A Novel P/V/C Gene in a New Member of the Paramyxoviridae Family, Which Causes Lethal Infection in Humans, Horses, and Other Animals

Abstract

In 1994, a new member of the family Paramyxoviridae isolated from fatal cases of respiratory disease in horses and humans was shown to be distantly related to morbilliviruses and provisionally called equine morbillivirus (K. Murray et al., Science 268:94–97, 1995). To facilitate characterization and classification, the virus was purified, viral proteins were identified, and the P/V/C gene was cloned and sequenced. The coding strategy of the gene is similar to that of Sendai and measles viruses, members of the Paramyxovirus and Morbillivirus genera, respectively, in the subfamily Paramyxovirinae. The P/V/C gene contains four open reading frames, three of which, P, C, and V, have Paramyxovirinae counterparts. The P and C proteins are larger and smaller, respectively, than are cognate proteins in members of the subfamily, and the V protein is made as a result of a single G insertion during transcription. The P/V/C gene has two unique features. (i) A fourth open reading frame is located between those of the C and V proteins and potentially encodes a small basic protein similar to those found in some members of the Rhabdoviridae and Filoviridae families. (ii) There is also a long untranslated 3′ sequence, a feature common in Filoviridae members. Sequence comparisons confirm that although the virus is a member of the Paramyxovirinae subfamily, it displays only low levels of homology with paramyxoviruses and morbilliviruses and negligible homologies with rubulaviruses.

The virus known as equine morbillivirus (EMV) first appeared in Hendra, a suburb of Brisbane, Australia, in September 1994 and was responsible for a brief but dramatic respiratory-disease outbreak, which resulted in the deaths of 14 horses and a horse trainer (29, 38). Over 1 year later, the virus was shown to be the cause of encephalitis and death of a horse breeder in Mackay, approximately 800 km north of Brisbane (31). The horse breeder probably became infected with EMV the previous year by assisting in the autopsies of horses which died with severe respiratory disease and were subsequently shown to have been infected with EMV (16). No direct link has been made between the two disease outbreaks (35). The recent finding that some 15% of Australian fruit bats (genus Pteropus), commonly known as flying foxes, have antibodies to EMV and the isolation of EMV-like viruses from bat uterine fluids suggest that fruit bats are a natural host for the virus (13, 30).

Virological and morphological analyses and sequence determinations of the M and F genes and part of the P gene indicated that the virus responsible for the original outbreak belonged to the family Paramyxoviridae (12, 17, 29). Sequence comparisons indicated that although homologies with other viruses were limited, the virus was more closely related to members of the Morbillivirus genus than to other Paramyxoviridae genera. Members of the Paramyxoviridae family contain predominantly monocistronic genes which are initiated at a single start codon and produce a single primary translation product. The P/V/C gene is a notable exception because it can produce multiple translation products both by using a series of initiation codons and overlapping reading frames (ORFs) and by an unusual cotranscriptional insertion of nontemplated nucleotides (6, 23). In addition to the major P protein, members of the Paramyxovirus and Morbillivirus genera produce a small basic (SB) protein (C) and a V protein, which shares the N-terminal region of P; they may generate other translation products, such as D and W proteins (23, 33). In members of the Paramyxovirus and Morbillivirus genera, as exemplified by Sendai virus (SeV) and measles virus (MeV), respectively, the primary mRNA transcript encodes the P protein and a single G insertion generates the V mRNA (2, 44). In contrast, members of the Rubulavirus genus do not produce a C protein and the unedited transcript of the gene codes for the V protein. A nontranscriptional insertion of two G residues generates the P protein mRNA (23).

Here we report the cloning and characterization of the P/V/C gene and the P protein of the new member of the Paramyxoviridae family known as EMV. The P protein is significantly larger than are other P proteins within the family, whereas the C protein is the smallest Paramyxoviridae C protein sequenced so far. Although the overall coding strategy for the P/V/C gene is similar to those of SeV and MeV, i.e., the primary transcript and one-G-insertion mRNA code for P and V proteins, respectively, there are features unique to this virus, such as the presence of an ORF between the C and V ORFs and an extended 3′ untranslated tail in the P gene mRNA. These novel features and the lack of extensive serological cross-reactivities with members of the Morbillivirus genus (29) indicate that the provisional name of the virus, EMV, is inappropriate. Murray et al. (30) have suggested that the virus be called Hendra virus (HeV), after the Brisbane suburb from which it was first isolated. The length of the virus genome and the presence of long untranslated sequences in N, M, F, and G mRNAs (unpublished data) distinguish HeV from other Paramyxoviridae viruses and reinforce the need not only to adopt a more appropriate name for the virus but also to consider whether it should be classified in a new genus within the Paramyxovirinae subfamily (30).

MATERIALS AND METHODS

Cells and virus.

Vero cells were grown in Eagle’s minimal essential medium supplemented with 10% fetal calf serum. HeV was isolated and plaque purified as previously described (17, 29). Stock preparations (titer, 2 × 10⁷ to 4 × 10⁷ 50% tissue culture infective doses [TCID₅₀]/ml) were generated by infection at a multiplicity of infection of 10⁻⁴ TCID₅₀/cell. Virus was grown under biohazard level 4 conditions, in which laboratory workers wear positive-pressure encapsulating suits supplied with breathing air.

Virus purification.

Vero cells were infected at a multiplicity of infection of 10⁻² TCID₅₀/cell and incubated in Eagle’s minimal essential medium containing 1% fetal calf serum at 37°C. At 48 h postinfection, culture medium was removed and clarified by centrifugation at 10,000 × g for 10 min in a type 19 rotor (Beckman). Virus was pelleted by centrifugation at 150,000 × g for 20 min in a type 55.2 rotor, resuspended in TNE (10 mM Tris, 100 mM NaCl, 1.5 mM EDTA [pH 7.4]) at 4°C overnight, layered onto linear 15 to 50% sucrose gradients in TNE, and centrifuged at 200,000 × g in an SW41 rotor for 20 min at 10°C. The virus band was removed, diluted in TNE, and pelleted at 200,000 × g in an SW41 rotor for 20 min at 10°C. The pellet was resuspended in TNE and used as a source of genomic RNA or ribonucleoprotein or inactivated by heating in 2% sodium dodecyl sulfate (SDS) at 100°C for 2 min prior to removal from the biohazard level 4 laboratory.

Construction of a cDNA library in the pZEro-1 cloning system.

RNA was extracted from SDS-treated virus by standard methods, and 2 to 4 μg was used for cDNA synthesis with a TimeSaver cDNA synthesis kit from Pharmacia Biotech. Random hexamer primers (0.037 μg) were used for the synthesis of relatively long cDNA. Double-stranded cDNA with EcoRI adapters at both ends was ligated with EcoRI-digested vector pZEro-1 (Invitrogen, San Diego, Calif.), followed by electroporation into Escherichia coli host strain TOP10F′ {F′ [lacI^q Tn10 (Tet^r)] mcrA Δ(mrr-hsdRMS-mcrBC) φlacZΔM15 ΔlacX74 deoR recA1 araD139 Δ(ara-leu)7697 galU galK rspL (Str^r) endA1 nupG} and selection for the growth of recombinants in the presence of zeomycin (1). PCR screening of approximately 100,000 recombinants in a primary library with primers which anneal to flanking regions of the cloning site in the vector revealed that 70 to 75% clones contained inserts in the range from 0.2 to 2.0 kb. Plasmid DNAs from these recombinants were isolated as follows. About 2 μg of purified plasmid DNA was digested in separate tubes with restriction enzymes PstI, SacI, and XbaI, which cut once in the vector multiple cloning site. Linearized plasmid DNA was purified from a 1% agarose–TAE (0.04 M Tris-acetate, 2 mM EDTA) gel with a GeneClean kit (Bio 101, San Diego, Calif.), and 4- to 7-kb fragments from the three digestion mixtures were combined, self-ligated, and reintroduced into TOP10F′ by electroporation to form a secondary library. PCR screening indicated that about 90% of clones in the secondary library contained inserts ranging from 0.4 to 2 kb, with a majority around 0.7 to 1.2 kb.

To test library quality and determine the frequency of positive clones, the secondary library was screened with PCR-generated DNA probes to the M and F genes, designed according to previously published sequence (12). Both probes detected positive clones at a rate of 0.1 to 0.2%. This was much lower than predicted on the basis of the average size of inserts in the library and a genome size of 15 to 16 kb for Paramyxoviridae family members. The low frequency of positive clones was confirmed when the sequences of 30 random clones were not found in the 4.5 kb of known sequence derived from the middle of the genome (12). A genome walking strategy, rather than random sequencing, was used to determine viral sequences in the library.

Library screening and characterization of overlapping cDNA clones.

The following two PCR approaches were used for the generation of specific gene probes to screen the cDNA library: (i) conventional PCR amplification with two specific primers that anneal to the ends of a known sequence and (ii) suppression PCR amplification with only one gene-specific primer (39). Approximately 1 μg of plasmid DNA purified from the primary library was digested with blunt-end-generating enzymes RsaI, HaeIII, and HincII, and fragments were ligated with 20 pmol of an adapter (top strand, 5′ CTAAT ACGAC TCACT ATAGG GCTCG AGCGG CCGCC CGGGC AGGT-3′; bottom strand, 5′-P_i-ACCTG CCC-NH₂-3′) to form a random adapter-anchored library to be used as template DNA in suppression PCR. One microliter of a 1:500 dilution of the ligation mixtures discussed above was used for PCR amplification with the following cycle parameters. For primary PCR, the first step consisted of 7 cycles at 94°C for 25 s and 72°C for 3 min and the second step consisted of 32 cycles at 94°C for 25 s and 67°C for 3 min, followed by an additional 7 min at 67°C after the final cycle. Similar conditions were used in secondary PCR, except that 5 and 20 cycles were used for the first and second steps, respectively. Primers AP1 (5′-GGATC CTAAT ACGAC TCACT ATAGG GC-3′) and AP2 (5′-AATAG GGCTC GAGCG GC-3′), which anneal to the adapter sequence, were used in primary and secondary PCRs, respectively, together with a gene-specific primer that anneals to the end of a known cDNA sequence. PCR fragments (usually in the range of 150 to 300 bp) generated by either of these approaches were purified from a 2% agarose–TAE gel with a GeneClean kit before being labeled with [³²P]dATP by using a Ready-To-Go DNA labeling kit (Pharmacia Biotech). Colony hybridization was carried out with a colony/plaque screen hybridization transfer membrane (NEN Research Products, Du Pont, Boston, Mass.). From four rounds of screening by a genome walking strategy, a total of six overlapping cDNA clones, which covered the entire P gene, were isolated. Sequences from the overlapping clones were determined with universal and reverse sequencing primers, which annealed to the flanking regions of the pZEro-1 vector cloning sites, and gene-specific primers derived from information obtained by genome walking analyses. The nucleotide at each position was sequenced at least twice, either from two overlapping clones or from both strands. The complete cDNA sequence derived from overlapping cDNA clones was corroborated by direct sequencing of PCR fragments obtained from the viral genome by reverse transcription-PCR (RT-PCR) (see below). Except for two nucleotide changes (only one of which resulted in a change in the amino acid sequence), the genome-derived sequence was identical to that derived from individual cDNA clones.

Synthesis of cDNA by RT-PCR.

Total nucleic acid was extracted from virus-infected Vero cells with TRIZOL reagent (Life Technologies) by a single-step isolation procedure (3). Virus-specific cDNAs were synthesized by using the Superscript preamplification system (Life Technologies) with gene-specific primers, followed by PCR amplifications with the following parameters: for primary PCR, denaturation at 94°C for 1 min, annealing at 50°C for 2 min, and elongation at 72°C for 2 min for 30 cycles, followed by an additional 7 min at 72°C after the final cycle; for secondary PCR, the same conditions as those used for primary PCR with internal primers.

Determination and analysis of nucleotide sequence.

PCR products, derived from cloned cDNA or genomic RNA by RT-PCR, were sequenced with Sequenase PCR sequencing kits (Amersham Life Science). Sequences were manipulated by using the Clone Manager program (version 4.01; S&E Software). Multiple-sequence alignments were done by using ALIGN Plus (version 3.0; S&E Software), CLUSTAL W (version 1.6) (42), and the SAM sequence alignment and modelling software system (22). Phylogenetic analyses were performed by maximum-parsimony (PROTPARS program) and distance matrix (PROTDIST and NEIGHBOR programs) methods in the PHYLIP package (10). Trees were generated from 100 random data sets by using the SEQBOOT program, and majority-rule bootstrapped trees were constructed by using the CONSENSE program (10). Only branches with greater than 50% bootstrap support were included.

In situ enzymatic digest and peptide mapping.

The method of Moritz et al. (28) was adapted to obtain internal amino acid sequence data after enzymatic digestion of proteins on gels. Proteins of SDS-treated virus were separated by polyacrylamide gel electrophoresis (PAGE) on Novex precast gels. Samples in 2% SDS were reduced and incubated for 5 min at 100°C. Two 10% gels were prerun with electrophoresis buffer containing 15 mg of reduced glutathione (Sigma) per liter for 90 min at 3 mA. Samples containing approximately 200 μg of total protein were loaded on each gel and separated at 100 V with glutathione-free electrophoresis buffer. Gels were stained for 30 min in 0.1% Coomassie brilliant blue R-250 and destained in 5% acetic acid–20% methanol. P protein bands were excised from gels and kept in airtight containers at −20°C. Two additional pieces of each gel containing no apparent protein were used as negative controls. Gel fragments were cut into 2-mm-long pieces and washed in 2 M NH₄HCO₃–50% acetonitrile for 30 min at room temperature with occasional mixing. The washing step was repeated twice. Gel pieces were dried in a SpeedVac concentrator (Savant) for 30 min and rehydrated in 20 μl of digestion buffer (100 mM NH₄HCO₃, 50% acetonitrile) for 30 min at 37°C. Modified trypsin or endoproteinase Glu-C (sequencing grade; Boehringer Mannheim) was used at an enzyme/substrate ratio of 1:10 in digestion buffer. Gels were incubated overnight at 37°C in 150 μl of digestion buffer and pelleted in a microcentrifuge. Supernatants containing peptide fragments were retained. Residual gel pieces were washed with 200 μl of 1% trifluoracetic acid (TFA) at 37°C for 30 min, centrifuged, and washed again with 200 μl of 0.05% TFA–80% acetonitrile at 37°C for 30 min. Supernatants were combined, concentrated in a SpeedVac to approximately 50 μl, diluted to 100 μl with 0.05% TFA in water, and subjected to reverse-phase chromatography.

Reverse-phase chromatography.

Chromatography of enzymatically digested protein samples was performed on a System Gold high-performance liquid chromatograph (Beckman) with diode array detectors (model 168; Beckman). Peptides were separated on a Vydac C₁₈ reverse-phase column (2.1 by 250 mm; Separations Group, Hysperia, Calif.) operated at 43°C and by using a binary gradient with 0.05% TFA as buffer A and 0.045% TFA–80% acetonitrile as buffer B. The gradient was developed as follows: 2 to 37.5% buffer B over 0 to 50 min, 37.5 to 75% buffer B over 50 to 75 min, and 75 to 100% buffer B over 75 to 85 min at a flow rate of 150 μl/min. About 40 fractions were collected manually and stored at −20°C prior to automated sequencing.

Edman degradation amino acid sequencing.

Protein and peptides were subjected to automated (Edman degradation) sequence analysis (8) with vapur phase delivery of critical reagents (15) in an automated sequenator (model 470A; Applied Biosystems) in conjunction with a PTH amino acid separation system (model 120A PTH analyzer; Applied Biosystems).

SDS-PAGE, Western blotting, and enzyme-linked immunosorbent assay.

Standard procedures were used for SDS-PAGE, Western blotting, and enzyme-linked, immunosorbent assay (45–47).

Nucleotide sequence accession number.

The sequence reported here has been assigned GenBank accession no. AF010304.

RESULTS

Virus proteins.

The proteins in purified HeV are shown in Fig. 1. Based on the presence of L, P, and N proteins in CsCl buoyant density-purified HeV ribonucleoprotein, the Triton X-100-mediated release of proteins G, F₀, F₁, and F₂ from purified virus, and the reactions of proteins on Western blots with antisera raised to bacterially expressed portions of individual proteins (unpublished data), the proteins were designated as indicated in Fig. 1. Like those of other Paramyxoviridae family members, the P protein of HeV is phosphorylated (data not shown). While the molecular masses of most proteins calculated on the basis of electrophoretic mobility were similar to those of proteins from other Paramyxoviridae viruses, the estimate of 98 kDa for the P protein was significantly larger than the masses of P proteins from other family members, such as human parainfluenza virus type 3 (hPIV3) (83 kDa) (48), MeV (70 kDa) (41), and mumps virus (68 kDa) (27). The amino acid sequences of many Paramyxoviridae P proteins have been predicted from nucleic acid sequence data, and P protein sizes range from 241 amino acids (aa) for pneumoviruses to 507 and 603 aa for parainfluenza viruses and morbilliviruses, respectively. To characterize the HeV P protein, the gene and protein were sequenced and compared with the P proteins of other Paramyxoviridae viruses.

FIG. 1.

Analysis of HeV structural proteins by PAGE. The molecular weights of Coomassie brilliant blue-stained marker and HeV proteins are indicated on the left and right, respectively.

Characteristics of the P/V/C gene and its coding capacity.

The complete sequences of the P/V/C gene and its potential coding regions are presented in Fig. 2. The predicted P/V/C gene starts and stops with highly conserved transcription initiation and termination signals, respectively (Fig. 2). The trinucleotide intercistronic sequence CTT, conserved in Paramyxoviridae viruses, was observed both upstream and downstream of the P gene (not shown). As summarized in Table 1, the mRNA is 2,698 nucleotides in length and appears to be capable of coding for several proteins, as has previously been observed for the P/V/C genes of other Paramyxoviridae viruses (23).

FIG. 2.

DNA, editing site, and ORF sequences of the HeV P/V/C gene. DNA and deduced amino acid sequences are shown in the first and second lines, respectively, of each set. Gene initiation and termination signals are underlined. ORF designations consisting of the encoded proteins and reading frames are indicated in bold and are followed by the deduced amino acid sequences of the encoded proteins. The conserved editing site for the insertion of a single G residue is in bold.

TABLE 1.

Size comparisons of P/V/C gene transcripts and translation products between HeV and selected paramyxoviruses and morbilliviruses

Virus	Size
	RNA features (nt)a			Potential coding regions (aa)
	mRNA	5′ noncoding	3′ end after P stop codon	ORF-P	ORF-C	ORF-V
HeV	2,698	105	468	707	166	54
SeV	1,893	103	83	568	204	71
hPIV3	2,014	80	126	603	199	69c
hPIV1	1,893	103	83	568	204	NAb
MeV	1,654	81	71	507	186	72
CDV	1,655	81	72	507	174	72
DMV	1,655	59	75	506	160	76

^ant, nucleotides. 

^bNA, not applicable (hPIV1 does not produce a V protein). 

^cGene present but not expressed (11). 

The largest ORF exists in the +1 reading frame and encodes a polypeptide of 707 aa (Table 1). Sequence homology analyses and protein sequencing (see below) confirmed this to be the P protein ORF. The size of the HeV P protein (78,324 Da) is consistent with its slower electrophoretic mobility, compared with those of cognate proteins from other Paramyxoviridae viruses. It has a calculated pI of 4.5 and contains 48 negative charges at neutral pH.

In the +2 reading frame, there are two potential ORFs. Downstream from the ATG codon of ORF-P, there are two in-frame, tandem ATG codons, followed by an ORF coding for a protein of 166 aa (Table 1). It is named ORF-C based on its similarity in location to the C proteins of other Paramyxovirinae viruses. The deduced C protein has a calculated molecular mass of 19,647 Da and a pI of 8.6. In the +2 ORF, there is another small ORF with potential coding capacity. It is named ORF-SB (for SB protein) because of the highly basic nature of the deduced polypeptide. The putative SB protein contains 65 aa, has a calculated molecular mass of 7,598 Da and a calculated pI of 11.9, and is arginine rich. A similar SB protein has previously been found in an overlapping reading frame in the P gene of vesicular stomatitis virus (VSV) (40) and the P gene equivalent, VP35, of Marburg virus (9).

In addition to these ORFs, there is short ORF (ORF-V) which could be expressed as a fusion to the P protein after the insertion of a unique nontemplated G, as has previously been observed for other Paramyxovirinae viruses (Table 1) (23). There is a highly conserved, AG-rich sequence (Fig. 2) which has previously been shown to be the G insertion site in other members of the Paramyxovirinae subfamily. The insertion of a single G residue at this site in HeV results in a P-to-V shift. ORF-V contains 54 aa, is cysteine rich, and has homologies with other Paramyxovirinae viruses (12) (see below).

The major features of the P/V/C gene and the P, V, and C proteins which are potentially expressed from it are summarized in Table 1 in comparison with counterparts from selected paramyxoviruses and morbilliviruses. The putative product of ORF-SB is not included in Table 1 because it appears to be unique to HeV and does not have correlates in other Paramyxoviridae family members. The expression of P, C, and V proteins in HeV-infected cells was confirmed with antibodies generated to bacterially expressed P-, C-, and V-specific peptides (data not shown).

Amino acid sequence analysis of the P protein.

Given the very low levels of sequence homology observed between the P/V/C gene of HeV and those of other Paramyxoviridae viruses (see below), amino acid sequence determinations of peptides derived from the P protein were used to corroborate the amino acid sequence deduced from the P/V/C gene sequence. The HeV P protein separated on SDS-PAGE gels as a well-defined protein band (Fig. 1) that was excised from gels or from membranes after electrotransfer. Direct N-terminal amino acid sequencing of P protein immobilized on a polyvinylidene difluoride membrane gave no sequence data, indicating that the N terminus of the protein was chemically blocked. Similar results were obtained in experiments with aldehyde-free acetic acid used in staining procedures, indicating that the N terminus of the P protein may be modified. Trypsin and Glu-C endoproteinase were used to generate internal peptides from protein samples excised from polyacrylamide gels. As summarized in Table 2, the 71% of the internal amino acid sequence obtained from these peptides was found to be identical with that derived from DNA analysis. No sequence information could be obtained for the blocked N-terminal peptide (aa 1 to 22) or for three long peptides (aa 459 to 530, 540 to 553, and 558 to 587) associated with the protein’s hydrophobic core. These peptides were not identified in proteolytic digests, as they were possibly not eluted from the reverse-phase column due to their hydrophobic natures.

TABLE 2.

Amino acid sequences of high-performance liquid chromatography (HPLC)-purified peptides obtained from in situ proteolytic digests of HeV P protein^a

Digest	HPLC peak no.	Amino acid sequence	Sequence positions
Tryptic	1	EIQK	23–26
	2	TYGR	27–30
	3	YVR	428–430
	4	ELR	674–676
	5	THIK	668–671
	6	GGPVK	319–323
	7	THIKDR	668–673
	8	AVPDTK	150–155
	9	NFR	614–616
	10	VLAK	554–557
	11	SSIQQPSTK	31–39
	12	GLPPNQESK	421–439
	13	EGSHPGGSLR	348–357
	14	GLPPNQESK	431–439
	15	LREPPQSSGNR	358–369
	16	EGSHPGGSLR	348–357
	17	TMIR	664–667
	18	VLPNAPK	156–162
	19	TTVPEEVR	163–170
	20	SQYVGTEDVPGSK	410–422
	21	LNHIEEQMK	531–539
	22	TGDAASPGGVQR	378–389
	23	DGSLTDEPYGGVAR	617–630
	24	GWSYHMSGTHDGNVR	135–149
	25	GNVNLDSIK	228–236
	26	TNPELKPVIGR	588–598
	27	NVELDSSVTSSDGTIG	74–89
	28	SVTAENVQLSAPSAVTR	440–456
	29	IYTSDDEDENQLEYEDEF	237–254
	30	NQSTPTEEPPVIPEYYYGSGR	197–217
	31	SELMDYLRN	677–685
	32	AWEDFLQSTSGEHEQAEGGM	44–63
	33	SSSEVVIDTTPEDND	257–271
	34	AWAEDPDDIQLDPMVTDVVYHDH	97–119
	35	NILEQQELFSFDNLK	599–613
	36	RLPMLSEEFECSGSDDP	324–340
	37	IRDDLILPELNFSETNASQFVPLADDASK	631–659
	38	EIDLIGLEDK	171–180
	39	AETDEEVQEVANTVNDI	686–702
	40	SELMDYLNRAETDEEVQE	677–694
Glu-C	1	SKSVTA	438–443
	2	CSGSDDPIIQ	334–343
	3	DFLQSTSGEHEQA	47–59
	4	IQKTYGRSSIQQPSTKD	24–40
	5	YYYGSGRRGDLSKSPPRG	211–228
	6	NVQLSAPSAVTRN	445–457
	7	TKVLPNAPKTTVPEEVR	154–170
	8	RGWSYHMSGT	134–143
	9	DKFASAGLNP	179–188
	10	LFSFDNLKNFRDGSLTDE	606–623

^aSequence information was obtained from multiple peptides within a peak. 

Multiple-sequence alignments and phylogenetic analysis of P/V/C gene-encoded proteins.

An alignment of the P/V/C gene-encoded proteins of HeV by using the ALIGN Plus, SAM, and CLUSTAL W programs indicated that HeV diverges significantly from other members of the Paramyxoviridae family. Rubulaviruses in particular are only distantly related, and even with the more closely related paramyxoviruses and morbilliviruses, the HeV P protein displays amino acid sequence identities of only 11 to 14 and 12 to 16%, respectively (Table 3). Low levels of homology are also characteristic of the HeV C protein, with amino acid sequence identities of 7 to 13 and 10 to 13% to C proteins from paramyxoviruses and morbilliviruses, respectively. These figures are substantially lower than those observed for P and C proteins of viruses within either genus. The amino acid sequence identities for P and C proteins in morbilliviruses are 44 to 48 and 37 to 42%, respectively. The corresponding figures for Paramyxovirus genus members are 23 to 53 and 36 to 69% (Table 3). In contrast to the P and C proteins, the HeV V-specific region shows 30 and 52% homologies with paramyxovirus and morbillivirus V-specific proteins, respectively. The latter figure lies within the range of 50 to 71% exhibited by morbilliviruses (Table 3). No significant sequence homology was detected between sequences in protein data banks and the putative small basic protein.

TABLE 3.

Amino acid sequence identities of P-gene encoded proteins between HeV and selected paramyxoviruses and morbilliviruses

Virus	Identity (%)a
	P protein						C protein						V-specific protein
	SeV (568)	hPIV3 (603)	hPIV1 (568)	MeV (507)	CDV (507)	DMV (506)	SeV (204)	hPIV3 (199)	hPIV1 (204)	MeV (186)	CDV (174)	DMV (160)	SeV (71)	hPIV3 (69)	hPIV1b	MeV (72)	CDV (72)	DMV (76)
HeVc	14	14	11	14	12	16	13	7	11	11	10	13	46	30		50	43	52
SeV		25	53	13	13	13		36	69	11	14	9		34		32	28	38
hPIV3			23	13	14	14			37	10	8	11				21	24	21
hPIV1				11	13	12				8	13	8
MeV					44	48					42	38					50	71
CDV						47						37						56
DMV

^aParenthetical data are the numbers of amino acid residues in the indicated proteins for the viruses listed. 

^bhPIV1 does not produce a V protein. 

^cThe HeV P, C, and V-specific proteins consist of 707, 166, and 54 aa, respectively. 

ALIGN Plus, SAM, and CLUSTAL W programs were consistent in revealing that the 11 to 16% homology observed between HeV and Paramyxovirinae P proteins occurs predominantly in the C-terminal domain, in the region between the insertion site and the carboxy terminus. An alignment of the P proteins of HeV, MeV, canine distemper virus (CDV), and dolphin morbillivirus (DMV) by CLUSTAL W confirmed the strong homology between the nontemplated-nucleotide insertion site (AAAAAGGG) of the P/V/C gene of HeV and morbilliviruses (12) and revealed sequence identity for only 16 aa in the region from the amino terminus of the P protein to the insertion site (approximately 4% of HeV amino acids), compared with 49 aa in the region from the insertion site to the carboxy terminus of the protein (approximately 16% of HeV amino acids) (Fig. 3A). The sequence of a stretch of 33 aa, starting at aa 549 of the HeV P protein, contains 16 identical and 9 conserved aa (Fig. 3A). An alignment of HeV and paramyxovirus P proteins revealed sequence identity for only 2 aa and conservation for only 5 aa in the same 33-aa region (data not shown). In contrast to morbilliviruses, the sequences of HeV, SeV, hPIV1, and hPIV3 P proteins are identical in only 9 aa between the amino terminus and the insertion site (approximately 2% of HeV amino acids) and 17 aa from the insertion site to the carboxy terminus (approximately 6% of HeV amino acids) (data not shown).

FIG. 3.

Sequence alignment revealing regions of homology between HeV P and V proteins and cognate proteins of selected Paramyxovirinae viruses. (A) Alignment of the P protein from the insertion site to the carboxy terminus. Amino acids encoded by the conserved insertion sequence are indicated by asterisks. Amino acid numbers of the HeV P protein are indicated above the sequence. (B) V protein alignment. Insertions in one protein are indicated by dashes in the sequences of cognate proteins. Amino acid identities and conserved amino acid characteristics are indicated below sequences by asterisks and periods, respectively.

The lack of significant homologies for the amino-terminal half of the HeV P protein to cognate morbillivirus proteins is due in part to the increased length of the HeV P protein. An alignment by CLUSTAL W revealed that an additional 109 aa of HeV P protein located at the amino terminus has no homology to any other Paramyxoviridae P protein. An alignment of the HeV V-specific protein with the V proteins of selected members of the Paramyxovirinae subfamily is shown in Fig. 3B. There is a more random distribution of conserved and identical amino acids, compared with that observed in the P protein.

The phylogenetic relationships among HeV, paramyxoviruses, and morbilliviruses were calculated with the PHYLIP package (10) by distance matrix–neighbor-joining and maximum-parsimony methods. Majority-rule consensus bootstrap trees were generated for the P/V/C gene and its products (Fig. 4); only branches with >50% support were included. The phylogenetic trees shown in Fig. 4 represent unrooted trees with branch lengths not to scale, and no assumptions were made about the evolution of these viruses. All of the trees were essentially identical and showed HeV on branches midway between paramyxoviruses and morbilliviruses. On the basis of these unrooted trees, HeV could not be classified as either a paramyxovirus or a morbillivirus but formed a distinct branch in all phylogenetic trees.

FIG. 4.

Phylogenetic trees for the HeV P/V/C gene (A) and P (B), C (C), and V (D) proteins. Phylogenetic trees were produced from 100 random data sets generated by using the SEQBOOT program from PHYLIP (10) and are shown as majority-rule consensus, radial trees constructed by maximum-parsimony methods (PROTPARS and CONSENSE programs [10]). Only branches with >50% bootstrap support are indicated, and trees are unrooted with branch lengths not drawn to scale.

DISCUSSION

The P/V/C gene of HeV differs from other Paramyxovirinae genes in a number of ways. (i) The gene is much larger. (ii) The P protein is the largest described to date. (iii) HeV P mRNA has a much longer untranslated 3′ tail. (iv) HeV C protein is smaller and less basic than cognate proteins in other Paramyxovirinae viruses. (v) All Paramyxovirinae viruses express at least two ORFs from their P/V/C genes. Only the P protein is expressed in all of the viruses studied; the C protein is absent in rubulaviruses, and some parainfluenza viruses do not produce the V protein. However, in addition to P, C, and V proteins, the P/V/C gene in HeV has the coding capacity for a putative SB protein.

The recognition that more than one protein is generated from a single mRNA in Paramyxovirinae viruses was an early example of multiple translational initiation, which led to modification of the scanning model for ribosomal initiation (20). In the modified model, scanning ribosomes can occasionally bypass 5′-proximal ATGs in an unfavorable context and initiate protein synthesis at downstream ATGs. According to Kozak, the optimal context for translation initiation has the consensus sequence GCCA(G)CCATGG, in which the A (sometimes G) residue at −3 and the G residue at +4 play the most important roles in modulating initiation efficiency. In the HeV P/V/C gene, we have found TGACAAATGG for ORF-P, TCAATGATGG for ORF-C, and ATTACTATGG for ORF-SB (Fig. 2). Since all three sequences have a G residue at +4, their initiation efficiencies are most likely determined by the residue at −3. Both ORF-C and ORF-SB have the optimal residue, A, at this position, whereas ORF-P has a nonpurine residue, C, making the ORF-P initiation site the weakest. This observation is consistent with the hypothesis that ORF-P contains a leaky initiation site for translation. For ORF-C, there are two tandem ATG codons at the beginning of the ORF; the second ATG was used in this comparison since it gives a better initiation context. ORF-C and ORF-SB have identical residues at −3 and +4. In addition, ORF-SB has a C residue at −2, whereas ORF-C has a C residue at −5. It is therefore difficult to rank the initiation environments for these two ORFs.

In relation to coding capacity for multiple proteins in different reading frames of the HeV P/V/C gene, it is interesting that the protein encoded in the +1 reading frame is very acidic (the P protein), those encoded in the +2 reading frame are basic (C and SB proteins), and the protein encoded in the +3 reading frame is close to neutral (V-specific protein).

Within the P/V/C genes of Paramyxovirinae viruses, the C-terminal domain of the P protein, i.e., the region after the mRNA editing site, is the most conserved. It is essential for genome replication (4), oligomerization of the P protein (14, 23), binding to the L protein (25), and association with the N protein both in nucleocapsids (36) and in unassembled N-P complexes (5). For most Paramyxovirinae viruses, this is also the V-specific coding region. In the case of the HeV P protein, the domain after the mRNA editing site also represents the region which displays the best, albeit low, homologies with the P proteins of morbilliviruses and paramyxoviruses. It is interesting that although the overall size of HeV P protein is much larger than its counterparts in other Paramyxovirinae viruses, this C-terminal domain (304 aa) is only slightly longer than those in other P proteins (253 to 277 aa).

The presence of an ORF coding for an SB protein between ORF-C and ORF-V is a unique feature of the HeV P/V/C gene. A similar SB protein, encoded in an overlapping reading frame within the P gene, has previously been identified in both the New Jersey and Indiana serotypes of VSV (21, 40). The expression of VSV SB in infected cells has previously been experimently demonstrated, and the protein does not appear to be associated with virions. Sequence analysis indicated that the SB protein is conserved in the Vesiculovirus genus of the Rhabdoviridae family but not in the Lyssavirus genus. It is interesting that both VSV and HeV SB proteins are 65 aa in length and contain eight arginine residues, which contribute to the high pIs of 11.0 and 11.9 for VSV SB and HeV SB, respectively. A similar SB-encoding ORF has also previously been found in the VP35 gene, the equivalent of the P gene, of Marburg virus in the Filoviridae family (9). If the HeV and Marburg virus SB proteins are synthesized in vivo, SB proteins may represent a class of molecules which are shared among the three families within the Mononegavirales order. Although SB proteins are not encoded by all viruses within a family and there seems to be no conservation of primary sequence, it is interesting that these P-overlapping molecules in members of the Mononegavirales order have very similar molecular sizes and characteristics. Spiropoulou and Nichol (40) used the letter C to designate the SB protein synthesized by VSV because its location at the N terminus of ORF-P and its overall basic nature resemble that of the C protein of a Paramyxovirinae virus. However, we believe that it may be more appropriate to name them SB proteins for two reasons. First, these proteins are much smaller, approximately one-third of the size of C proteins identified in Paramyxovirinae viruses. Second, we have identified in HeV not only a C protein but also a coding region that is potentially capable of generating an SB protein which is equivalent in size and with characteristics similar to the SB proteins identified in vesiculoviruses and Marburg virus.

The P genes of almost all of the viruses in the order Mononegavirales have the capacity to code for more than one protein either by multiple translational initiation or mRNA editing, but proteins such as V and C are not found in all viruses and may therefore be described as accessory proteins. Among the various proteins expressed from the P/V/C gene, the P protein is the only one known to be essential for virus replication and synthesized by all viruses in the family. The function of the C protein is not clear, and this protein is not present in rubulaviruses. The V protein is interesting for two reasons. First, it is cysteine rich, contains a zinc finger, and has previously been shown to bind zinc (24, 32). Second, V proteins are the most conserved proteins encoded by the P/V/C gene, arguing for their importance in virus evolution. Until recently, the V protein was regarded as nonessential, at least in tissue culture cells (7, 37). In 1997, Kato et al. (18, 19) demonstrated that SeV V protein codes for as a function that is required for viral replication and expression of pathogenesis in vivo. Whether the in vivo function of the V protein is unique to SeV remains to be seen because the closely related virus hPIV1 lacks the capacity to code for a V protein (26, 34) and hPIV3 does not seem to express a V protein, although the coding sequence is present in the P/V/C gene (11). In addition to V and C proteins, HeV may code for a third accessory protein, SB. It is tempting to speculate that the large P gene of HeV and its capacity to code for up to three unrelated, accessory proteins may be responsible, at least partially, for the ability of the virus to replicate in species as diverse as horses, humans, cats, rabbits, and fruit bats (49, 51) and in a wide range of cultured cells (29). HeV is pathogenic in humans, horses, cats, and guinea pigs (50).

The unusually large size of the HeV P protein, the nonconserved nature of its N-terminal domain, and the fact that the HeV P/V/C gene and its products have sequences that diverge significantly from those of other members of the Paramyxoviridae make the construction of multiple-sequence alignments difficult to interpret. Attempts were made to overcome this by (i) validating the phylogenetic groupings with bootstrap resampling to give confidence limits to the trees and (ii) using two methods of phylogenetic analysis, distance matrix–neighbor-joining and maximum-parsimony methods. The trees produced were consistent with those generated for the more conserved M protein (12) and gave bootstrap values of >50%. The results show clearly that HeV cannot be classified as either a paramyxovirus or a morbillivirus; it forms a distinct branch in all phylogenetic trees, suggesting that HeV should be classified in a new genus within the Paramyxovirinae subfamily (30).