Genome Analysis Linking Recent European and African Influenza (H5N1) Viruses

Steven L. Salzberg,* Comments to Author Carl Kingsford,* Giovanni Cattoli,† David J. Spiro,‡ Daniel A. Janies,§ Mona Mehrez Aly,¶ Ian H. Brown,# Emmanuel Couacy-Hymann,** Gian Mario De Mia,†† Do Huu Dung,‡‡ Annalisa Guercio,§§ Tony Joannis,¶¶ Ali Safar Maken Ali,## Azizullah Osmani,*** Iolanda Padalino,††† Magdi D. Saad,‡‡‡ Vladimir Savic',§§§ Naomi A. Sengamalay,‡ Samuel Yingst,‡‡‡ Jennifer Zaborsky,‡ Olga Zorman-Rojs,¶¶¶ Elodie Ghedin,### and Ilaria Capua† *University of Maryland Center for Bioinformatics and Computational Biology, College Park, Maryland, USA; †Istituto Zooprofilattico Sperimentale delle Venezie, Padova, Italy; ‡The Institute for Genomic Research, Rockville, Maryland, USA; §Ohio State University, Columbus, Ohio, USA; ¶Animal Health Research Institute, Giza, Egypt; #Veterinary Laboratories Agency, Addlestone, England, UK; **Central Laboratory of Animal Pathology, Bingerville, Côte d'Ivoire; ††Istituto Zooprofilattico Sperimentale dell'Umbria e delle Marche, Perugia, Italy; ‡‡Department of Animal Health, Hanoi, Vietnam; §§Istituto Zooprofilattico Sperimentale della Sicilia, Palermo, Italy; ¶¶National Veterinary Research Institute, Vom. Plateau State, Nigeria; ##Food and Agriculture Office of the United Nations, Tehran, Iran; ***Ministry of Agriculture, Animal Husbandry and Food, Kabul, Afghanistan; †††Istituto Zooprofilattico Sperimentale della Puglia e Basilicata, Foggia, Italy; ‡‡‡US Naval Medical Research Unit No. 3, Cairo, Egypt; §§§Croatian Veterinary Institute, Zagreb, Croatia; ¶¶¶University of Ljubljana, Ljubljana, Slovenia; and ###University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
Abstract

To better understand the ecology and epidemiology of the highly pathogenic avian influenza virus in its transcontinental spread, we sequenced and analyzed the complete genomes of 36 recent influenza A (H5N1) viruses collected from birds in Europe, northern Africa, and southeastern Asia. These sequences, among the first complete genomes of influenza (H5N1) viruses outside Asia, clearly depict the lineages now infecting wild and domestic birds in Europe and Africa and show the relationships among these isolates and other strains affecting both birds and humans. The isolates fall into 3 distinct lineages, 1 of which contains all known non-Asian isolates. This new Euro-African lineage, which was the cause of several recent (2006) fatal human infections in Egypt and Iraq, has been introduced at least 3 times into the European-African region and has split into 3 distinct, independently evolving sublineages. One isolate provides evidence that 2 of these sublineages have recently reassorted.

The first cases of human infection with highly pathogenic avian influenza (HPAI) strain H5N1 occurred in Hong Kong in 1997; it was brought under control by massive culling of the chicken population (1,2). An antigenically distinct strain emerged in 2002, in the same location, and has since spread to hundreds of millions of birds (3,4). More alarming has been the growing number of human influenza (H5N1) infections; by September 2006, 251 human cases had been reported, resulting in 148 deaths (2). From late 2005 to early 2006, HPAI (H5N1) was detected for the first time in birds in eastern Europe, the Middle East, and northern Africa, indications that the virus was spreading, possibly aided by wild bird migration. Human cases were reported beginning in January 2006 in Egypt, Iraq, Turkey, Djibouti, and Azerbaijan.
Methods

We sequenced and analyzed the genomes of 36 recent isolates of highly pathogenic influenza (H5N1) viruses collected from Europe, northern Africa, the Middle East, and Asia. We used high-throughput methods described previously (5).
Sample Collection
Samples primarily consisting of pooled trachea and lung tissue, pooled intestines, or tracheal and cloacal swabs collected from dead or moribund animals were processed for attempted virus isolation as described (6). Hemagglutinating isolates were typed either by reverse transcription–PCR (RT-PCR) or by serologic methods (7). RNA was extracted with the High Pure Extraction Kit (Roche, Indianapolis, IN, USA), according to manufacturer's instructions.
Primer Design

Sequences from recent human and avian influenza (H5N1) isolates were downloaded from GenBank and were aligned with MUSCLE (8). Degenerate primers were designed on the basis of consensus sequences generated with BioEdit (9). An M13 sequence tag was added to the 5´ end of each primer to be used for sequencing. Four of the reactions were analyzed by electrophoresis on an agarose gel for quality control purposes. Primer design was optimized by analysis of the sequence success rate of each primer pair. Primers that did not perform well were redesigned and replaced in the primer set. Primers were designed to produce ?500-nt overlapping amplicons to provide 2× coverage of each genomic segment. Additionally, a second set of primers was designed to produce 500-nt amplicons offset ?250 nt from the original primer pair, which gave at least 4× sequence coverage of each segment.
cDNA Synthesis

Amplicons tiling the genome of the influenza isolates were generated with a OneStep RT-PCR kit (QIAGEN, Valencia, CA, USA). They were treated with shrimp alkaline phosphatase-exonuclease I (U.S. Biologicals, Swampscott, MA, USA) before sequencing. Sequencing and Assembly

Sequencing reactions were performed as described previously (5). After sequencing, each segment was downloaded, trimmed to remove amplicon primer-linker sequence as well as low-quality sequence, and assembled. A small genome assembler called Elvira, based on the open-source Minimus assembler (http://cbcb.umd.edu/software), has been developed to automate these tasks. The Elvira pipeline delivers exceptions, including failed reads, failed amplicons, insufficient coverage of a reference sequence (as obtained from GenBank), ambiguous consensus sequence calls, and low-coverage areas. Additional sequencing and targeted RT-PCR were conducted to close gaps and to increase coverage in low-coverage or ambiguous regions.

All sequence data used in this study are available from GenBank and also from ftp://cbcb.umd.edu/pub/data/flu. GenBank accession numbers are available in the supplementary data (Technical Appendix 1).
Phylogenetic Analysis
Multiple sequence alignments of nucleotide data were performed by using MUSCLE (8) with default parameters. Most alignments of segments within a subtype lack internal gaps. Leading and trailing gaps were not considered in tree-length calculations, but all nucleotide positions were considered.