As a result of a disease outbreak in chickens in Chile (18), 13 hemagglutinating isolates (number 176822 obtained in May 2002 and numbers 4322, 4325, 4345, 4346, 4347, 4348, 4418, 4458, 4957, 4966, 4968, and 4977 obtained in June 2002) were submitted to Office of International Epizooties (OIE) reference laboratories for characterization. All viruses were propagated in the allantoic and amniotic cavities of 10- to 11-day-old embryonated specific pathogen free (SPF) chicken eggs for 48 to 72 h at 35°C to 37°C. Viruses submitted as isolates from Chile and other sources were passaged once in SPF eggs after receipt.

Subtype identification was done by hemagglutination-inhibition (HI) and neuraminidase-inhibition (NI) tests by using polyclonal chicken antisera against a panel of influenza A reference strains, which had been prepared in SPF chickens. Virulence was assessed by the standard intravenous pathogenicity (IVPI) test by using 6-week-old SPF chickens and the standard pathotyping test by using 4- to 6-week-old SPF chickens (19,20). Both tests give a standard volume of virus intravenously to SPF chickens, but the IVPI test characterizes both illness and how many days the chickens remain alive to produce an index of virulence from 0 to 3. An IVPI index of >1.2 or death in chickens of >75% in the standard pathotyping test indicates a highly pathogenic avian influenza virus. Representative low and highly pathogenic viruses were also tested for their ability to replicate in chicken embryo fibroblast cells with or without trypsin (0.25 μg/mL) in the cell culture media.

RNA from the isolates examined in this study was extracted with either the Trizol LS reagent (Invitrogen, Carlsbad, CA), the RNeasy mini kit (Qiagen, Valencia, CA), or the QIAmp Viral RNA mini kit (Qiagen) from infectious allantoic fluid from embryonated chicken eggs before reverse transcriptase–polymerase chain reaction (RT-PCR) amplification. The RT-PCR amplification was performed with either a one-step or a two-step RT-PCR reaction. The one-step reaction used the Onestep RT-PCR kit (Qiagen) with incubation steps of 45°C for the PA, PB1, and PB2 genes and 50°C for the other genes for 30 min, and 95°C for 15 min and PCR incubation steps of 30 cycles of 53°C annealing for the PA, PB1, and PB2 genes and 56°C annealing for the HA, NA, M, NP genes for 15 s, 72°C extension for 60 s, and 94°C denaturation for 30 s. For the amplification of the nonstructural and matrix gene segments, primers in the 5′ and 3′ noncoding region of the RNA segment were used to amplify the complete coding sequence to be used for direct sequencing. The N3 gene segment was RT-PCR amplified in two parts and used for direct sequencing. Electrophoreses was performed on HA, NP, PB1, PB2, and PA PCR product, amplified with specific primers from the noncoding sequence, in a 1% agarose gel, and the products were extracted with the Qiaquick gel extraction kit (Qiagen) and cloned by using the pAMP ligation independent cloning system (Invitrogen). Colonies were screened by using PCR with internal primers; positive cultures were grown overnight, and plasmid was extracted using the Qiaprep spin miniprep kit (Qiagen). Alternatively, a two-step RT-PCR amplification with the Vgen primer (5′ AGCAAAAGCAGG) with MMLV reverse transcriptase (Promega, Madison, WI) and PCR with gene-specific primers was performed. For both direct PCR sequencing and plasmid sequencing, the ABI PRISM Bigdye terminator sequencing kit (Perkin Elmer, Foster City, CA) was used, and the reactions were run on ABI 3700 or 310 automated sequencers (Perkin Elmer).

The sequencing information was compiled with the Seqman II program (DNASTAR, Madison, WI), and nucleotide sequences were aligned with sequences from the influenza sequence database with the Megalign program (DNASTAR, Madison, WI) by using the Clustal V alignment algorithm. Pairwise sequence alignments were also performed in the Megalign program to determine sequence similarity between A/chicken/Chile/176822/02 and other published sequences for each gene segment. The origin of a 30-nt insertion at the HA cleavage site was determined by using the best-local-homology rapid search procedure (BLAST) against GenBank sequences. Phylogenetic comparisons of the aligned sequence for each gene segment were generated by using either the maximum parsimony method with 100 bootstrap replicates in a heuristic search with the PAUP 4.0b10 software (Sinauer Associates, Inc, Sunderland, MA) or with the maximum likelihood by using the PHYLIP phylogenetic inference package, version 3.57c (21) with transition/transversion ratios calculated by PUZZLE (22). Sequence data was submitted to GenBank with accession no. AY303630-AY303666.

All 13 hemagglutinating isolates were identified as influenza A viruses of the H7N3 subtype. Six of the 7 isolates, 4322, 4325, 4418, 4957, 4968, and 4977, tested by the IVPI test were shown to be HPAI viruses with indices between 2.43 and 3.00. An IVPI of >1.2 is classified as a highly pathogenic avian influenza virus. Isolate 176822 was characterized as LPAI with an IVPI index of 0.00. Similar results were observed in the standard pathotyping test. The H7N3 virus was reisolated from chickens that died after intravenous injection with isolates 4322 and 4975. The amino acid sequence of the HA cleavage site was determined and shown to be the same as the injected virus. The low pathogenic and two highly pathogenic Chilean viruses, 176822, 4322, and 4957, were grown in chicken embryo fibroblast cell culture with and without the addition of trypsin to the media. The low pathogenic virus, 176822, did not plaque without the addition of trypsin. The HPAI viruses, which included a representative with and without the additional lysine at the cleavage site (see HA0 sequence below), plaqued with and without the addition of trypsin.

The deduced amino acid sequence at the hemagglutinin cleavage site for the H7N3 virus of low pathogenicity isolated in May 2002, 176822, was PEKPKTR/GLF. All the HPAI viruses had a 10-aa insert (basic amino acids are underlined) at the HA cleavage site, but the insert varied by a single nucleotide between some of the isolates (PEKPKTCSPLSRCRETR*GLF [isolates 4322, 4418 and 4977] and PEKPKTCSPLSRCRKTR*GLF [isolates 4325, 4957 and 4968] that resulted in an aa change from glutamic acid to lysine.

The insertion at the HA cleavage site is unlike any previously reported, and a BLAST search of this nucleotide sequence showed the most closely related sequence in the GenBank database was the nucleoprotein (NP) gene of A/gull/Maryland/704/77 at position 1268–1297, with 28 of 30 nucleotide identities. Nucleotide sequencing of the NP gene of the HPAI Chilean viruses 4077, 4346, 4957 and the LPAI virus 176822 indicated a 100% nucleotide sequence identity of the HA cleavage site insertion with the deduced amino acid sequence CSPLSRCRET. Three other conservative amino acid changes in the hemagglutinin were different between the LPAI and all the HPAI viral sequences including at position 38 A>T, position 141 D>N, and position 146 A>T. None of these changes involved potential N-linked glycosylation sites, and no additional potential N-linked glycosylation sites were observed in the Chilean viruses.

The complete coding sequence for all eight gene segments from the low pathogenic isolate (A/chicken/Chile/176822/02) and one of the highly pathogenic isolates (A/chicken/Chile/4957/02) was determined. Additional genes from other isolates were also sequenced and used for comparison. The low pathogenic isolate, A/chicken/Chile/176822/02, was chosen as the reference isolate for comparison with the influenza database, and pairwise sequence analyses were performed to identify the most closely related isolates for each gene segment. The closest nucleotide sequence similarities ranged from 82.1% for the HA gene to 96.3% for the matrix gene. The amino acid sequence similarity was much higher for all gene segments (Table). Phylogenetic trees were constructed from nucleotide sequences for all eight gene segments, and in general the Chilean isolates clustered most closely with North American avian isolates. Although the most closely related gene segments to the PB2 and PB1 gene segments were viruses isolated from swine, in both cases the swine isolates were thought to be the result of a recently introduced or reassorted avian influenza virus in North America (23,24). Exceptions were the nucleoprotein (Figure 1) and polymerase acid protein (PA) genes, which were most closely related to H7N7 equine influenza viruses which had internal genes derived from the equine type 2 H3N8 viral lineage.

For all gene segments, the Chilean viruses formed a distinct subgroup from other influenza viruses. Analyses of the HA gene indicated that all the Chilean H7N3 viruses were closely related and that the HPAI viruses emerged from the LPAI virus. The H7 phylogenetic tree, when using Equine/Prague/1/56 as the outgroup, shows the avian isolates divided into two main branches that are further subdivided into geographically defined groups (Figure 2). One branch includes the North American avian and the Chilean viruses, and the second branch includes the Eurasian avian and Australian avian viruses. Although the Chilean viruses are most like avian viruses of North American origin they are distinctly different from these viruses and form a unique branch on the tree.

For the N3 tree little sequence data were available to make meaningful observations about the phylogeny. In conserved internal proteins like the matrix and nucleoprotein genes, differences in North American avian and Eurasian avian influenza viruses can be observed at the nucleotide level, but almost all tree topology structure is lost when comparing the same isolates in phylogenetic trees based on amino acid sequence. This observation is common for avian influenza viruses which often have high sequence conservation at the amino acid level, but with large differences at the nucleotide level (25). The results of nucleotide analyses are presented for PB2 and NS genes as unrooted phylograms (Figure 3). Of the genes examined for the six different isolates, little sequence difference was observed between the low pathogenic and highly pathogenic viruses except for the H7 gene segment where an insertion of 30 nucleotides was present at the HA cleavage site.