INTRODUCTION
Influenza or flu is caused by a group of viruses called influenza viruses.
These categories of A, B and C viruses are the common pathological agents (Suzuki
and Nei, 2002). Genomes of these viruses are segmented, single stranded
and have (-) RNA. They are associated with epidemics and pandemics in mammals
and birds. Wild waterfowl and other aquatic birds are the natural reservoirs
of influenza viruses (Holmes et al., 2005). Influenza
epidemics are accountable for causing 10,000-15,000 deaths in humans every year
(Holmes et al., 2005).
Swine influenza also called swine flu is caused by a strain of influenza virus
named H1N1 that usually infect pigs. Pandemic influenza has created havoc in
1918 (HIN1), 1957 (H2N2) and 1962 (H3N2) resulting in numerous deaths worldwide
(Cox and Subbarao, 2000; Webby and
Webster, 2003), while H5N1 assumed epidemic proportions in Asia in the years
2003-2005 (Holmes et al., 2005). Very recently,
outbreaks of swine flu have sent shock waves in Mexico and United States, with
the World Health Organization (WHO) issuing warning for possibility of worldwide
pandemic. Although, the origin of this strain is still unknown, some reports
point out that it has not been found in pigs. WHO reported (www.who.int/mediacentre/news/statements/2009/H1N1-20090427)
that the mutated form of the virus might have been transmitted between humans
and causes symptoms of influenza, such as runny nose, fever, coughing and headache
etc. The H1N1 form of swine flu is reported to be a form of the causative agent
that caused the pandemic in humans in 1918-1919 (Taubenberger
and Morens, 2006). The descendants of the 1918 H1N1 virus have persisted
among humans as well as pigs throughout the 20th century, with some seasonal
bouts of influenza (Taubenberger and Morens, 2006).
New variants of influenza viruses arising out of reassortment of the segmented
RNA genome pose severe threat to public health (Gog et
al., 2007). The 2009 swine flu strain of influenza is reported (www.inspection.gc.ca/english/
corpaffr/newcom/2009) to be a reassortment of four strains that includes
influenza A virus subtype H1N1, one endemic in humans, one in birds and two
in swine.
The availability of the sequences of the segments of some of the latest strains of H1N1 virus has given an opportunity to look into the pattern of codon usage, gene expression levels, determine protein isoelectric points, aromaticity and hydropathicity indicates the solubility of the proteins: positive GRAVY (hydrophobic), negative GRAVY (hydrophilic) of the amino-acids in addition to studying their molecular phylogeny.
Synonymous codons are unequally used between genomes. Compositional bias, translational
selection and mutational pressure account for codon usage variation amongst
organisms (Sur et al., 2008). Highly expressed
genes have tendency for codons with high concentration of related tRNA molecules,
whereas low expressed ones have consistent codon usage (Zhou
et al., 2005). It has been reported that mutational pressure plays
a significant role in influencing codon patterns in human viruses (Zhou
et al., 2005). Estimation of the CAI (codon adaptation index) values
(Sen et al., 2008) indicates the nature of gene
expression levels of the respective genes. The physical properties of the proteins
are fundamental in the normal functioning of an organism (Knight
et al., 2004). Properties such as isoelectric point, hydropathicity
and aromaticity play a role in protein functioning. Environment and GC content
is known to play a crucial part in influencing amino-acid usage in organisms
(Tekaia and Yeramian, 2006). The correlation of GC3
and GC content of the organisms with isoelectric point and amino acid frequencies
of the proteins is expected to throw light into the molecular nature of swine
flu viruses. The results obtained for the latest strains will be compared with
that of the older strains of H1N1 and other flu viruses like H5N1, H2N2, H9N2,
H3N2, influenza B and influenza C virus.
On the other hand phylogeny study of swine flu and other related viruses will
shed some light on their evolutionary relationship. An important method of phylogeny
developed by a group of laboratories including ours use nucleotide triplet based
condensed matrix method (Mondol et al., 2008).
Phylogenetic studies using sequence alignment and structures are insufficient
in portraying the evolution of genes given that sequence comparison becomes
unreliable at identity levels lower than 25% (Mondol et
al., 2008). It also turns out to be tough to distinguish among properly
aligned homologs and discrete sequences. Structure based methodologies are also
insufficient given that number of structures are scarce to represent any significant
conclusion. The condensed matrix method that relies on nucleotide triplet based
phylogeny, is free from the aforesaid limitations as it takes into account,
full length of the genes for creating phylograms (Mondol
et al., 2008).
In a nutshell the aim of the present study is to look into the important characteristics of the genes and proteins of the swine flu and related viruses to infer upon their lifestyle and evolutionary relationships.
MATERIALS AND METHODS
The research work was started in the spring of 2009. It was virtually done in two laboratories. The software was developed in Department of Chemistry, Raiganj College. All bioinformatics analysis were performed at NBU Bioinformatics Facility, NBU while interpretation of results and paper writing were done in both the laboratories.
Sequences of the genome segments of Influenza A viruses [A/California/04/2009(H1N1);
A/California/05/2009 (H1N1); A/California/07/2009(H1N1); A/Texas/04/2009 (H1N1)
and A/Texas/05/2009(H1N1)] were obtained from the NCBI website (http://www.ncbi.nlm.nih.gov)
and that of other Influenza A viruses like [A/Goose/Guangdong/ 1/96(H5N1); A/Korea/426/68
(H2N2); A/Hong Kong/1073/99(H9N2); A/New York/392/ 2004(H3N2); A/Puerto Rico/8/34(H1N1)]
and Influenza B virus and Influenza C virus were obtained from the IMG database
(http://img.jgi.doe.gov/cgi-bin/pub/main.cgi)
(Markowitz et al., 2006).
The ACUA (Umashankar et al., 2007) was utilized
to compute GC content, GC3 composition (amount of G or C codons in the third
position), Nc (Effective number of codons) and CAI (codon adaptation index)
values. The Nc measures codon bias whose value ranges from 20 to 61 (Sur
et al., 2009). The CAI computes the relative adaptation of codon
usage of genes towards codon usage of highly expressed genes (Wu
et al., 2005). The CAI values vary from 0 to 1 with higher values
signifying that, gene of concern has a codon usage pattern analogous to highly
expressed genes. Certain viruses like bacteriophages have their own tRNA and
it is inferred that though phages use most of the cells translational machinery
and complement it with their own genetic information to attain higher fitness
(Bailly-Bechet et al., 2007). However, in swine
flu viral genomes there is no gene coding for any tRNA assuming that the swine
flu viruses entirely depend upon host cells translational machinery. Therefore,
we have used codon usage table of Homo sapiens as a reference for determining
CAI values. Hydropathicity (GRAVY score) and aromaticity of the genes were determined
using Codon W (http://mobyle.pasteur.fr/cgi-bin/MobylePortal/portal.py?form=codonw)
(Peden, 1999). Hydropathicity or GRAVY score is calculated
as the arithmetic mean of the sum of hydropathic indices of each amino acid,
whereas aromaticity determines amino acid usage, provided inequality in amino
acid composition includes application for evaluating codon usage (Lobry
and Gautier, 1994). Distribution of isoelectric points (pI) in a proteome
is one of the most important aspects of proteins (Kiraga
et al., 2007). Protein isoelectric points were calculated using DAMBE
(http://dambe.bio.uottawa.ca).
Correspondence analysis was computed for codon usage on codon count using Codon
W
(http://mobyle.pasteur.fr/cgi-bin/MobylePortal/portal.py?form=codonw)
(Peden, 1999). Major trends in codon usage variation among
the genes within the genomes are estimated with this analysis.
Determination of frequency of triplets of nucleic acid bases: Our own
program written in Turbo C++ was used to count all the possible triplets of
the nucleotide sequences of the whole genomes from the studied viruses. The
matrices were formed using all the triplets. Introduction of a 4x4x4 cubic matrix
having 64 possible entries resolves the frequency of incidence of all the possible
64 triplets in a DNA sequence. Here, it is possible to obtain three groups of
4x4 matrices {M1, M2, M3, M4}, {M5,
M6, M7, M8}, {M9, M10,
M11, M12} each one having every entry of cubic matrix
(Randic et al., 2001). Usually, the group of
4x4 matrices {M1, M2, M3, M4} are
taken as representative of the cubic matrix. Our technique gives equal weight
to all positions since it considers nucleotide triplets, not only starting from
1st codon position but all the three positions. Thus, addition or deletion of
bases taking place during the course of evolution is taken care of. The methodology
depicts DNA by condensed matrix counting the rate of presence of adjoining base
pairs (Randic, 2000).
Calculation of eigen value and construction of phylogram: Leading eigenvalues
were calculated using MATLAB (version 5.0.0.4069) software. Eigenvalues are
a special set of scalars associated with a linear system of equations, usually
matrix equations that are often regarded as characteristic roots, characteristic
values and proper values or latent roots (Mondol et al.,
2008). Evaluation of DNA sequences for similarity or dissimilarity is normally
aided by the convenience of leading eigenvectors calculated by this method.
Diversity between eigenvalues was used to study sequence similarity/dissimilarity
keeping in mind the characterization of a sequence by leading eigenvalue (Nandy
et al., 2006). Matrices linked to each sequence are estimated and
the leading eigen values computed. Variations in leading eigen values concurrent
to the string are estimated and the relationships between genes investigated.
Distance matrixes of the studied sequences were constructed by summing up the
square of the difference of eigen values. Phylograms were built by cluster analysis
of the similarity matrix using PHYLIP (Ver 3.65) and drawn with PHYLODRAW (Ver
0.8).
RESULTS AND DISCUSSION
It is seen from the values depicted in Table 1 that the H1N1
viruses, influenza B and C viruses as well as other avian flu viruses are poor
GC content genomes with GC values hovering around the 44% mark. This characteristic
has been previously reported (Zhou et al., 2005)
for H5N1 viruses. Subsequently, their GC3 content is also poor. The GC3 content,
which is regarded as an important parameter in studying codon usage variation
(Sen et al., 2008), reveals homogeneity in present
study. There is very little difference in the values of GC and GC3 content in
the individual genes of the genomes (data not shown). Influenza B and C viruses
have comparatively poorer GC and GC3 content with respect to H1N1 viruses and
avian flu viruses. The Nc values representing the effective number of codons
in a gene are quite high. However, the Nc values of the HIN1 CAL/04/2009 strain
are much lower compared to other H1N1 strains. Influenza C and B viruses too
have lower Nc values compared to other viruses undertaken in this study. The
CAI values representing the expression level of the genes are quite high in
all the studied genomes with H5N1 goose virus having the highest CAI value of
0.726.
| Table 1: | Mean
and standard deviation values of various parameters for the studied viruses |
 |
Data
are expressed as Mean±SD |
Generally in predicted proteomes the major pI values are classified as belonging
to acidic cluster (pI less than 7.4), neutral cluster (pI between 7.4 and 8.1)
and basic cluster (pI greater than 8.1) (Nandi et al.,
2005). The pI values depicting the isoelectric points of the proteins showed
a bi-modal distribution in the studied viral proteomes. All the HIN1 strains
as well as Influenza C, H3N2 and H2N2 had a pI in the range greater than 8.1
with H1N1 CAL07/2009 having a pI value of 8.87. On the other hand H5N1, H9N2
and Influenza B had pI values in the neutral cluster. There is good deal of
variation in the pI values amongst the proteins in the organisms as exemplified
by the standard deviations.
Table 2 shows the correlations of the CAI, GC, GC3, GRAVY,
Aromaticity and pI values. The CAI values were correlated with GC and GC3 content
(Sen et al., 2008) for the viruses. It was found
that CAI showed positive correlations with GC content in all the strains while
CAI showed strong positive correlations with GC3 content in some of the strains
except H1N1 CAL04/2009, H1N1 CAL05/2009 and H1N1/TEX05/2009.GC3 showed strong
correlation with GRAVY representing hydropathicity in case of H1N1 TEX04/2009,
H5N1/Goose/Guangdong, H2N2 Korea, H9N2 Hong Kong, Influenza C and H3N2 New York.
When GC3 content was correlated with aromaticity values it was found that GC3
showed strong negative correlations with H1N1 CAL04/2009, H1N1 CAL05/2009, H1N1
CAL07/2009 and H1N1 TEX04/2009, respectively. The isoelectric point values were
correlated with GC3 and GC content of the viruses. It was noticed that GC3 content
had a strong negative correlation with TEX04/2009, H5N1 Goose/Guangdong and
positive correlations with H1N1 Puerto Rico, H3N2 New York, Influenza C, H9N2
Hong Kong, H3N2 Korea and H1N1 TEX05/2009. Insignificant correlations were found
for the other strains.
Correspondence analysis of codon count (CACC) (Peden, 1999)
was computed to infer upon the role of amino-acid compositions in codon usage
variations. CACC revealed two major axes of variation. Majority of the genes
remained scattered with the exception of Influenza B and C strains were they
remain clustered in the centre of the axes. Figure 1a-l
show the distribution of the genes along the two major axes of variation. The
first major axis of variation was correlated with GC3 content, CAI, Nc, aromaticity
and hydropathicity scores.
Figure 2 shows the phylogram constructed for the complete genomes of the studied viruses. The phylogram reveals a rooted tree, which shows two major clades; Clade A and Clade B which contain subclades. Most of the new strains of H1N1 lie together in the same clade with the exception of H1N1 CAL04/2009 that lies in a different clade. The older H1N1 Puerto Rico strain lies in a different clade but in the same major clade with H1N1 CAL04/2009. Among the other viruses Influenza C and Influenza B viruses lie in different clades. H2N2, H5N1, H3N2 and H9N2 lie in same clade. In Clade A it is observed that H1N1 CAL07/2009and H1N1TEX05/2009 co-segregate. Influenza C lie sister to H1N1 strains in Clade A. In Clade B, the Influenza B viruses lie near the H3N2 New York strain. To be specific, these viral strains have more or less similar root distances and remain co-segregated as evident from Fig. 2.
The results obtained for GC3 and GC imply that there is a degree of homogeneity
among the genes in the studied viral strains and they are AT rich. Interestingly,
earlier reports (Sen et al., 2008) revealed a
good deal of heterogeneity in GC rich genomes. Although Nc values varied among
the genes in the organisms, high mean Nc values implied low bias. Although,
the genomes are rich in AT content yet they are markedly biased. This feature
has been previously reported for a Rhizobium phage (Sur
et al., 2009). Low bias may be due to the high mutation rates and
no contribution from translational selection in influencing codon bias. Comparatively
lower Nc values of H1N1 CAL04/2009 and Influenza B and C viruses indicate that
they are to some extent different with respect to this feature from other viruses.
High CAI values for the studied viruses are quite expected, since all of them
are highly pathogenic and need to survive against host defences.
| Table 2: | Correlation
results between different indices and principal axis of correspondence
analysis on codon count |
 |
| IC: Inconsequential result |
|
| Fig. 1: |
Correspondence analyses of codon count for the studied viral
strains. Axis 1 corresponds to X axis and Axis 2 corresponds to Y axis.
(a) CAL 04/2009 (b) CAL 05/2009 (c) CAL 07/2009 (d) TEX 04/2009 (e) TEX
05/2009 (f) TEX 05/2009 (g) H5N1/Goose (h) H2N2/Korea (i) H9N1/Honkong (j)
Influenza B (k) Influenza C and (l) H1N1/Puertorico |
|
| Fig. 2: | Phylogram
of the whole genome segments for the studied viral strains. Numbers depict
the root distances |
Most of the genes including hemagglutinin, neuraminidase and structural genes
are essential for the survival and manifestation of diseases and the high expression
levels of these genes act as a strategy. It is very well known that these viruses
have a high rate of mutation in response to drug treatments and the outbreaks
that occur after a span of some years (Holmes et al.,
2005). The high expressivity of the genes associated with pathogenicity
may play a significant role in undergoing reassortment thus giving rise to seasonal
outbreaks.
It is shown from Fig. 1 of CACC that there is little difference in variation among the different viruses. However, the clustering of majority of these genes at the centre of the axes for influenza B and C viruses indicate that they may be conserved in nature.
Strong correlations of GC3 and GC content with CAI values point out that expression
levels play a significant role in synonymous codon bias in these viruses and
GC, GC3 too have an important role to play. Strong correlations of GC3 content
with Axis 1 in most of the new H1N1 strains and the two influenza viruses imply
that GC compositional content play a significant role in influencing codon bias
in these organisms. GC3 content has also been found to show strong negative
correlations with grand average hydropathy value (GRAVY) in most of the studied
viruses except some of the H1N1 strains. This implies that hydropathy levels
increases with the decrease of GC3. Aromaticity levels increase with the decrease
in GC3 content as exemplified by strong negative correlations with GC3 in some
of the strains. Bimodal distributions of pI values were observed for the viral
proteomes. The average pI values for the viruses reveal that the H1N1 strains
have more basic proteomes compared to the other viruses. GC3 content shows moderate
to low correlations with isoelectric points in the studied viral genomes. Present
findings for most of the correlations of GC3 content in the studied viruses
strongly support the concept of Kiraga et al. (2007)
that low GC rich organisms code for more basic proteomes. However, there have
been some exceptions in case of Influenza B, H9N2 and H2N2. Negative correlations
of the GC3 content and pI may be attributed to the increase in basic lysine
encoded by the AT rich organisms while positive correlations are attributed
to the increase in basic arginine. This feature has been previously reported
(Kiraga et al., 2007). On the basis of Kiraga
et al. (2007) observation, present studied viral proteomes are either
basic or neutral, with most of the basic proteomes being influenced by mutational
pressure.
Strong correlations of the principal axis of correspondence analysis of Axis
1 with grand average hydropathy value (GRAVY) entails that genes associated
with the hydrophilic proteins are favoured by the translationally optimal codons.
Correlations of Axis 1 with aromaticity scores point out that aromaticity plays
an important role in influencing codon usage patterns in the studied viruses.
Similarly gene expression levels strongly influence codon usage bias as exemplified
by strong correlations with Axis 1. Negative correlations of the principal axis
of variation with Nc values may be attributed to the decrease in codon bias
among genes lying left of axis 1, while positive correlations indicate the reverse.
The phylogenetic pattern obtained for the studied viruses using the condensed
matrix method point out very clearly that reassortment has a part to play in
influencing evolution of these viruses. Clade A and B contained viruses taken
on a global basis. Although, most of the H1N1 strains are present together in
a single clade two other are placed in a different clade altogether. It is also
observed that lineages of one virus are occurring amongst lineages of other
viruses. Present results for the whole genome phylogeny reveal that the viral
segments being subjected to re-assortments are obtained from various lineages.
In this respect our findings support the results of Holmes
et al. (2005) that these flu viruses co-circulate, endure and re-assort
time-to-time depending upon the environment and susceptibility of the host.
The results obtained from the H1N1 strains commensurate with the aims and objectives
of the study. The evolutionary pattern of the new strains has been well explained
with the new methodology. The role of mutational pressure as the most important
force in guiding the codon usage patterns has been interpreted. Besides, other
properties like isoelectric point, aromaticity, hydropathicity, GC content has
been shown to influence the lifestyle of the viruses (Tekaia
and Yeramian, 2006).
Present results obtained from the condensed matrix based phylogenetic study revealed a slight difference from that obtained by previous studies with respect to the placements of the older H1N1 Puerto-Rico strain and one new strain H1N1 CAL04/2009 that lie in a different clade compared to other new H1N1 strains. Since our methodology focuses on the quantitative as well as qualitative characteristics of the DNA giving equal weightage to all codon positions; the role of mutations in reassortment of these viruses has been interpreted. The cladogram obtained in this study is correct because the result has been further corroborated with the data obtained from isoelectric point with basic proteomes being influenced by mutational pressure.
CONCLUSION
Present findings revealed that synonymous codon usage is less biased in H1N1
virus. Synonymous codon usage study in genes encoded by different influenza
A viruses show that they are conserved and mutational bias was the main factor
that drives the codon usage variation among these viruses. Low bias may be attributed
to the high mutation rates and inability of the contribution of translational
selection. High expression of pathogenicity related genes confirm its role as
potentially dangerous pathogen. Present studied viral proteomes are either basic
or neutral, with most of the H1N1 basic proteomes
influenced by mutational pressure implying the role played by mutations in
influencing the nature of the viruses. Genes associated with the hydrophilic
proteins are favoured by the translationally optimal codons. Phylogenetic analysis
by the condensed matrix method portrays the role played by re-assortments in
controlling the evolution of the studied strains. While majority of the new
strains lie in the same clade, H1N1 CAL04/2009 lies in the other clade along
with H1N1 Puerto Rico. The phylogeny results reaffirm that flu viruses co-circulate
and mutate by reassortment depending upon the environment and susceptibility
of the host (Holmes et al., 2005).
ACKNOWLEDGMENTS
The authors thank Department of Biotechnology, Government of India for providing financial assistance in setting up Bioinformatics Faculty in the University of North Bengal. A.S. acknowledges the receipt of DBT Overseas Fellowship to University of New Hampshire, USA.
Subscribe Today
