Full-length genome analysis of natural isolates of vesicular stomatitis virus (Indiana 1 serotype) from North, Central and South America

Genomic termini

We sequenced the genomic termini directly from RT–PCR products using L forward and N reverse primers on purified virion RNA ligated with T4 RNA ligase. The terminal 50 nucleotides at the 3′ terminus were highly conserved among all the viruses with only one nucleotide change (G→A) observed at position 25 in sequence J02428 (Fig. 2⇓). The 5′ terminus was less conserved, with two nucleotide substitutions (A→C) at positions 20 and 26 from the 5′ terminus of 94GUB and 98COE, respectively (Fig. 2⇑). A high degree of sequence conservation was expected for the virus termini in the natural isolates, since these regions contain signals necessary for initiation of transcription and also for replication of the viral RNA (Pattnaik et al., 1995⇓ ; Whelan & Wertz, 1999⇓ ).

Non-coding regions and gene junctions

All four non-coding gene junctions in all the viruses studied contained a 23 nucleotide conserved sequence: 3′ AUAC(U)₇NNUUGUCNNUAG 5′. These sequences have been the subject of intense investigation since they contain cis-acting signals for transcription stop, polyadenylation and transcription start of each mRNA. As expected, the sequence 3′ AUAC(U)₇ 5′ was conserved in all viruses since it is critical for both termination and polyadenylation of mRNA, and in vitro studies have shown very little tolerance to mutations at these sites (Whelan & Wertz, 1999⇓ ).

The (U)₇ polyadenylation signal was followed by two non-transcribed nucleotides. These non-transcribed intergenic nucleotides are an important part of the cis-acting signals involved in termination and reinitiation of transcription in VSV (Barr et al., 1997a⇓ ; Stillman & Whitt, 1997⇓ ). The non-transcribed intergenic nucleotides 3′ GA 5′ and 3′ CA 5′ between N and P and between P and M, respectively, were conserved among all viruses sequenced (Table 1⇓). In contrast, the M–G and G–L intergenic nucleotides were more permissive to changes in the natural isolates from Central and South America, with each virus lineage having different intergenic nucleotides at the M–G junction, including the trinucleotide 3′ GGA 5′ in the South American lineage and either 3′ GG 5′or 3′ GA 5′ in the Central American lineage (Table 1⇑). We determined the sequence of the intergenic nucleotides between M and G and between G and L for five additional viruses from each genetic lineage and found them to be identical to the representative full-length sequence within the North American or South American lineages. However, the G–L junction was variable in the Central American lineage with three different intergenic nucleotides found among the five viruses sequenced (Table 1⇑).

The sequence variability observed at the gene junctions was unexpected, given the importance of these nucleotides in regulating transcription. However, each of the intergenic nucleotide sequences, except for the trinucleotide observed in the South American lineage, had been previously tested in vitro by Barr et al. (1997b⇓ ) and found to efficiently stop readthrough and attenuate transcription of the downstream mRNA. Our finding of the naturally occurring functional intergenic trinucleotide 3′ GGA 5′ increases the known permissive range of intergenic nucleotides.

Nucleocapsid

N is the most abundant viral protein in infected cells. It binds both genomic and replicative intermediate viral RNA and, together with the L and P proteins, is an indispensable component of the transcription–replication complex. We found that N, which is 422 amino acids long, was the most conserved protein among all the viruses (Fig. 3⇓). Only five amino acid substitutions (positions 33, 259 and 360 in 94GUB and positions 11 and 110 in 85CLB) were found among all three viruses sequenced. None of these changes were located within the last 60 amino acids of the C-terminal end of N, where sequences required for interaction with the phosphoprotein and also for encapsidation are believed to be located (Pattnaik et al., 1995⇓ ). Its high conservation, coupled with the fact that N is the most abundant transcript in infected cells, makes it an ideal target for diagnostic probes.

Phosphoprotein

Three major domains have been described in P, which play an essential role in transcription and replication. Domain I (aa 1–137) is responsible for the association of P with L and needs to be phosphorylated in several sites for optimal transcription activity (Takacs et al., 1992⇓ ; Pattnaik et al., 1997⇓ ). Domain II (aa 229–242) binds to L and contains two seemingly essential serines, whose phosphorylation by L regulates transcription. Domain III (aa 243–265) is positively charged and is located at the C terminus (Das et al., 1997⇓ ). Although its length was 265 amino acids in all the viruses studied, P was the most variable of the five structural proteins (Fig. 3⇑). Of a total of 33 amino acid changes, 21 occurred within functional domain I. Six of these changes involved threonines or serines, resulting in a net loss of one phosphorylation site in 94GUB. Domain II was conserved among all viruses, with only one amino acid change at position 237. All basic amino acids in domain III, which are essential for transcription and binding to L, were conserved among all viruses, with only one substitution observed at position 261. Despite the fact that P was the most divergent major viral structural protein (range of divergence 3·4–8·7%), the amino acid changes did not seem to involve any of the critical positions in the functional domains.

C proteins

Two small basic proteins, C′ and C (67 and 55 amino acids, respectively), are encoded in a second open reading frame (ORF) within the P gene of VSV-IN1 (Spiropoulou & Nichol, 1993⇓ ; Kretzschmar et al., 1996⇓ ). The role of these proteins in the VSV life cycle remains unknown, since viruses in which C protein expression has been abrogated grow in vitro similarly to the wild-type strain (Kretzschmar et al., 1996⇓ ). All viruses had the second ORF with conserved start codons for both C’ and C. In all cases, the AUG of the smaller C protein was in better context for translation, indicating that it is preferentially translated (Spiropoulou & Nichol, 1993⇓ ; Kozak, 1986⇓ ). The C protein was the most divergent among all the VSV proteins (followed by P, which was the most divergent among the structural proteins) (Fig. 3⇑). However, the predicted isoelectric point, dictated by the number of arginines in each protein, remained between 10·86 and 11·28 in all viruses. Interestingly, C was the only VSV gene product with a substantially higher number of non-synonymous versus synonymous substitutions (Fig. 3⇑). This indicates that C is under different selective pressures than P, despite the fact that both proteins are encoded by the same mRNA.

Matrix protein

M is an important component in the virus structure, playing a major role in virus morphology, assembly and budding. It is also involved in inducing host-cell cytopathic effect, inhibition of host gene expression, nuclear transport and apoptosis (Kopecky et al., 2001⇓ ; Desforges et al., 2001⇓ ; Petersen et al., 2000⇓ ; Wagner & Rose, 1996⇓ ). M, which is 229 amino acids in length, was very conserved at the amino acid level, with only seven differences among all the viruses (positions 54, 133 and 225 in virus J02428; 70 and 221 in 98COE; and 126 and 171 in 94GUB) (Fig. 3⇑). At position 24–27, all viruses had a PPPY motif similar to the late domains in retroviral Gag proteins, which seems to be required for the efficient release of viral particles (Jayakar et al., 2000⇓ ).

Glycoprotein

G has several functions, including fusion with the cell membrane during virus entry and membrane budding during exit from the cell. It is also the target of neutralizing antibodies and cell-mediated immune responses to VSV (Wagner & Rose, 1996⇓ ). Trimers of G form the spikes protruding from the viral envelope, which interact with cellular receptors. Functional domains identified in G include glycosylation and palmitylation sites, a membrane anchor, a cytoplasmic domain and membrane fusion domains (Coll, 1995⇓ ).

After P, G was the second least conserved of the major viral proteins, with 95·5–97·5 amino acid conservation (Fig. 3⇑). There were a total of 41 amino acid differences among the four viruses; 16 of these occurred within 30 amino acids of either the N terminus in the ectodomain, or the C terminus in the cytoplasmic domain. No changes occurred at or near glycosylation or palmitylation sites. Two amino acid changes (position 258 in 98COE and 259 in 94GUB) were observed at or near epitopes where neutralization-resistant mutations have been selected in vitro (Luo et al., 1988⇓ ). This could suggest that there is immune selection among these viruses under natural conditions. However, no accumulation of changes in these areas of the glycoprotein was observed among viruses from Central and North America previously sequenced (Rodriguez et al., 2000⇓ ). Two domains involved in fusion, one at amino acid position 118–139 (Shokralla et al., 1998⇓ ) and the other at position 395–462 (Fredericksen & Whitt, 1998⇓ ) were completely conserved among the four viruses compared.

Insertion in the genomic 5′ non-coding region of the G gene

The 5′ (genomic sense) non-coding region of the glycoprotein gene is highly variable both in length and sequence composition, among the genetic lineages of VSV. A reiterative 175 nucleotide insertion of 3′ UUAAAAA 5′ was found between the stop codon of G and the G–L gene junction of 94GUB. We sequenced the G gene 5′ non-coding regions of nine other VSV-IN1 isolates from throughout Central America, and found inserts of variable length only in viruses within one genetic lineage from northern Central America (Guatemala, Honduras, El Salvador) and not in viruses from southern Central America, North or South America (Fig. 1⇑). Insertions had been previously noticed in the VSV-IN1 G mRNA by Bilsel & Nichol (1990)⇓ . Until now, it was not clear whether this insert was added by the viral polymerase by stuttering during transcription of the mRNA, or if it was present in the viral genome. In order to clarify this, we used gradient-purified virion RNA as template for RT–PCR of this region and confirmed the presence of the insert in the genome. How this insert arose and was maintained in this lineage is not certain. Our data supports the idea that it was created by stuttering of the viral polymerase on the sequence 5′ AAUUUUU 3′ near the U₇ track in the G–L junction of the positive-sense RNA template during virus replication, as first proposed by Bilsel & Nichol (1990)⇓ . The polymerase stuttering event might have been favoured by the presence of the polyadenylation signal 3′ AUAC 5′, which occurs prior to the 5′ AAUUUUU 3′ only in the northern Central American lineage. Once this insert arose in an ancestral virus, it could have become fixed due to the lack of a recombination mechanism in Mononegavirales (Pringle, 1981).

Since the length of the 5′ non-coding sequence in the G gene was variable among the viruses studied, we wanted to determine its diversity within the quasispecies of a single virus strain. In order to test this, we cloned the resulting RT–PCR product of the G–L region of 94GUB, from the stop codon of G to the start codon of L, into the TOPO vector and selected ten individual colonies for sequencing. Four of the colonies had a 1 nucleotide deletion at position 14, four had a 1 nucleotide deletion at position 55 and one colony had a 30 nucleotide deletion at position 142–171 (Fig. 4⇓). Despite this variability, we found the same consensus 175 nucleotide sequence in at least three independent 94GUB RNA preparations when RT–PCR products were directly sequenced. Interestingly, the G–L region is also the site where other rhabdoviruses such as snakehead rhabdovirus and rabies virus (Kurath et al., 1997⇓ ; Johnson et al., 2000⇓ ; Ravkov et al., 1995⇓ ) have additional genes. It seems that the G–L junction of rhabdoviruses in general is very permissive to insertions and deletions and perhaps is the site of extinct genes in VSV.

L protein

The VSV polymerase (L) has several functions when associated with P and N including: ATPase, methylase, guanylyl transferase, capping enzyme and poly (A) polymerase among others (Feldhaus & Lesnaw, 1988⇓ ). This protein was 2109 amino acids in length, with high conservation (97·2–98·6%) among the four viruses (Fig. 3⇑). However, the amino acid substitutions observed were not randomly distributed, but rather were concentrated in several discrete regions, particularly at the N- and C-terminal ends.

The exact domains within L associated with each of its functions are not clear but conserved motifs have been described in at least six areas of conservation among the Mononegavirales and these have been proposed as putative active sites (Poch et al., 1990⇓ ). A sliding window analysis of the L amino acid alignment showed that, for the most part, these areas were of high amino acid conservation among the four virus sequences (Fig. 5⇓). The polymerase motif QGDNQ was found at position 712–716 in all viruses in an area of low divergence within domain III (Fig. 5⇑). The putative template recognition sequences within domain II at amino acids 530–660 were also completely conserved. Since these L sequences are the first from field strains without passage in tissue culture, they could be useful in determining the functional domains of this protein.

The purpose of this work was to determine and analyse the full-length genomic sequences of natural isolates, each representing a distinct genetic lineage of VSV-IN1, and to compare these sequences with the only full-length sequence of VSV-IN1 deposited in GenBank. The results showed that overall structure and organization were similar among all the viruses, except for one genetic lineage from northern Central America that had an insertion of variable length at the 5′ non-coding region of the G gene. Protein comparisons showed that N was the most conserved protein, making it a good target for viral detection in diagnostic tests. Based on this finding, we are currently developing detection assays using real-time RT–PCR. Non-coding sequences involved in regulation of transcription and gene expression were completely conserved, but intergenic nucleotides varied in sequence and length in the M–G and G–L gene junctions. These full-length sequences could serve as a baseline for studies on genomic function using reverse genetics.