Short Regions of Recombination in Human Influenza
Recombinomics Commentary 12:20
March 27, 2008
Using an exhaustive search and a nonparametric test for mosaic structure, we identified 315 sequences (2%) in five different RNA segments that, after a multiple comparisons correction, had statistically significant mosaic signals compatible with homologous recombination.
Given our inability to exclude the occurrence of mixed infection and template switching during amplification, laboratory artifact provides an alternative and likely explanation for the occurrence of phylogenetic incongruence in these two cases. We therefore conclude that, if it occurs at all, homologous recombination plays only a very minor role in the evolution of human influenza A virus.
The above comments from the abstract of the upcoming paper, "Homologous Recombination is Very Rare or Absent in Human Influenza A Virus” are curious. Previously, the authors had discounted homologous recombination in negative sense RNA viruses. In the introduction they acknowledged the recent data demonstrating homologous recombination in Ebola, a negative sense RNA virus, as well as bioinformatics analysis of influenza of recombination in Canadian swine. However, the Ebola data was qualified with an “if true” and the influenza data was cited in a paragraph on the controversial data on recombination in influenza.
The paper repeatedly cited influenza dogma on frequent copy errors as the source of single nucleotide polymorphisms associated with genetic drift, and discounted the paper's multiple examples of statistically significant recombination. In NA from H3N2 isolates, the p value was less than 10 to the minus 10, indicating the likelihood that the data was due to chance was less than one in a billion. However, there were internal inconsistencies in the data, including no examples of recombination in HA or two other gene segments. Similarly, the overall frequency of recombination was markedly lower in H1N1 than H3N2 and the authors repeatedly cited contamination or co-infections as possible causes of in vitro recombination during amplification. However, no data was provided to show that the examples in the database were due to amplification.
An alternate explanation for the internal inconsistencies in the level and distribution of examples may be linked to the experimental design of the studies. The authors created a database of full sequences from human influenza. Partial sequences, as well as influenza sequences from other sources, such as birds or swine, were not included. Moreover, the study required the identification of the recombinant as well as both parental sequences, so the limited database would lower the number of identified examples.
The limitations could be easily demonstrated with a series of six HA recombinants from South Korea. The six isolates were from human influenza samples collected in 2002. All six had obvious recombination with human H3N2 sequences in circulation a decade earlier. However, the Korean sequences were about 1650 base pairs in length, and all HA sequences in the study were greater than 1700 base pairs, so these six recombinants would not be included in the database used in the study. Moreover, most public HA sequences from isolates collected in the early nineties were about 1000 base pairs in length, so these parental sequences would also be excluded from the database. Thus, even though these six isolates had clear examples of recombination, none were included in the examples in the paper, and the paper found no HA examples (in either H3N2 or H1N1).
The six recombinants were obvious examples. The first and last third of the HA gene matched other 2002 H3N2 isolates, while the central third of the gene matched human H3N2 sequences from isolates collected a decade earlier. The sequences from these six recombinants also provided a mechanism for the generation of shorter regions of recombination. Although all six recombinants had crossovers in the same region, three of the recombinants (A/Cheonnam/323/2002, A/Cheonnam/338/2002, A/Cheonnam/340/2002) had a nested region of 2002 sequences, which split the 1991 sequences in half. This nested segment split the larger segment so instead of being all 1991, it became 1991-2002-1991. The other three isolates, (A/Kyongbuk/320/2002, A/Daejeon/258/2002, A/Incheon/260/2002) had a longer region which could be used in phylogenetic analysis.
This same type of nesting was seen in the swine influenza sequences, where two recombinants shared outer crossover points, but the recombined region had nested sequences which reduced the size of the recombined region and produced alternating sequences as demonstrated above. Thus, the identification of small regions of recombination would not be unexpected.
Moreover, a template switching mechanism of recombination would favor recombination between closely related genomes, because the regions of identity would be larger, creating more opportunities for the nascent RNA to re-anneal. Similarly, co-infections would more commonly involve closely related sequences because region specific subclades are common. Moreover, the newly acquired sequences would have a limited number of new polymorphisms, and would look like point mutations, or single nucleotide polymorphisms.
These examples have been described for H5N1 polymorphisms from clade 2.2 isolates, and these examples are common. This type of recombination would not be identified with the methods used in the paper. Moreover, recombination with closely related sequences would also reduce the size of sequences acquired in earlier recombination events, and create short regions such as those that were statistically significant in the upcoming paper.
Thus, much of the positive data in the paper could be explained by multiple recombination events, which were not addressed. Therefore, although the recombination identified was not of sufficient length for phylogenetic analysis, the finding of short regions of recombination is supported by prior results and was not unexpected.
Media Links
Recombinomics Presentations
Recombinomics Publications
Recombinomics Paper at Nature Precedings