Evolution of Influenza Neuraminidase and the Detection of Antiviral

Nov 13, 2013 - congruent with sequence based trees. Such trees can be built solely from the masses of the proteolytic peptide ions of viral proteins r...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/ac

Evolution of Influenza Neuraminidase and the Detection of Antiviral Resistant Strains Using Mass Trees Kavya Swaminathan and Kevin M. Downard* School of Molecular Bioscience, University of Sydney, Sydney, New South Wales NSW 2006, Australia ABSTRACT: A new approach employing mass trees is described and implemented which enables the evolution of influenza neuraminidase across all subtypes (N1−N9) in human and animal hosts to be monitored and charted without gene or protein sequencing. These mass trees are shown to be congruent with sequence based trees. Such trees can be built solely from the masses of the proteolytic peptide ions of viral proteins recorded by a mass spectrometer. They are shown to be able to correctly chart the evolutionary history of human pandemic influenza viruses, which originated in animal hosts, and can also resolve antiviral resistant from sensitive strains. Furthermore, experimental mass map data recorded for a circulating strain is correctly positioned onto a mass tree so as to quickly establish its evolutionary history and identify whether it is resistant or sensitive to the antiviral inhibitor oseltamivir. This new computational approach is expected to find wider application for evolutionary studies of organisms more generally.

I

evolution of the virus can render primers that target these genes to be ineffective so as to require their continual redesign. Most strains in circulation, therefore, are only characterized at the non-molecular level using simpler assays and screens.12 We have recently reported on a new approach with which to chart the evolutionary history of the influenza virus, and by extension to any other organism, without the need for gene or protein sequencing.13 This has been achieved using so-called “mass trees”. These are constructed using the theoretical or experimental masses of peptides derived from the proteolytic digestion of proteins.13 The masses of these proteolytic peptides reflect the sequence of the protein. Homologous proteins will yield sets of peptide masses that are identical mass in the sections of the protein that share a common sequence and peptide masses which are different in segments of the protein where the sequence differs. When high-resolution mass spectrometry is employed, even subtle changes to each peptide’s sequence will alter its mass.13 Only the substitution of leucine for isoleucine is undetectable. Lists of peptide masses, rather than sequences, can therefore be used to construct phylogenetic style “mass trees” using an algorithm that has been written specifically for this purpose.13 The processing of mass map data used to identify bacterial strains for tree generation purposes has been recently considered in an independent study by others14 and an alternative mass based phylogenetic approach has also been implemented on a small

nfluenza neuraminidase is a glycoprotein on the surface of the influenza virus that facilitates the release of progeny viruses from the host cell at the end of the viral replication cycle.1 It does so by the hydrolysis of sialic acid receptors on the host cell.2 Antibodies or antiviral inhibitors that target the protein can impede the release of progeny viruses and thereby reduce the viral load and the duration symptoms and severity of infection.3,4 A loss of host immunity to the influenza virus is attributed in part to the accumulation of mutations within influenza neuraminidase through a process known as antigenic drift.5 The same mechanism confers resistance to antiviral neuraminidase inhibitors if such mutations occur within the active site of the sialidase enzyme.6 Antiviral resistance mutations, such as H274Y, have been widely observed in strains associated with regional epidemics.7 Monitoring the evolution of this protein is a key requirement of the global surveillance of the influenza virus to mitigate its impact on human and animal health.8 The evolution of the influenza virus is most often studied using phylogenetic trees.9 In the case of influenza neuraminidase, multiple gene sequences that encode the protein are aligned across strains to establish common sequence motifs and to chart their evolutionary origins.10 Such phylogenetic analyses have provided insights into how influenza viruses evolve but are limited by the time required to conduct full-length gene sequencing of a large number of strains in circulation.10 Despite recent developments in gene sequencing technology, the characterization of influenza viruses employing PCR based methods is only performed on a limited set of circulating strains and has been described as ad hoc and reactive rather than systematic and widespread.11 Furthermore, the unpredictable © 2013 American Chemical Society

Received: September 11, 2013 Accepted: November 13, 2013 Published: November 13, 2013 629

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Figure 1. Mass tree constructed for 1620 N1 neuraminidase sequences of all human HxN1 virus strains.

set (of 50 neuraminidase sequences). A much larger data set comprising over 12 000 translated gene sequences for all neuraminidase subtypes (N1−N9) in all hosts (human and animal) was also downloaded and used to generate the unrooted phylogram. The FluSim13,16 algorithm was used to generate theoretical monoisotopic masses for protonated tryptic peptide ions using the sequences within each data set. Note that only the isomeric residues leucine and isoleucine can be substituted without a change to the mass of a peptide or protein. The experimental mass spectral data was acquired by MALDI mass spectrometry on a high resolution mass spectrometer as previously described.17−19 Deisotoped spectral data containing only the measured monoisotopic masses of all peptide ions were exported from the DataAnalysis software (Bruker Daltonics, Breman Germany) and input into the MassTree algorithm. The high resolution mass spectrometer used routinely achieves a mass precision at or below 5 ppm using an external calibration, and this mass error threshold was therefore used by the MassTree algorithm. Basis of the MassTree Algorithm and MassTree Construction. The MassTree algorithm13 reads t sets of monoisotopic m/z values theoretically calculated on the basis of a protein sequence or detected in a mass spectrum following the proteolytic digestion of a protein. The m/z values contained within pairs of sets are compared to establish the number of indistinguishable mass values, within a specified mass tolerance (default 5 ppm), and those that differ in mass but which correlate with a single amino acid substitution. A distance score is then computed on the basis of the number of matching mass values within each set. This scoring

scale using bacterial nucleic acids cleaved with restriction enzymes.15 Mass trees can be used to trace the evolutionary history of the influenza virus following the digestion of viral proteins. This can be achieved post their recovery or even within whole viral digests where the FluShuffle algorithm16 is first employed to identify protein constituents. This affords a significant time saving over gene or protein sequencing given that mass spectral data of protein digests can be recorded within a fraction of a second. Here, the approach is employed to study the evolution of N1 and N2 neuraminidase within human influenza viruses and across all subtypes (N1−N9) in human and animal hosts. The ability to distinguish antiviral resistant strains from sensitive strains is demonstrated as well as the fitting of experimental mass map data onto a mass tree to establish strain susceptibility.



EXPERIMENTAL SECTION Generation of Theoretical and Experimental Mass Data. Translated, full-length neuraminidase gene sequences for various data sets were downloaded from the NCBI Influenza Virus Resource Database (IVRDB) on 30 May 2012. All 3263 N1, 1620 human N1, and 1468 human N2 sequences were downloaded with those containing identical sequences collapsed into a single nonredundant entry. Subsets of these sequences containing one strain accession from each year, with every region represented, were constructed. These contained 595 N1 and 240 N2 neuraminidase sequences, respectively. From these subsets, broader regional and temporal filters were employed to generate smaller subsets consisting of two N1 data sets (of 250 and 50 neuraminidase sequences) and one N2 data 630

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Figure 2. Mass (left) and sequence (right) trees constructed for a subset of 595 N1 neuraminidase sequences of human HxN1 virus strains.



approach minimizes the distance between two sets of mass values that contain nearly identical masses, including those that differ only by a single amino acid substitution, where less weight is given to the latter. A distance matrix is then generated through pairwise comparison of mass values across all data sets in order to construct the mass trees. The most common distance based method, a relaxed neighbor joining approach,20 was employed using the Clearcut algorithm.21 The branch lengths of the mass tree reflect the ratio of the number of different mass values to the total within each set. Construction of Corresponding Sequence Trees. The corresponding sequence trees were generated using translated neuraminidase gene sequences, for strains identical to those used to construct mass trees, extracted from the NCBI Influenza Virus Resource Database in FASTA format. Multiple sequence alignment was performed with the Clustal X algorithm (v2.1),22 and the corresponding sequence trees were generated using the neighbor-joining method. Both the mass trees and sequence trees were visualized and midpoint rooted with Archaeopteryx software.23 A custom written TreeCompare algorithm was used to visualize the similarities among trees by highlighting matching clades in each tree using a unique color. Comparison of Mass and Sequence Trees. The topologies of the mass and sequence trees were compared using the Compare2Trees24 and MAST algorithms25,26 as previously described.13 These algorithms report overall topology scores, congruence indices, p-values, and corrected pvalues (or q-values calculated using the R algorithm) that can be used to establish whether the trees are more congruent than by chance.

RESULTS AND DISCUSSION

Construction of Influenza Neuraminidase N1 and N2 Mass Trees. Influenza neuraminidases of the N1 and N2 subtype represent those found in the human population among the type A H1N1, H3N2 and, less often, H5N1 strains. Mass trees were constructed for representative N1 and N2 neuraminidase sequences acquired from the Influenza Virus Resource database and then selected according to the year of isolation. One strain accession from each year, for every represented region, was chosen to obtain a subset of representative sequences consisting of 595 N1 and 240 N2 neuraminidase sequences. From these subsets, broader regional and temporal filters were used to generate smaller subsets consisting of two N1 data sets (of 250 and 50 neuraminidase sequences) and one N2 data set (of 50 neuraminidase sequences). Simulated theoretical mass data were generated for these data sets using the FluSim algorithm.13,17 The masses for these tryptic peptides were used to construct the respective mass trees. The N1 neuraminidase mass tree constructed for 1620 HxN1 human strains of the influenza virus is shown in Figure 1. All human H5N1 strains were found to cluster in the upper subtree. The human H1N1 strains were found to cluster according to their year of isolation in periods ranging from 1933 to just before the 2009 pandemic. Among the latter strains post 2007, a separate subclade of strains that contain the neuraminidase H274Y (based on N2 numbering) oseltamivir resistance mutation (shown colored in red) was found to have evolved from other antiviral sensitive circulating strains in the same period. The same chronological separation of strains into 631

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Table 1. Results of Two Tree Comparison Algorithms When Mass and Sequence Trees Generated from Different Numbers of N1 and N1 Neuraminidase Sequences Were Compared MAST subtype and number of sequences N1 N1 N1 N2 N2 a

human human human human human

− − − − −

595a 250 50 240 50

Compare2Trees, %

Icong

54.7 55.5 69.3 66.4 71.7

3.13 2.32 4.16 2.61

p-value 4.49 4.14 2.99 1.96

× × × ×

10−29 10−11 10−42 10−13

q-value 1.80 1.66 1.20 7.84

× × × ×

10−28 10−10 10−41 10−13

The subset of 595 N1 sequences was too large to allow for a comparison with the MAST algorithm.

Figure 3. High resolution MALDI mass spectrum of the tryptic digest of the recombinant N1 neuraminidase derived from the A/California/04/2009 (H1N1) strain. Ions denoted with ∗, Λ, and # represent oxidized, pyroglutamate, and carboamidomethylated peptides, respectively.

resistant H274Y mutant strains post 2007 through 2009 (highlighted in red). The avian HxN1 strains form part of a separate subtree and have evolved solely in birds from the early 1930s through 2011 (shown in pink). Swine H1N1 influenza that evolved from avian strains sometime during this period and in circulation from 2002 subsequently crossed to human hosts and resulted in the swine-originating pandemic human H1N1 viruses that appeared in the population in early 2009. These have remained in circulation as the predominant H1N1 human strains ever since. A separate cluster below these strains is associated with avian HxN1, including the H5N1, strains in circulation from the mid-1970s from which the highly pathogenic avian and human H5N1 strains evolved in the past decade. The Compare2Trees24 and MAST algorithms25,26 were employed to facilitate a quantitative comparison of the N1 neuraminidase mass and sequence trees (Figure 2). These algorithms generated overall topology scores, congruence indices, and p-values (Table 1). Only the Compare2Trees algorithm was able to analyze the largest subset of 595 N1 neuraminidase sequences. It returned an overall topology score of 54.7% (Table 1). This value is consistent with that obtained upon the comparison of mass and sequence trees for H1

subtrees and clades, according to their year of isolation, was evident for N2 neuraminidase (data not shown). Noteworthy is the separate subtree that is observed for the 2009 pandemic and post-pandemic strains. These strains have evolved differently from other circulating strains in 2009 consistent with reports of their animal origins.27 Also of note is that current strains, post 2009 to the present, all derive from the pandemic strains of 2009. These have now established themselves as the most common H1N1 strains in circulation on a seasonal basis. Comparison of N1 Mass and Sequence Trees Using the Compare2Trees and MAST Algorithms. To assess how well the mass tree reflects the evolution of N1 neuraminidase, it was compared with a conventional sequence based tree produced from the same subset of 595 neuraminidase sequences. Both phylogenetics tree are shown side-by-side in Figure 2. The tree topologies are highly similar. In both cases, the subtree contains H1N1 swine strains in circulation from 1930 to 2002 (shown in green). In the rest of the subtree are clades containing human H1N1 strains in co-circulation across a similar period through 2008/9 (shown in blue) with a separate clade showing the subsequent evolution of antiviral 632

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Figure 4. Mass tree of all N1 neuraminidases across all hosts. Swine strains are colored in green for human strains, in blue or red for H274Y mutant strains, and in pink for avian strains. Experimental mass data for the neuraminidase of the A/California/04/2009 strain is mapped onto the tree as shown by the brown arrow (see also enlarged insert on right-hand side).

pink) and swine (shown in green) strains of the H5N1, HxN1, and H1N1 subtypes. The N1 neuraminidase of the A/California/04/2009 (H1N1) strain was correctly located on the subtree containing other human H1N1 pandemic 2009 and post-pandemic strains. As shown in the insert to Figure 4, it resides in a clade containing 2009 pandemic strains, including the very same strain positioned on the basis of theoretical masses that represent the entire sequence. The experimentally derived mass data places the strain on a separate branch, but still within this clade, due the lower number of mass values input to the mass tree algorithm. The experimental mass values reflect peptide segments that span 19% of the sequence versus 100% sequence coverage for all strains positioned on the basis of the theoretical mass values. Previous work13 has shown that, even with mass values that span only 17% to 47% of the sequence, influenza proteins can be correctly positioned on a mass tree and separated according to the viral subtype. Potentially increasing the sequence coverage by using LC-ESI-MS, over MALDI-MS, limits the desired throughput of such studies and is challenged by the low viral protein levels and contaminants when clinical specimens are analyzed for this purpose27 using whole viral digests and the FluShuffle algorithm.16 The evolutionary history of the 2009 pandemic N1 neuraminidase, reflected in the mass tree in Figure 4, is consistent with its reported gene origins.28 According to the study of Garten and co-workers,29 the N1 neuraminidase of the human 2009 pandemic strain crossed from avian to swine hosts

hemagglutinin derived from type B human influenza viruses for a similar sized data set.13 As the trees composed of 595 N1 neuraminidase sequences were too large to be compared by the MAST algorithm, mass and sequence trees were constructed for two subsets composed of 250 and 50 sequences (data not shown). The Compare2Trees algorithm recorded topology scores of 55.5% and 69.3%, respectively, for these trees, while the MAST algorithm returned congruence indices of 3.13 and 2.32, respectively (see Table 1). Note that in both cases the p-values and corrected p- (or q-) values were very low and also very similar to one another and ranged from 10−11 to 10−29. This reflects that the trees are far more congruent than is probable by chance. Establishing the Evolutionary History of an Influenza Strain by Fitting Mass Spectral Data to a Mass Tree. The use of mass trees to establish the origins of an influenza neuraminidase subsequently analyzed by mass spectrometry was investigated using the MALDI mass spectrum of the tryptic digest of a recombinant N1 neuraminidase. Its sequence is based on the A/California/04/2009 (H1N1) strain (Figure 3). The experimental peptide ion m/z values, shown in the mass spectrum, were added to those for all N1 neuraminidases across all human and animal hosts. These were used to build the allN1 mass tree shown in Figure 4. This tree is similar to the N1 neuraminidase mass tree of Figure 2, where human strains are shown colored in blue and red, the latter containing the strains bearing the H274Y mutation. It also contains avian (shown in 633

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Figure 5. Unrooted mass tree depicting the divergence of all neuraminidase subtypes (N1−N9) of type A influenza viruses across all hosts for all known influenza neuraminidase sequences (>12 000). Avian, human, and swine strains are shown colored in pink, blue, and green, respectively, and outgroups are designated by an asterisk. Canine and equine N8 strains are colored in brown.

as early as 1979. Using bioinformatics approaches, it was found that the pandemic neuraminidase gene was closest to those of swine H1N1 viruses isolated in Europe between 1991 and 1993. Smith et al.30 reported that the time that the most recent common ancestor of the swine NA gene crossed to humans was in August of 2008. Origins and Divergence of Group 1 and Group 2 Neuraminidases Reflected in a Mass Tree. The MassTree algorithm was subsequently used to construct trees from much larger data sets so as to enable the divergence profile of all known influenza neuraminidases to be studied. Simulated theoretical mass data derived from all available full length neuraminidase sequences for all type A influenza strains within the IVRDB (exceeding 12 000 sequences) were used to generate a single mass tree. The resulting tree was visualized as an unrooted phylogram, given the parallel evolution of many strains, using the Archaeopteryx software (Figure 5). The tree shows two distinct phylogenetic groups of neuraminidases separated by the near vertical bold dividing line. The group 1 neuraminidases, clustered to the right-hand side of the tree, consist of the N1, N4, N5, and N8 subtypes. The group 2 neuraminidases that are clustered toward the left consist of the N2, N3, N6, N7, and N9 subtypes. A comprehensive analysis of the clustering profile represented

for each neuraminidase group on the tree has revealed that it is consistent with the known evolutionary histories of influenza neuraminidase. Evolution of Group 1 Neuraminidases (N1, N4, N5, and N8). The group 1 neuraminidase cluster comprises five main subtrees (ST1−5). Four of them are composed of the N1 neuraminidases (ST2−5). The N1 neuraminidases have shown the greatest host diversity and include the H1N1 strains that caused the human pandemics of 1918 (ST3) and 2009 (ST4) as well as the highly pathogenic avian influenza strains (H5N1) that caused a serious outbreak of the disease in humans in 1997 and continues to pose a looming pandemic threat (ST5). The N1 neuraminidase from H1N1 swine viruses form a separate subtree (ST2) composed of two clades. One of these clades represents the strains in circulation since 1930 referred to as classical swine viruses.30 Following a reassortment event of the classical swine with a triple reassortant H3N2 virus in 1998,28 the H1N1 swine viruses began to steadily evolve within the swine population through antigenic drift. This is reflected by the formation of a separate clade with strains from early 2002 and is consistent with reports of these viruses continuing to be endemic in pigs.31 The other known swine influenza H1N1 viruses are of the Eurasian swine lineage (ST4). The neuraminidase of this virus 634

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

Figure 6. Mass tree constructed for all H1 human hemagglutinin showing the location of strains (highlighted in red) which contain the H274Y mutation in their corresponding neuraminidase sequence.

neuraminidase stalk region.38 These mutations, combined with the unusual addition of multiple basic residues within the hemagglutinin HA0 protein that is cleaved into the two subunits HA1 and HA239 and its altered glycosylation pattern, are reported to have enhanced the virus’ pathogenicity.38,39 The remaining subtree (ST1) consists of strains of the N4, N5, and N8 subtypes. This subtree is found far removed from the other group 1 subtrees, owing to their exclusivity in avian hosts. In only a few sporadic cases have equine and canine HxN8 viruses been reported and characterized. Evolution of Group 2 Neuraminidases (N2, N3, N6, N7, and N9). Among group 2 neuraminidases, those that predominant have been isolated from avian hosts. Those of the N3, N7, N6, and N9 subtypes are seen to cluster together in a subtree (ST10) at the base of the tree (Figure 5). Human neuraminidases derived from sporadic infections caused by the avian HxN9 viruses were found to be indistinguishable.40 The mass tree reflects a remarkably low level of divergence in the neuraminidase of these subtypes, similar to the group 1 neuraminidases that predominated in avian hosts (N4, N5, and N8) (ST1). This is attributed the virus being well adapted in birds.41 Goup 2 neuraminidase viruses, however, evolve rapidly when they enter into mammalian hosts. The increased evolutionary activity is evidenced by the divergence of viruses in the mass tree for the N2 subtype (Figure 5). The N2 neuraminidase is known to have had a complicated evolutionary dynamic, shuttling at various time points between avian, swine, and human hosts.42

lineage is known to have evolved solely from avian viruses in circulation in Europe and Asia.32 Acordingly, it is placed in a separate subtree close to the avian HxN1 viruses. This subtree also contains the 2009 human pandemic N1 neuraminidases. The 2009 pandemic H1N1 virus was a result of recombination of viral proteins from three different viral lineages (a triple reassortant) with the neuraminidase derived from strains of a Eurasian swine lineage.28,29 In between the classical swine and pandemic neuraminidase subtrees are the human N1s from the H1N1 viruses (ST3) that have circulated since 1918.33 These human viruses were found be antigenically similar to the classical swine H1N1 strains.29,33 This accounts for their placement in the mass tree. The 1918 viruses circulated until 1957 when they were replaced completely by an H2N2 strain.34 The H1N1 viruses reemerged in 1977 after a two decade absence. Their reemergence is attributed to suspected accidental release from a laboratory.35 Between 1977 and 2009, the human H1N1 virus is known to have undergone substantial antigenic drift.29 This is reflected in the pre 2009-pandemic subtree (ST3) that shows several divergent branches. The ST5 subtree contains neuraminidase from avian influenza viruses that are referred to as highly pathogenic influenza (HPI).36 These H5N1 viruses diverged from the other avian viruses and were associated with the “bird-flu” that caused devastating infections in poultry with occasional human infection.37 The divergence of these highly pathogenic strains, from their low pathogenic counterparts (on ST4) in the mass tree, is attributed to the deletion of 22 amino acids in the 635

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

circulating strains. This is both time-consuming and adversely impacted by the evolution of the virus that can render PCR based amplification methods ineffective. Mass trees offer a viable alternative with which to characterize and chart the evolution of influenza viruses without gene or protein sequencing. Mass trees can be generated from experimental mass map data more rapidly and directly than PCR based gene sequencing. Subject to the mass spectrometer employed, there are benefits in terms of infrastructure costs, and mass map data can be generated with high sample throughput with currently available instrumentation. The evolution of type A influenza neuraminidase within human N1 and N2 viruses, and across all subtypes (N1−N9) in human and animal hosts, have been monitored and charted using only mass data. The results demonstrate that such trees can correctly chart the evolutionary history of human pandemic influenza viruses and identify and distinguish antiviral resistant ones from sensitive ones.

HxN2 viruses, originated from birds, have been characterized as early as 1963. These form two clades on a separate subtree (ST9) that are diverged from the other avian group 2 neuraminidases (of subtree ST10). A reassortment event between avian H2N2 viruses of this subtree and human H1N1 viruses, in circulation at the same time, resulted in the formation of a novel H2N2 human virus bearing the avianoriginating N2 gene. This H2N2 virus was responsible for a pandemic in 1957.43 The H2N2 virus subsequently evolved in humans through antigenic drift44 as seen in a separate subtree (ST8). An avian H3N2 virus strain reassorted with this virus in 1968 resulting in a new H3N2 strain that was responsible for a pandemic outbreak.43,44 The human H3N2 strain carried avian viral genes (HA and PB1) but retained the N2 neuraminidase gene derived from the human H2N2 virus it replaced.42 Both the N2 neuraminidases are therefore clustered together in the same subtree (ST8). The subsequent evolution of human H3N2 viruses post 1982 is evident in the ST6 and ST7 subtrees. The latter contains the antigenically distinct Fujian H3N2 human seasonal strains. An interesting feature observed in the N2 neuraminidases within the tree is the placement of the swine neuraminidases along with the human in almost every subtree. This observation is in line with the co-infection and recombination events between the human and swine influenza viruses at various time points. Compensatory Mutations in Influenza Hemagglutinin in Drug Resistant Strains. Having established that oseltamivir antiviral resistant H274Y mutations cluster in the neuraminidase mass tree, the ability of the MassTree algorithm to similarly cluster the hemagglutinin protein in the same viral strains was evaluated. Compensatory mutations have been acquired by influenza hemagglutinin in the drug resistant strains in order to achieve a functional balance between the viral hemagglutinin and neuraminidase.45 Such compensatory mutations have been shown to underlie the viral fitness of the oseltamivir resistant strains,46 since the presence of the H274Y mutation in their neuraminidase impedes the virus’ ability to detach itself from the host cells and thus compromises its sialidase activity. As with the neuraminidase mutants, the scoring function in the MassTree algorithm is expected to allow for the clustering of hemagglutinin in the same strains that have acquired one or more point mutations. These include various point mutations including D35N, T82K, Y94H, K141E, R189K, R209K, and E274K.46 The mass tree generated for all human H1 influenza hemagglutinin sequences reported is shown in Figure 6. It resembles that previously reported.13 The hemagglutinins derived from the H274Y mutant strains up to early 2009 were found to form a separate clade within the cluster of other pre-pandemic H1N1 strains (see Figure 6 highlighted in red). Those post the 2009 pandemic appear scattered throughout the lower subtree amidst strains that are mostly sensitive to oseltamivir.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: +61 (0)2 9351 4140. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by an Australian Research Council Discovery Project grant (DP120101167) awarded to Kevin M. Downard and Jason W. H. Wong.



REFERENCES

(1) Air, G. M.; Laver, W. G. Proteins: Struct., Funct., Genet. 1989, 6, 341−356. (2) Shtyrya, Y. A.; Mochalova, L. V.; Bovin, N. V. Acta Naturae 2009, 1, 26−32. (3) Colman, P. M. Protein Sci. 1994, 3, 1687−1696. (4) von Itzstein, M.; Wu, W. Y.; Kok, G. B.; Pegg, M. S.; Dyason, J. C.; Jin, B.; Van Phan, T.; Smythe, M. L.; White, H. F.; Oliver, S. W. Nature 1993, 363, 418−423. (5) Luther, P.; Bergmann, K. C.; Oxford, J. S. J. Hyg., Camb. 1984, 92, 223−229. (6) Collins, P. J.; Haire, L. F.; Lin, Y. P.; Liu, J.; Russell, R. J.; Walker, P. A.; Skehel, J. J.; Martin, S. R.; Hay, A. J.; Gamblin, S. J. Nature 2008, 453, 1258−1261. (7) Hurt, A. C.; Holien, J. K.; Parker, M. W.; Barr, I. G. Drugs 2009, 69, 2523−2531. (8) Nelson, M. I.; Holmes, E. C. Nat. Rev. Genet. 2007, 8, 196−205. (9) Nei, M.; Kumar, S. Molecular evolution and phylogenetics; Oxford University Press: Oxford UK, 2000. (10) Xu, J.; Davis, C. T.; Christman, M. C.; Rivailler, P.; Zhong, H.; Donis, R. O.; Lu, G. PLoS One 2012, 7, No. e38665. (11) Butler, D. Nature 2012, 483, 520−522. (12) World Health Organisation Global Interim Epidemiological Surveillance Standards for Influenza Manual, July 2012; http://www. who.int/influenza/resources/documents/INFSURVMANUAL.pdf. (13) Lun, A. T. L.; Swaminathan, K.; Wong, J. W. H.; Downard, K. M. Anal. Chem. 2013, 85, 5475−5482. (14) Simmuteit, S.; Schleif, F. M.; Villmann, T.; Hammer, B. Knowl. Inf. Syst. 2010, 25, 327−343. (15) von Wintzingerode, F.; Böcker, S.; Schlötelburg, C.; Chiu, N. H.; Storm, N.; Jurinke, C.; Cantor, C. R.; Göbel, U. B.; van den Boom, D. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 7039−7044. (16) Lun, A. T. L.; Wong, J. W. H.; Downard, K. M. BMC Bioinf. 2012, 13, 208.



CONCLUSIONS Molecular based studies of influenza neuraminidase are of vital importance in order to chart the evolution of the virus and identify and monitor mutations that confer strains with resistance to current antiviral inhibitors. Phylogenetics approaches are central to this goal but, to be effective, require the large scale sequencing of neuraminidase genes across many 636

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637

Analytical Chemistry

Article

(17) Wong, J. W. H.; Schwahn, A. B.; Downard, K. M. BMC Bioinf. 2010, 11, 266. (18) Schwahn, A. B.; Wong, J. W. H.; Downard, K. M. Anal. Chem. 2009, 81, 3500−3506. (19) Schwahn, A. B.; Wong, J. W. H.; Downard, K. M. Analyst 2009, 134, 2253−2261. (20) Evans, J.; Sheneman, L.; Foster, J. A. J. Mol. Evol. 2006, 62, 785−792. (21) Sheneman, L.; Evans, J.; Foster, J. A. Bioinformatics 2006, 22, 2823−2824. (22) Larkin, M. A.; Blackshields, G.; Brown, N. P.; Chenna, R.; McGettigan, P. A.; McWilliam, H.; Valentin, F.; Wallace, I. M.; Wilm, A.; Lopez, R.; Thompson, J. D.; Gibson, T. J.; Higgins, D. G. Bioinformatics 2007, 23, 2947−2948. (23) Han, M. V.; Zmasek, C. M. BMC Bioinf. 2009, 10, 356. (24) Nye, T. M. W.; Lio, P.; Gilks, W. R. Bioinformatics 2006, 22, 117−119. (25) Kubicka, E.; Kubicki, G.; McMorris, F. R. J. Classification 1995, 12, 91−99. (26) de Vienne, D. M.; Giraud, T.; Martin, O. C. Bioinformatics 2007, 23, 3119−3124. (27) Fernandes, N. D.; Downard, K. M. J. Clin. Microbiol., in press. (28) Smith, G. J.; Vijaykrishna, D.; Bahl, J.; Lycett, S. J.; Worobey, M.; Pybus, O. G.; Ma, S. K.; Cheung, C. L.; Raghwani, J.; Bhatt, S. Nature 2009, 459, 1122−1125. (29) Garten, R. J.; Davis, C. T.; Russell, C. A.; Shu, B.; Lindstrom, S.; Balish, A.; Sessions, W. M.; Xu, X.; Skepner, E.; Deyde, V. Science 2009, 325, 197−201. (30) Gorman, O. T.; Bean, W. J.; Kawaoka, Y.; Donatelli, I.; Guo, Y. J.; Webster, R. G. J. Virol. 1991, 65, 3704−3714. (31) Peiris, J. S.; Poon, L. L.; Guan, Y. J. Clin. Virol. 2009, 45, 169− 173. (32) Pensaert, M.; Ottis, K.; Vandeputte, J.; Kaplan, M. M.; Bachmann, P. Bull. W. H. O. 1981, 59, 75−78. (33) Tumpey, T. M.; García-Sastre, A.; Taubenberger, J. K.; Palese, P.; Swayne, D. E.; Basler, C. F. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 3166−3171. (34) Nelson, M. I.; Viboud, C.; Simonsen, L.; Bennett, R. T.; Griesemer, S. B.; St. George, K.; Taylor, J.; Spiro, D. J.; Sengamalay, N. A.; Ghedin, E.; Taubenberger, J. K.; Holmes, E. C. PLoS Pathog. 2008, 4, No. e1000012. (35) Zimmer, S. M.; Burke, D. S. New Engl. J. Med. 2009, 361, 279− 285. (36) Li, K.; Guan, Y.; Wang, J.; Smith, G.; Xu, K.; Duan, L.; Rahardjo, A.; Puthavathana, P.; Buranathai, C.; Nguyen, T. Nature 2004, 430, 209−213. (37) Claas, E. C.; Osterhaus, A. D.; Van Beek, R.; De Jong, J. C.; Rimmelzwaan, G. F.; Senne, D. A.; Krauss, S.; Shortridge, K. F.; Webster, R. G. Lancet 1998, 351, 472−477. (38) Banks, J.; Speidel, E.; Moore, E.; Plowright, L.; Piccirillo, A.; Capua, I.; Cordioli, P.; Fioretti, A.; Alexander, D. Arch. Virol. 2001, 146, 963−973. (39) Suarez, D. L.; Senne, D. A.; Banks, J.; Brown, I. H.; Essen, S. C.; Lee, C.-W.; Manvell, R. J.; Mathieu-Benson, C.; Moreno, V.; Pedersen, J. C. Emerging Infect. Dis. 2004, 10, 693−699. (40) Peiris, M.; Yuen, K.; Leung, C.; Chan, K.; Ip, P.; Lai, R.; Orr, W.; Shortridge, K. Lancet 1999, 354, 916−917. (41) Suarez, D. L. Vet. Microbiol. 2000, 74, 15−27. (42) Xu, J.; Davis, C. T.; Christman, M. C.; Rivailler, P.; Zhong, H.; Donis, R. O.; Lu, G. PLoS One 2012, 7, No. e38665. (43) Smith, G. J.; Bahl, J.; Vijaykrishna, D.; Zhang, J.; Poon, L. L.; Chen, H.; Webster, R. G.; Peiris, J. M.; Guan, Y. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 11709−11712. (44) Webster, R. G.; Bean, W. J.; Gorman, O. T.; Chambers, T. M.; Kawaoka, Y. Microbiol. Rev. 1992, 56, 152−179. (45) Wagner, R.; Matrosovich, M.; Klenk, H. D. Rev. Med. Virol. 2002, 12, 159−166. (46) Ginting, T. E.; Shinya, K.; Kyan, Y.; Makino, A.; Matsumoto, N.; Kaneda, S.; Kawaoka, Y. J. Virol. 2012, 86, 121−127. 637

dx.doi.org/10.1021/ac402892m | Anal. Chem. 2014, 86, 629−637