H5N1 Recombination Evidence Excluded from Genbank
Recombinomics Commentary
October 26, 2006
H5N1 bird flu sequences were released this month via the NIAID Influenza sequencing project (see list here). The samples were submitted by the Capua lab in Italy. Although HA sequences from several of the isolates had been reported previously, many were partial sequences and in most cases the sequences from the other seven gene segments were not released.
The NIAID program generates full sequences on all eight gene segments, which provide powerful profiles that can be used to trace origins or newly emerging genes, which can be used. These origins can be used to predict new sequences, as was done for S227N in Turkey at the beginning of this year.
Recent data from northern China has included compelling evidence that H5N1 in China is evolving via recombination. These sequences are transmitted and transported by migratory birds, which has led to widespread detection of the Fujian strain, which was present in all human cases in China, and subsequently detected in wild birds in Hong Kong, as well as Laos and Malaysia earlier this year. Moreover, these sequences are also appearing in human Indonesian isolates, which do not match the poultry isolates.
Thus, a full sequences of all eight gene segments are essential for predicting new sequences generated by recombination. The full sequence of the HA gene of a Qinghai isolate from Afghanistan demonstrates how recombined regions are excluded in partial sequences, and raises questions on why the H5N1 sequences generated by St Jude and Hong Kong University are largely partial sequences. The partial sequences are from H5N1 isolates in Hong Kong, China, and Vietnam and are important in mapping the interactions in dual infections involving H5N1 in Asia.
The original HA sequence from a chicken A/chicken/Afghanistan/1207/2006 was 1602 BP and was missing the 5’ portion of the gene, the sequence from the NIAID program was a full sequence of 1734 BP and the region missing from the initial submission included a string of four polymorphisms between positions 42 and 62. These polymorphisms are defined by Qinghai H5N1 HA sequences, and clustering of such polymorphism is a strong signal for recombination.
The first polymorphism (T42C) was also found in a 2003 duck isolate from China (see detail here). The second polymorphism was found in two 2004 duck H5N2 isolates from Malaysia (A/duck broiler/Malaysia/F118/08/04 and A/duck broiler/Malaysia/F189/07/04). The third polymorphism, C53, was found in a large number of low and high path H5 including all three isolates above (C53T is found in Qinghai isolates from Nigeria). The fourth polymorphism, G62A, has a much more limited distribution, which includes high path H5N1 in Mongolia and China as well as the two low path ducks from Malaysia. The matching of 3 of the 4 continuous polymorphisms strongly suggests that these three polymorphism were acquired from a sequence similar to the low path duck sequences in Malaysia.
These acquisitions are further supported by two additional polymorphisms downstream from the cluster of three. G175A is present in the full sequence from NIAID. The wild type sequence is in the early isolate. G175A also has a limited distribution among H5 isolates, including the two ducks from Malaysia. Similarly, C268T also has a limited distribution in H5N1 isolates from Asia (Russia, Mongolia, and Indonesia) as well as the two low path ducks from Malaysia. C268T is in both Afghan sequences. The same type of distribution is found for T868C, which is in high and low path H5 isolates from Asia and Europe, including the two low path ducks from Malaysia.
Thus, of the six polymorphisms shared by the Malaysian ducks and the chicken from Afghanistan, only two were present in the original partial sequence. The first three were missing because only a partial sequence was submitted, and the fourth polymorphism difference in the two submitted sequences. The presence of the two low path duck sequences in the origin Afghan sequence suggested recombination. However, the presence of six polymorphisms in the full sequences strongly suggests that these sequences were present in the high path Qinghai sequence because of recombination with a low path H5 sequence similar to the two sequences from Malaysia. Confirmation via additional sharing in the other 7 gene segments was limited because the N in the low path was N2 instead of N1 in the Qinghai strain, and the other six gene segments from the Malaysian isolates have not been submitted.
This example highlights the limitations created by partial sequences and the absence of information for all eight gene segments. This missing data, limits the tracing of the polymorphisms, which identifies parental strains as well as prior dual infections.
It remains unclear why the majority of H5N1 sequences from Hong Kong, China, and Vietnam submitted by St Jude and Hong Kong University are partial sequences, particularly for the four genes most likely to show recombination (PB2, PB1, PA, and NP), but full sequences should be generated and placed in the database to provide a clearer picture of recombination and dual infections involving the various versions of H5N1 in Asia.
Media Sources
Phylogenetic Trees