing the sequence and antigenic relationships within and among the APMV serotypes and for developing vaccines and diagnostic reagents against these viruses.

Virus and Cells
APMV-5 strain budgerigar/Kunitachi/74 was kindly provided by Dr. Ian Brown, the Veterinary Laboratories Agency, Weybridge, Surrey, UK. The virus was propagated in the African green monkey kidney (Vero) cell line. Vero cells were grown in Dulbecco's minimum essential medium (DMEM) containing 10% fetal bovine serum (FBS) and incubated at 37uC under 5% CO 2 . Cell monolayers were infected with a 10 3 dilution of the original virus stock; after 2h of adsorption, the viral inoculum was replaced with maintenance medium containing 2% FBS. The infected cells were observed daily for cytopathic effects (CPE), which was evident beginning on day 3. The infected cell culture supernatant was examined daily for hemagglutination (HA) activity using chicken RBCs, but this failed to detect virus growth. On day 5 post-infection, when CPE was extensive, the infected cells were scraped into the cell culture medium and pelleted, and the cell pellet was resuspended in a small volume of collected medium and subjected to three cycles of freezing and thawing. This suspension was clarified by low speed centrifugation and the collected medium was pooled and stored at 280uC as virus stock. The virus titer was determined by plaque assay in Vero cells using 0.8% methylcellulose overlay and staining with 1% crystal violet 5 days post-infection. Growth of the virus was evaluated in twelve established cell lines namely, Vero, chicken embryo fibroblast (DF-1), Madin-Darby Canine Kidney (MDCK), human epidermoid carcinoma (HEp-2), Baby Hamster Kidney (BHK-21), Bovine Turbinate (BT), Pig Kidney (PK-15), Quail fibrosarcoma (QT-35), Rabbit Kidney (RK-13), Human cervical carcinoma (HeLa), Madin-Darby Bovine Kidney (MDBK), and duck embryo (CCL-141), all the cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA). Growth was also evaluated in two primary cell cultures, namely chicken embryo fibroblast (CEF) and chicken embryo kidney (CEK) cells, which were cultured from 10 and 20-day-old specific pathogen free (SPF) embryonated chicken eggs, respectively, using standard procedures. A total of three serial passages of the virus were made in each cell line to examine virus growth. At each passage level, the cells were collected on day 5-x post-infection, when CPE was apparent and freeze-thawed as described above, and the one-tenth of the collected medium was used to inoculate the next passage. The virus failed to grow in the allantoic cavity of 9-day-old embryonated chicken eggs, but was successfully propagated in the amniotic cavity of 8-day-old embryonated chicken eggs. The infected amniotic fluid was harvested 3 days post-inoculation and the virus was titered using the plaque assay described above. The growth of virus in each cell type was observed with and without the presence of 10% allantoic fluid, or 1-5 mg/ml of acetyl trypsin (Invitrogen, USA), or 1-5 mg/ml of a-chymotrypsin (Sigma), as potential sources of protease for cleavage of the F protein if necessary. All research involving animals have been approved by the members of the institutional animal care and use committee (IACUC).

Virus RNA Isolation and Sequence Analysis
Viral RNA was isolated from infected Vero cells using RNeasy kit according to the manufacturer's instructions (QIAGEN, USA). Reverse transcription (RT) was performed using Superscript II RT (Invitrogen) and an oligo dT primer. The resulting first strand cDNA was PCR-amplified (Recombinant Taq DNA polymerase, Invitrogen) using degenerate consensus primers as the forward primers and either oligo dT or a degenerate consensus primer as the reverse primers. The degenerate consensus forward and reverse primers were designed using published sequences of the viral genomes of members of genera Avulavirus, Rubulavirus, Respirovirus and Morbillivirus (accession numbers given below). Specifically, the N, P, M, F, HN and L nucleotide sequences of selected viruses were aligned, and highly-conserved sequence segments were identified and used to design minimally-degenerate primers of 22-24 nt in length. The following degenerate consensus forward primers genes (N = A/C/G/T, S = G/C, M = A/C, W = A/T, K = G/T, R = A/G) were used in conjunction with oligo dT: (i) the N gene forward primer AP-5N380F (59-NTKCGTCWCTTGCTKCACGAARCA-39), which would prime at APMV-5 nt 382-405, within the N gene, yielding a 1350 bp length amplicon; (ii) the P gene forward primer AP-5PF 2315 (59-KCACWCTGCNACASATCAGMRTCC-39), which would prime at APMV-5 nt 2315-2338, within the P gene, yielding a 1305 bp length amplicon; (iii) the M gene forward primer AP-5MF 4465 (59-TGGWGTCTACAAARCTCA-TATCTG-39), which would prime at APMV-5 nt 4465-4488, within the M gene, yielding a 952 bp length amplicon; (iv) the F gene forward primer AP-5FF 5981 (59-TTGGTTTGGCAW-CTRCTRCACARR-39), which would prime at APMV-5 nt 5961-5984, within the F gene, yielding a 1505 bp length amplicon; (v) the HN gene forward primer AP-5HNF 7888 (59-TTAGATARRRTTWCTGTTRAGGTA-39), which would prime at APMV-5 nt 7888-7911, within the HN gene, yielding a 1824 bp length amplicon; and (vi) the L gene forward primer AP-5LF 14337 (59-CTRTTTACRTCAGCWGCKCRAGAC-39), which would prime at APMV-5 nt 14337-14360, within the L gene, yielding a 2368 bp length amplicon. In addition, a second internal L gene forward primer AP-5L1F (59-CCCTWTKKT-CTCTCATWGATACTA-39), which would prime at APMV-5 nt 10618-10642, within the L gene, was used in conjunction with reverse primer AP-5L1R (59-WCTTKRTCCTCKCAWTGC-TGWTCT-39), which would prime at APMV-5 nt 13595-13619, within the L gene, yielding a 3002 bp length amplicon. PCR was carried out using a cycling sequence of 95uC for 5 min followed by 35 cycles of 95uC for 1 min, 45uC for 1 min and 72uC for 2 min, which was then followed by a final extension of 72uC for 10 min. New sets of primers were designed from the sequence analysis of these virus-specific cDNAs for determining the remaining sequence. PCR products for determining the sequences of the 39 and 59 termini of the genome were made by rapid amplification of cDNA ends (39 and 59 RACE, respectively) from the viral RNA using manufacturer's protocol (Invitrogen) with modification from Subbiah et al., 2008. Briefly purified viral RNA from infected cells was ligated using T4 RNA ligase to a 59-phosphorylated and 39 blocked RNA oligonucleotide adapter (59-CCAAAACGC-CAUUUCCACCUUCUCUUC-39). The ligated RNA was subjected to RT using adapter 2 (59-GAAGAGAAGGTGGAAA-TGGCGTTTTGG-39) as the primer. The cDNA was PCR amplified using adapter 2 as the forward primer and a gene specific reverse primer obtained from the N gene (59-GTT-TAATCTTAAGAGCCCTCCTGTC-39), which would prime at APMV-5 nt 200-224, within the N gene. For determining the 59 end of the genome, a 59RACE kit (Invitrogen) was utilized. Purified viral RNA was reverse transcribed using an L gene specific primer (59-CTAACGCGCCTTCTTACATCCGACGT-TCAT-39), which would prime at APMV-5 nt 16423-16452, within the L gene. The resulting cDNA was subjected to a tailing reaction with dCTP using terminal deoxynucleotidyl transferase, and then was PCR amplified using same L gene specific primer and an anchored oligo dG primer. PCR products were gel purified and were sequenced directly or were cloned into the TOPO TA cloning vector (Invitrogen) and positive clones were sequenced using M13 forward and M13 reverse primers. DNA sequencing was performed in a 3130xl genetic analyzer (Applied Biosystems Inc, USA) according to the manufacturer's instruction. The entire genome sequence was determined at least three times: once by analysis of cloned cDNAs and twice by direct sequencing of uncloned RT-PCR products, providing a consensus sequence.

Sequence and Phylogenetic Tree Analysis
Sequence analysis, BLAST searches, and prediction of ORFs were carried out using the SeqMan and EditSeq programs, and PCR primers were designed using the PrimerSelect program in DNASTAR Lasergene 8 (software suite for sequence analysis, version 8.0.2(13) 412). Bootstrap values in the phylogenetic trees were calculated using 1000 replicas. Construction of phylogenetic trees and divergence analysis was performed by maximum parsimony method using MEGA 4 software (Molecular Evolutionary Genetics Analysis) [26].

Growth Characteristics of APMV-5
APMV-5 strain Kunitachi produced CPE in Vero cells, resulting in cell rounding and detachment as well as the formation of large syncytia observed 3 days post-infection. The virus produced distinct plaques in Vero cells under methylcellulose overlay. Supplementation with 10% fresh allantoic fluid, acetyl trypsin (1-5 mg/ml), or a-chymotrypsin (1-5 mg/ml) did not enhance the growth of the virus, indicating the lack of a requirement for external protease for efficient cleavage of the F protein. Consistent with previous reports [16,18,20], APMV-5 did not cause hemagglutination of chicken, turkey, guinea pig, horse or human-O-group RBCs. We note that weak hemagglutinin (HA) activity (up to 2 4 HAU) was observed when we purified and concentrated the APMV-5 strain Kunitachi virus (not shown). However, we observed a similar level of weak HA activity with other purified preparations of viruses that are known to lack HA activity and were analyzed in parallel as negative controls, namely blue tongue virus, infectious bovine rhinotracheitis, and infectious bronchitis virus, and even uninfected Vero cell lysates gave a titer of up to 2 2 HAU. Thus, we concluded that this weak HA activity was a non-specific artifact of using highly concentrated virus. These results identified strain Kunitachi as the only APMV analyzed to date that appears to lack a typical virion HA activity. Twelve different cell lines from mammalian and avian species (BHK-21 HEp-2, Vero, HeLa, MDCK, DF-1, MDBK, PK-15, RK-13, CCL-141, QT-35 and BT) and two primary chicken cell cultures (CEF and CEK) were evaluated to identify cell substrates that can support the replication of APMV-5. The virus was subjected to three passages in each cell line, and the peak titer was determined by plaque assay. The virus grew most efficiently in chicken DF-1 cells, followed by Vero and BHK-21 cells. The peak titers (PFU/ml) of the different cell lines tested were: Vero, 6.5610 7 ; BHK-21, 4.6610 5 ; DF-1, 4.4610 10 ; HeLa, 3610 5 ; MDCK, 2.8610 5 ; MDBK, 3610 3 ; HEp-2, 1.1610 3 ; QT-35, 2610 2 ; CEK, 1.3610 3 ; and CEF, 2.5610 3 . Virus growth was not detectable in PK-15, RK-13, BT and CCL-141 cells using plaque assay but RT-PCR analysis of the infected cell culture supernatant and the cell pellet indicated a low level of virus replication in each of these cell cultures. Interestingly, APMV-5 failed to grow in the allantoic cavity of 9-day-old embryonated chicken eggs, which is contrary to all other APMV serotypes. But APMV-5 grew in the amniotic cavity of 8-day-old embryonated chicken eggs, yielding a titer of 1.2610 3 PFU/ml. The virus did not cause death of the chicken embryos in a mean death time (MDT) assay even after 168 h post infection. The intracerebral pathogenicity index (ICPI) in one-day-old chicks for APMV-5 strain Kunitachi was zero, indicating the virus is non pathogenic for chickens.

Determination of the Complete Genome Sequence of APMV-5
We determined a complete consensus sequence of the genome of APMV-5 strain Kunitachi (GenBank accession no. GU206351).
RT-PCR products spanning much of the genome were prepared using degenerate PCR primers derived from consensus sequences identified by sequence alignment of multiple members of the Avulavirus, Rubulavirus, Respirovirus, and Morbillivirus genera (see Materials and Methods). Based on the sequences of these initial cDNAs, additional primers were designed to complete the analysis. The entire sequence was confirmed in uncloned PCR products, providing a consensus sequence.
The genome organization of APMV-5 strain Kunitachi is 39leader-N-P/V/W-M-F-HN-L-trailer-59 (Fig 1), similar to the organization of the other members of genus Avulavirus except for APMV-6, which has an additional small hydrophobic (SH) protein gene between the F and HN genes [19]. Thus, each of the APMV serotypes contains six genes except for APMV-6, which has seven. Major features of the genome of APMV-5 strain Kunitachi are summarized in Fig. 1 and are compared with prototype members of the other APMV serotypes. The APMV-5 genome consists of 17,262 nt and thus is the longest among the APMVs. The genome nt length is a multiple of six, conforming to the ''rule of six'', as is the case for all members of subfamily Paramyxovirinae analyzed to date [4,27].
Although the genome length of APMV-5 is substantially greater than for the other serotypes, this was not reflected in increased lengths for the encoded proteins: the amino acid lengths predicted for the APMV-5 proteins are very similar to those of the cognate proteins in the other APMV serotypes (Fig. 1). Instead, the increased genome length of APMV-5 was mostly due to the presence of unusually long 39 -292) and (69-291), respectively. Consistent with the longer lengths of some of the UTRs of APMV-5, the percentage of the APMV-5 genome that encodes proteins is 81%, which is less than average coding percentage (92%) of other members of subfamily Paramyxovirinae [1,28].
The 39-leader sequence of APMV-5 consists of 55 nt, a length that is conserved among all of the APMV and among almost all of the members of the subfamily Paramyxovirinae. The APMV-5 leader has a high degree of nucleotide sequence identity with those of the other APMVs (Fig. 2a). The sequence of the first 12 nt of the leader region of APMV-5 (39-UGGUUUGUUCCU-59, negativesense) is identical to that of APMV-2, -6, -7, -8 and -9. The first 8 nt of the leader sequence are the same for all the members of genus Avulavirus sequenced to date except for a 1-nt change in the third position of APMV-3, where residue G is replaced with  residue A, and 2 nt changes at the third and seventh positions of APMV-4, involving G to C and G to U changes, respectively.
The length of the 59-trailer region of APMV-5 is 552 nt, which is second longest next to APMV-3 (707 nt), whereas the shortest trailer region among the APMV serotypes is 17 nt, for APMV-4.
The length of the trailer region is variable among the members of the subfamily Paramyxovirinae. The sequence of the last 12 nt of the APMV-5 trailer sequence is identical to those of APMV-6 and -7, and 11 nt out of the last 12 nt are identical to those of APMV-1 and -9 (Fig. 2b). The first 12 nt of the leader and trailer sequences  of APMV-5 are 100% complementary to each other, which is suggestive of conserved promoter elements at the 39 termini of the genome and antigenome (Fig. 2c).

Sequences of the GS and GE Motifs and the IGS
The GS sequences of the various genes of APMV-5 are identical at positions 1, 3, 4 and 5, with a consensus sequence (negativesense) of 39-CCCCCUUNNN-59 ( Table 2). The GE sequences are identical at positions 1-3 and 6-10, with a consensus (negativesense) of 39-AAU(A/U)NU 5 -59 (Table 2). Interestingly, the GS and GE signals of the APMVs sequenced to date have a high degree of identity, implying that polymerase recognition of these cis-acting sequences is conserved within the genus ( Table 3)
The P gene contains a putative editing site 39-UUUUUUCCC-59 (negative-sense) at nt positions 537-546 in the P gene (that corresponds to nt positions 2361-2369 in the genome). The sequence of the APMV-5 editing site is exactly same as that in APMV-7 and is similar to those in the other APMV serotypes (Fig. 3a). The addition of a single G residue at the APMV-5 editing site would produce a V mRNA encoding a V protein of 277 aa, with a predicted MW of 31,180. The N-terminal 182 aa domain of the V protein is identical to the corresponding domain in the P protein and is followed by a V-specific C-terminal domain of 95 aa. This V-specific domain contains seven cysteine residues that are exactly conserved in number and spacing in other members of the subfamily Paramyxovirinae (Fig. 3b). The V protein of APMV-5 has an aa sequence identity of 25%, 28%, 19%, 26%, 30%, 27%, 30% and 23% with the V proteins of APMV-1, -2, -3, -4, -6, -7, -8 and -9, respectively. The addition of  two G residues at the editing site would produce a W mRNA encoding a W protein of 187 aa, with a predicted MW of 20,727. The putative W protein contains an N-terminal domain of 182 aa that is identical to that of P followed by five additional residues that are unique to W. The W protein of APMV-5 has an aa sequence identity of 16%, 22%, 15%, 28%, 28%, 22%, 22% and 24% with the W proteins of APMV-1, -2, -3, -4, -6, -7, -8 and -9, respectively.

The Matrix (M) Gene and the M Protein
The M gene of APMV-5 is 1755 nt long and encodes a M protein of 366 aa with a predicted MW of 39,999. The pI value of the M protein is 9.31: this basic nature is thought to mediate ionic interactions with the acidic N protein [1]. The M proteins of some members of the subfamily Paramyxovirinae contains a motif (242-KKTNRKGADRSVLQIKEKVRK-262 in APMV-5) that consists of a bipartite clustering of basic residues. This motif is thought to serve as a nuclear localization signal (NLS) for the protein [6,30], although the significance of this for infection remains unclear. A protein-protein interaction domain called the 'late domain', which contains the motif FPIV and is involved in assembly and budding of viruses, was first identified in the paramyxovirus SV5 [31].  (Table 5). In the late domain motif, the amino acid proline (P) in the second position is conserved in all the members of genera Avulavirus and Rubulavirus, with the consensus pattern a-P-W-W (a represents an aliphatic amino acid and W represents an aromatic amino acid). However, this motif is not observed in the M proteins of members of genera Respirovirus, Morbillivirus and Henipavirus of the subfamily Paramyxovirinae. The M protein of APMV-5 has an aa sequence identity of 33%, 43%, 31%, 28%, 55%, 45%, 42% and 29% with the M proteins of APMV-1, -2, -3, -4, -6, -7, -8 and -9, respectively ( Table 5). The APMV-5 M protein has an aa sequence identity of 28%, 22%, 20%, 21%, and 14%, with the M proteins of MuV (Rubulavirus), SeV (Respirovirus), MeV (Morbillivirus), NiV (Henipavirus), and AMPV (Metapneumovirus), respectively.

The Fusion (F) Gene and the F Protein
The F gene of APMV-5 is 2028 nt long and encodes a F protein of 544 aa with a predicted MW of 58,724 and pI of 8.34. The F protein is predicted to be a type I transmembrane protein similar to the F proteins of the other members of the family Paramyxoviridae. It contains a signal sequence of 25 aa (positions 1-25) at the N-terminal end and a predicted transmembrane domain located at aa sequence positions 494-516. The putative cleavage site of the APMV-5 F 0 protein precursor is G-K-R-K-K-RQF (aa positions 104-110), with the residue F representing the newly-created N-terminus of the F1 fusion peptide. The cleavage site sequence of APMV-5 conforms to the favored sequence motif for cleavage by the intracellular protease furin, R-X-K/R-RQF [32]. Interestingly, the cleavage site of APMV-5 contains five tandem lysine and arginine residues (KRKKR), which is the greatest number of basic residues for any of the APMV (Table 6) except for some strains of APMV-1 [33]. The F1 polypeptide of APMV-5 contains heptad repeats at aa positions 134-182 and 455-485, as predicted by the LearnCoil-VMF program, which correspond to the characteristic HRA and HRB heptad repeats that are found in paramyxovirus F proteins and are thought to be important for the conformational changes that lead to membrane fusion [34]. The APMV-5 F protein contains five potential Nlinked glycosylation sites located at aa position 74 of the F2 subunit and at aa positions 189, 359, 440 and 464 of the F1 subunit, as predicted by the NetNGlyc 1.0 program of the Expasy proteomics server. The F protein of APMV-5 has an aa sequence identity of 41%, 47%, 31%, 33%, 55%, 37%, 46% and 37% with the F proteins of APMV-1, -2, -3, -4, -6, -7, -8 and -9, respectively ( Table 5). The APMV-5 F protein has an aa sequence identity of 33%, 24%, 27%, 24%, and 19%, with the F proteins of MuV (Rubulavirus), SeV (Respirovirus), MeV (Morbillivirus), NiV (Henipavirus) and AMPV (Metapneumovirus), respectively.

Phylogenetic Tree Analyses
A phylogenetic tree was constructed from a sequence alignment of the complete genome nt sequences of the prototype members of the nine APMV serotypes (Fig. 4a). In addition, phylogenetic trees were prepared for the N, P, M, F, HN and L proteins of APMV-5 versus the cognate proteins of members of the other eight APMV serotypes and representative viruses of the other four genera of subfamily Paramyxovirinae (trees for the F and L protein are shown in Fig 4b and c, respectively). The phylogenetic trees clearly indicated that the APMV-5 was most closely related to the known APMVs, consistent with its inclusion in the genus Avulavirus. Furthermore, the phylogenetic analysis of the complete genome sequence and the sequences of the six proteins revealed a closer genetic relatedness of APMV-5 to APMV-6 than to the other APMV serotypes within the genus Avulavirus.

Discussion
The APMVs are frequently isolated from a wide variety of avian species and are grouped into nine serological types based on antigenic tests involving the HN protein. Until recently, very little was known about the molecular details of these common viruses, except for NDV (APMV-1). However, recent sequence analysis of prototypes of APMV-2, -3, -4, -6, -7, -8 and -9 has revealed many of the molecular details of these viruses [16,17,18,19,20,21,22]. Prior to this report, APMV-5 was the only remaining serotype whose genome sequence was not available. As a first step towards the molecular characterization of APMV-5, we have determined the growth characteristics and the complete genome sequence of APMV-5 strain Kunitachi. APMV-5 grew in many cell lines from different species, suggesting a broad in vitro host range for the virus. Interestingly APMV-5 grew in the amniotic cavity rather than the allantoic cavity of embryonated chicken eggs: the initial report describing the first isolation of APMV-5 had suggested that the virus did grow in the allantoic cavity of chicken eggs [24], but our findings are in agreement with a subsequent study reporting growth in the amniotic rather than allantoic cavity [23]. Also, APMV-5 strains have been reported to grow in the yolk sac of 6day-old embryonated chicken eggs [42]. Infection of embryonated chicken eggs with 10 3 Pfu/ml of APMV-5 via the amniotic sac did not cause embryo death even after 5 days post-infection, suggesting that the virus is avirulent in chickens.
Pairwise comparisons of the complete genome sequences of the prototype strains of the nine APMV serotypes showed that they shared 42 to 60% nt sequence identity (Table 1), and comparison of the amino acid sequences of cognate proteins of the various APMV serotypes also provided evidence of substantial divergence (Table 5 and data not shown). Thus, the serotype classification, which was originally based solely on the reactivity of antibodies that inhibit HA and NA activities, has now been substantiated by sequence analysis showing extensive divergence across the genome and involving all of the proteins. The genome of APMV-5 strain Kunitachi is 17,262 nt in length, which is the longest genome of all of the APMVs analyzed to date (which have the following lengths: APMV-1, 15,186 nt; APMV-2, 14,904 nt; APMV-3, 16,272 nt; APMV-4, 15,054 nt; APMV-6, 16,236 nt; APMV-7, 15,480 nt; APMV-8, 15,342 nt; and APMV-9, 15,438 nt). The APMV-5 genome also is longer than the average genome size (,15,500 nt) of members of Respirovirus, Morbillivirus, Rubulavirus and Avulavirus sequenced to date, but is somewhat shorter than those of HeV (18,234 nt) and NiV (18,252 nt) of genus Henipavirus as well as those of some of the unclassified members of subfamily Paramyxovirinae such as Beilong virus, which has the longest paramyovirus genome (19,212 nt) identified to date [43]. The increased lengths of the genomes of HeV and NiV are mostly due to long downstream non-coding regions, whereas the increased genome length of the Beilong virus is due to an unusually large attachment protein, the presence of additional genes, and the presence of genome regions that may be either unusually long UTRs or may contain additional functional open reading frames. In the case of APMV-5, the increase in genome length compared to the other APMV serotypes was not due to additional genes or longer proteins, but rather was due to unusually long downstream UTRs in the P, M, and HN genes as well as a long trailer region. This suggested that, in the process of divergent evolution by the APMVs the change that was most readily accommodated was in the length of the downstream UTRs and the trailer region.
Apart from its longer length, the genome of APMV-5 had general similarities with the genomes of the other APMVs. The APMV-5 genome has six non overlapping genes (N, P, M, F, HN and L), similar to the genomes of all of the AMPVs except APMV-6, which has in addition the SH gene between the F and HN genes [19]. The lengths of the predicted APMV-5 proteins were very similar to those of the cognate proteins in other APMVs. Like the other APMVs, the genome length of APMV-5 is consistent with the 'rule of six', which appears to be a characteristic of all members of the subfamily Paramyxovirinae [3]. The leader sequence of APMV-5 is 55 nt long, a length that was exactly conserved among the other eight APMV serotypes and is generally conserved among members of the subfamily Paramyxovirinae [1]. The first 12-13 nt of the leader sequences in the subfamily Paramyxovirinae are more closely related within, than between genera, and this criteria is used in the classification of new isolates. Consistent with this, the first 12 nt of the APMV-5 leader sequence were found to be conserved among APMV -2, -6, -7, -8 and -9, suggesting a close evolutionary relationship among these viruses. The trailer region of APMV-5 is 552 nt in length, which is longer than the typical length (40-60 nt) for most members of the family Paramyxoviridae [44], and is the second longest trailer among APMVs, next to APMV-3 (707 nt) [17]. The first 12 nt of the 39 leader sequence of APMV-5 are complementary to the last 12 nt sequence of the 59 trailer regions suggesting the presence of conserved elements in the 39 promoter regions of the genome and antigenome.
Comparison of the GS and GE sequences of APMV-5 with those of the other members of genus Avulavirus showed that the GE sequences are somewhat more highly conserved than the GS sequences. All of the GS sequences are rich in C residues (negative-sense), especially for the first five positions. However, only two positions were conserved among all nine serotypes, namely the C residues at positions 3 and 5. The GE sequences were rich in A and U (negative-sense): the first three positions of the signals of all nine serotypes were identical (AAU), and all of the sequences ended with a U tract that encodes the poly A tail. While all of the APMV genomes conform to the rule of six, there was no consistent pattern to the hexamer phasing positions of the GS and GE signals other than that of the N gene being at position 2 in all nine serotypes. Thus, while the importance of the hexamer phasing positioning is well-known for cis-acting signals involved in RNA replication [1], the importance of the phasing of transcription signals remains unclear. The IGS of APMV-5 differed from each other in sequence and varied in length from 4 to 57 nt. This variable, non-conserved nature of the IGS is shared with the other eight serotypes of APMV and also with the members of Rubulavirus, but differs from the conserved trinucleotide (usually GAA) IGS that is characteristic of Respirovirus, Morbillivirus, and Henipavirus [45]. With regard to the UTRs, the downstream UTRs (39 in positive-sense) were longer than the upstream UTRs. This also is the case for the other APMVs (with the exception of the UTRs of the APMV-2 F gene and APMV-4 HN gene), and thus APMV-5 conforms to this general pattern. However, APMV-5 has unusually long 39UTR in the P (458 nt), M (531 nt), and HN (378 nt) gene. Although the significance of UTR length in APMV-5 is not known, UTRs have been shown to modulate transcription and translation in APMV-1 [46], and measles [47] and canine distemper viruses [48].
APMV-5 strain Kunitachi also had other motifs in common with the other APMV serotypes and other members of the subfamily Paramyxovirinae. These include a conserved hydrophobic motif (F-X 4 -Y-X 4 -SYAMG) in the N protein that is thought to be involved in N-N self-assembly, a putative nuclear localization signal in the M protein, a putative ''late domain'' sequence in the M protein that may be involved in virion assembly, a putative sialic acid binding site in HN, proposed neuraminidase active-site residues in HN, and conserved polymerase and nucleotide-binding domains in the L protein. In addition, the editing motif and editing products of the APMV-5 P gene conform to those of the other APMV serotypes. The P gene of all members of the subfamily Paramyxovirinae is subjected to an RNA editing mechanism that causes a frame shift to access additional internal ORFs: the sole known exception is hPIV-1 of the genus Respirovirus. There are two patterns of editing within the subfamily. One pattern, which is followed by subfamily Paramyxovirinae except for genus Rubulavirus, involves expression of the P protein from the unedited mRNA and expression of the V protein from an mRNA that was edited to insert a single non-templated G residue. The insertion of two G residues results in an mRNA encoding the W, I, or D protein, depending on the virus. Editing by this group of viruses takes place at the editing motif 39 UUUUUCCC 59 (negative-sense), although this sequence varies somewhat in J virus and certain other unclassified viruses [1,49]. Members of the Rubulavirus genus follow a second pattern in which the unedited mRNA encodes the V protein, and mRNA edited by the insertion of two G residues encodes the P protein [1]. The insertion of a single G residue results in expression of the I protein. The editing site and predicted protein products of the APMV-5 P gene are consistent with those of the other APMVs analyzed to date, and conform to the first pattern described above.
The F protein of the family Paramyxoviridae mediates fusion between the viral envelope and host plasma membrane at neutral pH. The F protein is synthesized as an inactive protein (F 0 ) that is cleaved by host cell protease into the active form consisting of disulfide-linked F 2 2F 1 subunits [1]. The sequence of the F protein cleavage site is a major determinant of NDV pathogenicity in chickens because it plays a major role in determining where and how the F 0 precursor is cleaved [1,50]. The cleavage sites of all known virulent strains of NDV have multiple basic residues (bold and underlined) that contain the preferred cleavage site of the intracellular cellular protease furin (R-X-K/R-RQF) present in most cell types. In contrast, the cleavage site sequences of avirulent NDV strains characteristically have one or a few basic residues immediately upstream of the cleavage site that do not create a furin cleavage site. The ability of the NDV F protein to be cleaved by furin provides the opportunity for viral replication in a wide variety of tissues, which is essential for systemic spread of the virus. The presence of a non-furin cleavage site as in avirulent NDV strains restricts viral replication to the respiratory and enteric tracts, where secretory proteases necessary for F cleavage are present. The ability of NDV F to be cleaved also is affected by the amino acid that immediately follows the cleavage site and becomes the N-terminal residue of the F1 subunit: the presence of leucine in this position is associated with less efficient cleavage [34]. Surprisingly, however, some of the other APMV serotypes do not conform to the NDV paradigm. For example, for a number of the other eight serotypes, the sequence of the cleavage site does not predict whether exogenous protease is needed for growth in vitro. Specifically, the F protein cleavage sites of APMV-2 (PASRQF), APMV-4 (IQPRQF) and APMV-7 (PSSRQF) all contain a single basic residue (R) immediately upstream of the putative cleavage site, but do not require exogenous protease supplementation for growth in cell culture [16,18,20]. Other APMV serotypes do conform more closely with the NDV paradigm: specifically, they have a non-furin cleavage site and do require exogenous protease for replication in vitro. These viruses are APMV-3 (PRGRQL), APMV-8 (PQTRQL), and APMV-9 (REGRQI). It also is noteworthy that these three serotypes have leucine or isoleucine immediately downstream of the cleavage site. APMV-6 also has a single basic residue immediately upstream of the putative cleavage site (PEPRQL), but the requirement for trypsin varies with the strain (S. Xiao, P.L.C. and S.K.S, unpublished data). APMV-5 strain Kunitachi contains five tandem basic amino acid residues at the putative cleavage site ( 105 KRKKRQF 110 ), and thus contains even more basic residues than most of the virulent strains of NDV sequenced to date. The presence of this highly basic cleavage sequence in APMV-5 may be responsible for the high pathogenicity of this virus in budgerigars, where it can cause up to 100% mortality [23]. However, it is interesting that this same virus was completely avirulent in chicken eggs. This is suggestive of a host range difference that is not ameliorated by the presence of a multibasic cleavage site that otherwise would confer virulence in chickens. Similarly, some of the other APMV serotypes that did not require trypsin for cleavage in vitro nonetheless had low virulence in chickens (e.g. APMV-2, -4, and -7). Thus, for a number of the APMV serotypes, the cleavage site sequence did not predict the cleavage phenotype in vitro nor virulence in vivo.
All members of genus Avulavirus (APMV-1 to 9) have both hemagglutination and neuraminidase activities except for APMV-5 [23]. Although the APMV-5 HN protein has neuraminidase activity, it neither hemagglutinated nor hemadsorbed RBCs of chickens, turkeys, guinea pigs, horses, or human O-group, either at 37uC or at 4uC. The sequence of the APMV-5 stain Kunitachi HN protein provides evidence of the expected globular head, N-terminal-proximal signal/anchor domain, putative sialic acid binding motif (NRKSCS), seven conserved residues of the proposed neuraminidase active site, and eleven conserved cysteine residues that have been shown to be conserved among type II integral membrane proteins of subfamily Paramyxovirinae [38]. Thus, the lack of HA activity for APMV-5 cannot be ascribed to an obvious difference in HN structure. Non-hemagglutinating highly virulent APMV-1 strains which grow in the allantoic cavity, have been reported from cormorants (also a psittacine bird) in Canada [51]. These observations indicate that the HA activity of the HN protein is not required for replication of APMVs. APMV-5 failed to grow in the allantoic cavity of embryonated chicken eggs, in contrast to all hemagglutinating APMVs, which grow in the allantoic cavity. It is tempting to suggest that there is a link between these observations: specifically, that the lack of a strong HA in APMV-5 renders the virus unable to bind to receptors present on cells of the allantoic cavity. However, this hypothesis requires further study.
In summary, APMV-5 Kunitachi strain differs significantly from all other APMVs in the following ways; (i) having isolated only from budgerigars, (ii) the presence of an HN protein with no HA activity, (iii) the inability to grow in the allantoic cavity of embryonated chicken eggs, (iv) the presence of unusually long 39UTRs in P, M and HN genes, and (v) the longest genome among the APMVs. The sequence alignment and phylogenetic analysis showed that APMV-5 clusters with the other APMVs, justifying its classification in the genus Avulavirus. Phylogenetically, APMV-5 is more closely related to APMV-6 than to the other known APMVs. Further sequence analysis of additional APMV-5 strains will help us understand the antigenic and genetic variation among APMV-5 strains. Future studies on the epidemiology and pathogenicity of this virus in different avian species will be necessary to define the role of this virus in avian health.