Infection, Genetics and Evolution 25 (2014) 69–77

Contents lists available at ScienceDirect

Infection, Genetics and Evolution journal homepage:

Vaccination influences the evolution of classical swine fever virus Wei Ji, Dan-Dan Niu, Hong-Li Si, Nai-Zheng Ding, Cheng-Qiang He ⇑ College of Life Science, Shandong Normal University, China

a r t i c l e

i n f o

Article history: Received 13 December 2013 Received in revised form 2 April 2014 Accepted 9 April 2014 Available online 18 April 2014 Keywords: Classical swine fever virus Vaccination Evolution Recombination Selection pressure Codon usage bias

a b s t r a c t Classical swine fever is a serious, economically damaging disease caused by classical swine fever virus (CSFV). The CSFV is composed of two clades, according to phylogenetic estimates. Attenuated live vaccine such as HCLV, has been widely used to protect pigs from CSFV, but the influence of vaccination on the evolution of CSFV has not been studied. We conducted a systemic analysis of the impact of vaccination on the evolution of CSFV by comparing vaccine-related and non-vaccine-related CSFV groups. We found that vaccination may affect strain diversity and immune escape through recombination and point mutation. We also found that vaccination may influence the population dynamics, evolutionary rate and adaptive evolution of classical swine fever virus. Our evidence suggests that the vaccination might also change host adaptation through influencing codon usage of the virus in swine. These findings suggest that it is necessary to avoid excessive use of CSFV attenuated vaccines. Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction Classical swine fever virus (CSFV) is a single stranded, non-segmented, positive-sense RNA virus that belongs to the family Flaviviridae, genus Pestivirus (Francki et al., 1991). This lipid-enveloped RNA virus infects domestic and wild pigs worldwide, and causes major losses in stock farming. CSFV infection is associated with disseminated intravascular coagulation, thrombocytopenia and immunosuppression (Yin and Liu, 1997). The CSFV genome is approximately 12,300 nucleotides in length and comprises a long open reading frame (ORF) flanked by 50 and 30 untranslated regions (UTRs). The ORF is translated to a primary polyprotein and then subsequently processed into 12 mature proteins by cellular and viral proteases, including eight nonstructural proteins (Npro, P7, NS2, NS3, NS4A, NS4B, NS5A, NS5B) and four structural proteins (C, Erns, E1, E2; (Rice, 1996; Thiel et al., 1991). Classical swine fever (CSF) was first recognized in Tennessee, USA in 1810 (Agriculture, 2010). It was described in France in 1822 (Cole, 1962) and within a few years it was found all over the world, including the UK in 1864 (Agriculture, 2010), Japan in 1888 (Edwards et al., 2000), the Caribbean in 1930 (Cuba; (Edwards et al., 2000), and China in 1925 (Tu et al., 2001). Three CSFV genotypes have been recognized in 36 countries and are still in circulation (Food and Agriculture, 1990; Paton et al., 2000). Genotype1 consists of lentogenic and velogenic viruses and are mainly found in CSFV outbreaks in China (Zhou, 1980). The other ⇑ Corresponding author. Tel.: +86 531 86188690. E-mail address: [email protected] (C.-Q. He). 1567-1348/Ó 2014 Elsevier B.V. All rights reserved.

two genotypes are made up of virulent viruses and exist in other parts of the world. Vaccination has been widely used to protect pigs from CSFV. From 1940 to 1949, two attenuated strains (the ROVAC strain from the US and the SFA strain from England) were obtained by repeatedly adapting viruses to rabbits (Baker, 1946; Koprowski et al., 1946). In 1954, Lee and colleagues continued to adapt the strain ROVAC (250-generation passages) in rabbits and obtained a lapinized strain (LPC) after 800 generations (Lin Tc Fau - Shieh et al., 1974; Lin and Lee, 1981). In 1955, a stable lapinized strain was obtained by passaging the virulent Shimen strain in rabbits. This strain was named the Chinese vaccine strain (C-strain) or the hog cholera lapinized virus (HCLV) and has been widely used in mainland China since 1957. The HCLV strain has played a global role in controlling CSFV epidemic in domestic pigs (TC, 1980; Yin and Liu, 1997; Zhou, 1980). HCLV has been introduced to many other countries because it is believed to be safer and more effective than the other commercial vaccines (e.g. American ROVAC and British SFA; (Bognar K, 1963; Olah and Palatka, 1967). Many commercial vaccine strains have been derived from HCLV (TC, 1980), such as Pestiffa (France), SUVAC (Hungary), Lapest (Poland), SuiferinC (former East Germany), KR (former USSR), VADIMUN (USA), and Riems (Germany). By passage the virulent viruses in certain cells under low temperatures (29–30 °C), investigators also recovered some attenuated strains, for instance a Japanese guinea-pig exaltationnegative strain (GPE), derived from the virulent strain ALD (Sasahara, 1970; Sasahara et al., 1969) and a French cell culture adapted strain Thiverval, derived from virulent strain Alfort (Aynaud et al., 1970; Launais et al., 1978; Lunais M et al., 1974;


W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77

Lunais M, 1977; LunaisM, 1972). Currently, the routinely used vaccines are all live attenuated vaccines including C-strain, GPE, Thiverval and their derivative strains (Van Oirschot, 2003). All of these live attenuated vaccines belong to the genotype I group. The evolutionary impact of the CSFV vaccination has not been studied although it has successfully protected pigs. In this study, we investigate the role of vaccination in shaping the evolution of CSFV using available isolates. In particular, we focused on the evolutionary effects on HCLV. Our results demonstrate that vaccination can potentially result in novel CSFV forms through recombination between vaccine strains and wild strains and we provide evidence that the vaccine-related group experiences different mutation rates, population sizes, selection profiles and codon usage bias than the non-vaccine-related CSFV strains.

2. Results 2.1. Phylogenetic classification and origin of CSFV We recovered two monophyletic CSFV groups in the E2 gene tree (Figs. 2 and S1). One group included all of the attenuated

vaccine strains, named the vaccine-related group, and the other group is named the non-vaccine-related group. The BEAST analysis estimates that CSFV may have appeared in China in the 1920s (95% HPD: 1856–1970) and the two CSFV groups may have diverged in the early 1960s (Fig. 2). These results agree with observations of when CSFV originated in China (Tu et al., 2001; Yin and Liu, 1997; Zhou, 1980).

2.2. Attenuated CSFV vaccine can potentially influence CSFV evolution through recombination with wild virus Homologous recombination is an important evolutionary force and previous studies have found that homologous recombination can occur in CSFV (He et al., 2007). To know whether CSFV vaccines can recombine with circulating strains, we reanalyzed all available CSFV genomes in RDP4 with nine different algorithms (Martin et al., 2010). Five mosaic viruses were detected (7.8% of the 64 complete genomes) with two putative parents and p value less than 105 in each putative breakpoint. The five recombinants were HCLV strain, strain 39, SWH, Heb 52010 and strain ALD (Figs. 3 and S2–S5). Interestingly, the isolate associated with HCLV has a

Fig. 1. Schematic representation of cDNA clones to determine the nucleotide sequence of the C strain of CSFV.

Fig. 2. An MCC tree of the E2 gene of CSFV from China. The trees were scaled to time using the collection ages of some CSFV samples, the GTR + G6 substitution model and an uncorrelated exponential molecular clock. Median tMRCAs are shown for selected nodes with the posterior probability.

W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77


Fig. 3. Evidence of recombination analysis in CSFV using RDP and Simplot. (A) Recombination breakpoint map of five isolates detected by several different methods implemented in the RDP4 program. The nucleotide position of breakpoints in the full length genome is indicated. Each breakpoint is supported as significant (p < 103) by more than four methods. The nucleotide position of breakpoints in the concatenated genome is indicated. (B) Sequence similarity was compared between AF091507/HCLV/ China and the putative parents. AF091507/HCLV/China was used as the query and the nucleotide position of breakpoints is given. Each breakpoint is supported as significant (v2 = 54.091, p-value < 103; v2 = 119.296, p-value < 103). The sliding window is 450 bp wide with a step size of 30 bp. (C) Bootscan result of AF091507/HCLV/China and its putative parent sequences. The isolate AY646427 was used as the outgroup. The y axis displays the percentage of permutated trees using a 450 bp sliding window, with a step size of 30 bp. (D) Topological incongruence in phylogenetic trees reconstructed using (i) positions 1 to 2110, (ii) positions 2111–2526, and (iii) positions 2527–11697. The mosaic isolate and its parent are marked with triangles in lime-green, red, blue. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

mosaic E2 gene, a product of a recombination between C-strain and a wild strain. 2.3. Vaccination may influence evolutionary rates and population dynamics of CSFV We estimated the evolutionary rate and population dynamics using a coalescent Bayesian skyline approach (Drummond et al., 2002, 2012). This analysis was based on full length genomes of all available isolates, except the recombinant viruses. According to a Bayesian analysis, the tMRCA of each gene and the complete genome was estimated to be between 1583 and 1859, with an average value 1771 (Fig. 4A). The evolutionary rate of each gene and complete genome was estimated to be between 3.17E4– 1.16E3 substitution/site/year (Fig. 4B). A Bayesian skyline plot (BSP) was used to estimate how effective populations changed through time using Erns (n = 107) and E2 (n = 188) (Minin et al., 2008). The two genes maintained a constant effective population until the late 1980s. Between 1980 and 1992 there was a steady decrease in the Erns estimates. E2 and Erns estimates exhibit an abrupt decline in the population from 2005 until now (Fig. 6).

To determine if vaccine and non-vaccine related groups exhibit different evolutionary rates and population dynamics, we repeated the analysis based on the dated full length E2 gene from China and the world, respectively, since available E2 can provide a larger dataset for this analysis. The tMRCA of the vaccine and non-vaccine-related groups were estimated to be around 1879 and 1829, respectively (Table S3) and the tMRCA of CSFV from China were 1963 and 1964, respectively (Fig. 1). Interestingly, the mutation rates of vaccine-related groups are lower than in non-vaccine-related groups (Fig. 5, Table S3). BSP analyses demonstrated that each group had a unique population dynamic (Fig. 7A). Vaccine and non-vaccine-related genotypes from around the world maintained constant effective population sizes until 2000. In 2005, the vaccine-related genotypes exhibited an abrupt decline in the population, with recovery from this event in early 2008, while the non-vaccine-related group showed an abrupt decline in the population after 2006. Then we analyzed the duplicate datasets, based on the dated full length E2 gene of CSFV isolated in China (Fig. 7B). The dynamics of vaccine and non-vaccine-related groups are inconsistent with the E2 gene from around the world. The population dynamics of the


W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77

Fig. 4. tMRCA and evolutionary rate were analyzed in Beast. (A) The tMRCA of different gene alignments was analyzed in Beast. (B) Relative rate of all genes was analyzed. The data are median values with the 95% HPD, indicated by error bars with the best model, calculated by Bayes factors.

Fig. 5. Relative rate of the E2 gene analyzed in Beast. (A) Relative rate of E2 genes from the world, illustrating the difference between the vaccine and non-vaccine related groups. (B) Relative rate of E2 genes from China, representing the difference in mutation rates in the vaccine and non-vaccine groups. The data are median rates with a 95% HPD, indicated by error bars. The best estimation was calculated with Bayes factors.

non-vaccine-related group slowly increased before 2003, suddenly expanded between 2003 and 2004, decreased after 2004, and abruptly declined after 2008. In contrast, the vaccine-related group declined before 2008, but since then has steadily increased. 2.4. Selection profiles We compared the selection profiles of protein coding genes for vaccine and non-vaccine-related groups (Table 1). There were no positions under positive selection in C, P7, NS4A, or NS4B genes in the two groups. However, selection profiles of the Npro, Erns, E1, E2, NS2, NS3, NS5A, and NS5B genes were different in the vac-

cine and non-vaccine related groups. There were no shared sites under positive selection between the two groups. The Erns gene had four codons under positive selection in the vaccine-related group and only one in the non-vaccine-related group (Table 1). The E2 gene had three codons under positive selection in the vaccine-related group and two in the non-vaccine-related group. The E1 gene had one codon under positive selection in the vaccinerelated group and none in the non-vaccine-related group. In the non-vaccine-related group, there was one codon under positive selection in the Npro, NS2, NS5A, and NS5B genes. Thus, selection pressure of vaccination is focused on the three surface proteins (Erns, E1, E2).

W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77


Fig. 6. Bayesian skyline plots (BSP) depicting the effective population size of CSFV through time. Effective population size estimates are expressed in logarithmic scale on the y axis. The median estimates and 95% HPD intervals of the estimates are denoted with solid lines and dotted lines. The light yellow region indicates a population decline. (A) BSP based on E2 and Erns gene sequences from around the world. The figure highlights the interval from 1965 to 2012 and the entire estimate is displayed in the inset. (B) BSP based on E2 and Erns gene sequences from China.

Fig. 7. Bayesian skyline plots (BSP) depicting the effective population size of vaccine and non-vaccine related genotypes of CSFV over time. Effective population size estimates were expressed in logarithmic scale on the y axis. The median estimates and 95% HPD intervals of the estimates are denoted as solid lines and dotted lines. (A) BSP of vaccine and non-vaccine related genotypes based on E2 sequences from around the world. The large graph displays the interval between 1980 and 2012 and the entire estimate is displayed in the inset. (B) BSP of vaccine and non-vaccine related genotypes based on E2 sequences from China.

2.5. Codon usage bias of each gene Based on ENc values and the corresponding relative distribution of GC3%, the great majority of the points are under the theoretical curve (Fig. 8) (Wright, 1990). Vaccine-related and non-vaccinerelated groups of each gene and complete genomes show distinct distributions in the Nc plot (Figs. 8 and S6), suggesting different selection pressures on the two groups. In particular, the complete genomes of the vaccine and non-vaccine-related groups endure different translational selection (Fig. 9). Moreover, the vaccinerelated group has a higher Nc value (51.7804 ± 0.17692) and a lower Fop value (0.3938 ± 0.00295) than the non-vaccine-related groups (Nc, 51.6945 ± 0.51180; Fop, 0.4018 ± 0.00243). This result is in agreement with the idea that genes with a high frequency of optimal codons have low ENc values (Peden, 2000).

3. Discussion Our study investigated how vaccination has shaped the evolutionary history of CSFV. Recombination is a rapid evolutionary

force in positive strand RNA viruses and traces of homologous recombination have been found in CSFV (He et al., 2007). In this study we found recombinant strains derived from a vaccine strain and a wild strain (Fig. 3). This indicates that swine can be simultaneously infected with the live vaccine virus and other circulating wild types. Recombination between the wild type virus and vaccine strains is not unique to CSFV, vaccine recombinants of rabies virus (associated with fatal neuron disease; (He et al., 2012), poliovirus (associated with paralytic poliomyelitis; (Guillot et al., 2000) and bursal disease virus (Hon et al., 2008) have all been previously reported. We are concerned that vaccination may facilitate the emergence of new strains through recombination with circulating viruses. These new strains are phenotypically unpredictable and may pose greater dangers to swine health. Our estimate of the tMRCA of the CSFV is 1583–1859, which is consistent with the first record of CSFV outbreak in Tennessee in 1810 (Agriculture, 2010). We estimated that the Chinese strains of CSFV first arose in 1926, which is in agreement with the first recorded observation in 1925 (Tu et al., 2001). The origin of Chinese strains of HCLV was estimated to be 1963, which is in line with the first manufacture of the vaccine in 1954 (Tu et al., 2001).


W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77

Table 1 Site-specific selection analysis for each coding gene of vaccine and non-vaccine related groups. Gene

Total number of condens

Npro C Erns E1 E2 P7 NS2 NS3 NS4A NS4B NS5A NS5B

168 99 227 195 373 70 457 683 64 347 497 718

Global dN/dS€ (Conden position§) Vaccine-related group

Nonvaccine-related group

0.225151 (None) 0.189962 (None) 0.159732 (107*,q119q209*,q217q) 0.223594 (80*,w) 0.256627 (34*49*198q) 0.092 (None) 0.076 (None) 0.066 (481q673q) 0.051 (None) 0.074 (None) 0.076 (None) 0.073 (None)

0.265515 (27*) 0.250961 (None) 0.372508 (52*) 0.249775 (None) 0.429386 (197w283*) 0.120 (None) 0.169 (410**,w) 0.123 (None) 0.138 (None) 0.163 (None) 0.154 (306*) 0.137 (247*)

Strains used for selection analyses in both groups are mentioned in Table S1. Position of codon under positive selection with statistical supporting. Significant (P < 0.05) in FEL analysis. ** Significant (P < 0.01) in FEL analysis. q Significant (P < 0.05) in PAML analysis. w Significant (P < 0.01) in PAML analysis. € Rate of non-synonymous (dN) to synonymous (dS) substitution. § *

Fig. 8. Nc plots of CSFV genes. The continuous curve in each of the figures represents the relationship between Nc and GC3s (under H0, Nc = 2 + s+{29/[s2 + (1s)2]}). Ovals in dashed and solid lines represent vaccine and non-vaccine related groups, respectively. (A) The vaccine and non-vaccine groups of E2 gene from around the world are shown in green and red, respectively. (B) The vaccine and non-vaccine groups of the Erns gene from around the world are shown in cyan and blue, respectively. (C) The vaccine and non-vaccine groups of concatenated genes from around the world are shown in yellow and black, respectively. (D) The vaccine and non-vaccine groups of E2 genes from China are shown in red and cyan, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Bayesian skyline plots suggested that the demographic history of CSFV experienced an abrupt decline in population from 2005 to 2012. Although the factors responsible for such population changes are not fully known possible explanations could be the slaughter of pigs during the outbreak of blue ear disease in 2006–2007 (Tong et al., 2007; Zhou et al., 2008) or during bouts of swine influenza (H1N1 in 2008–2009, H3N2 in 2007–2008; (Belongia et al., 2010; Pasma and Joseph, 2010; Seef S Fau Jeppsson and Jeppsson, 2013). Compared to the non-vaccine-related group, the CSFV vaccinerelated group did not seem to be influenced by 2009 swine

population declines in China. Before 2005, the population of the vaccine-related group decreased, while the numbers of the nonvaccine-related group were on the rise. Both groups decreased in population between 2005 and 2008, which may be related to the swine slaughter in 2006 and 2007 in China (Barboza, 2007; Tian et al., 2007). After 2008, populations of the non-vaccine-related group continued to decrease, while the vaccine-related group increased. This may be due to vaccine strains continually introduced to susceptible populations as modified live vaccine. From the BSP analysis of CSFV in China before 2005, we suspect that mutations were accumulating in the vaccine-related group due

W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77


Fig. 9. Correspondence of CSFV genomes with their CAI values. Scatter plot of the position of CSFV genomes on the first axis generated by correspondence analysis against their CAI values. The genomes are represented by circles. (A) Scatter plot of non-vaccine related groups with r = 0.535, p < 0.01. (B) Scatter plot of vaccine related groups with r = 0.453, p < 0.05.

Fig. 10. Q-mode cluster analysis of the codon usage pattern in the CSFV vaccine related group, non-vaccine related group and swine. Codon usage is displayed as comparisons between the vaccine related group and the non-vaccine-related group (orange), the non-vaccine-related group and swine (blue), and the vaccine related group and swine (green). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

to the pressure of vaccination and therefore, in the future, HCLV may lose the function of protecting pigs from all CSFV strains. The evolutionary rate of each CSFV gene is different. Structural proteins have faster rates of change than nonstructural proteins and this may result from selective pressure exerted by the immune response (Hulst et al., 1993; Kwang et al., 1992; Wensvoort et al., 1990). Vaccine-related groups have slower evolutionary rates than non-vaccine-related groups and we infer that this is the reason why HCLV has maintained a stable virulence, after evolving from the Shimen strain (TC, 1980; Yin and Liu, 1997; Zhou, 1980). The global rate of non-synonymous to synonymous substitution (dN/ dS), x, for all protein coding genes in both groups was less than 1, inferring that purifying selection has been the major driving force in the evolution of CSFV. Nevertheless, positive selection has played a major role in the evolution of certain codons. In Erns, the positive selection site (209S ? R) was associated with SK6infected cells changing from a heparan sulfate-independent pathway to a heparan sulfate-dependent pathway (Hulst et al., 2000). In E2, the sites under positive selection (34, 49) were consistent with what has been found in previous studies (Pérez et al., 2012; Shen et al., 2011; Tang et al., 2008). The surface structural proteins of the vaccine-related groups contain more positive sites than other proteins of the vaccine-related groups and all proteins of the non-vaccine-related groups, suggesting influence from immune selection (Hulst et al., 1993; Kwang et al., 1992; Leifer et al., 2013; Wensvoort et al., 1990). Additionally, the Nc-plot is an effective strategy to analyze the patterns of synonymous codon usage by comparing the actual distribution of genes with the expected distribution under no selection (Wright, 1990). When codon choice is determined only by GC3 composition, the point will fall on or just below the curve of the expected values (He et al., 2013). Therefore, the great majority

of the points are well below the curve and codon usage variation in the CSFV genes seem to be influenced by pressures besides mutational bias. The vaccine and non-vaccine-related groups have different Nc-GC3 distributions. Previous studies have shown that CSFV has evolved as two separate clades and this is supported by their codon pair usage (Leifer et al., 2011; Tao et al., 2009). We hypothesize that divergence of the two clusters may be caused by selection pressure exerted by vaccination. Since viruses are parasitic organisms, their codon usage pattern would be subject to their host to some extent (Zhou et al., 2012). The RSCU of CSFVs and pigs have a similar synonymous codon usage bias (Table S4). In particular, the synonymous codon usage bias of the non-vaccine-related group is more similar to that of swine genes (Fig. 10), indicating the adaptation of non-vaccine-related group, which has fewer positive selection profiles Consequently, recombination can occur between the vaccine and the non-vaccine related groups, even though strains in these groups evolve at different rates, exhibit different evolutionary histories, have different selection profiles, and have different synonymous codon usage bias. These features suggest that the HCLV vaccine might lose its function of protecting pigs, and vaccination will boost the occurrence of new strains. And thus, it is necessary to avoid excessive use of CSFV attenuated vaccines and develop new vaccine to non-vaccine-related CSFV group. 4. Methods 4.1. Virus sequencing Classical swine fever virus C-strain was provided by the Veterinary Medicine Research Institute of China (Beijing). PK15 Cells were grown in Dulbecco’s modified eagle medium, supplemented


W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77

with 10% fetal calf serum and maintained in a 5% CO2 humidified incubator at 37 °C. Viruses were inoculated in PK15 cells and cultured for 3 days. Viral RNA was isolated and reverse transcribed into cDNA. The cDNA was used for a PCR assay with universal primers (Table 1). PCR products were identified with 1% agarose gel electrophoresis. PCR amplicons of the correct size were excised from the gel. The full length genome was cut into seven parts (Fig. 1) and cloned into a pMD18T-vector for sequencing. Positive plasmids were sent to the Beijing Genomics Institute for sequencing. Sequencing results were deposited with NCBI (Genbank accession number: AY382481).

into vaccine-related and non-vaccine-related viruses (Table S5). We used Fixed-Effect Likelihood (FEL) from and an ML approach implemented in CODEML (PAML package version 3.15) to investigate positive selection (Nielsen and Yang, 1998; Yang, 1998, 2007). Positive selection was supported by p-values < 0.05 in a FEL model. We also analyzed the genes in PAML, with a likelihood ratio test comparing M1a, M7 and M8a models that assume no positive selection (x < 1) with M2 and M8 models that assume positive selection (x > 1), leaving the naive Empirical Bayes (NEB) analysis result, ignoring Bayes Empirical Bayes (BEB) analysis result.

4.2. Sequence data collection and phylogenetic analyses

4.6. Synonymous codon usage analyses

Fully sequenced CSFV genomes (N = 64) and CSFV E2 gene sequences (N = 188) were retrieved from NCBI (Table S5). These sequences represent genotypic and geographic variation, as well as different isolation years. Sequences were aligned in MEGA version 5.1 (Tamura et al., 2011). A separate dataset was created for each coding gene (Npro, C, Erns, E1, E2, P7, NS2, NS3, NS4A, NS4B, NS5A and NS5B genes) using MEGA 5.1.

To estimate the synonymous codon usage of each gene and the full length genome of vaccine and non-vaccine related groups (Table S5), the effective number (Nc) and the GC frequency at the synonymous third codon position (GC3s) were calculated in Codon W1.4.4 (Liu et al., 2012; Wright, 1990). Nc values ranged from 20 (highly biased as only one codon is used for each amino acid) to 61 (all synonymous codons are used equally). The overall variation in codon usage was reflected by the relative synonymous codon usage (RSCU). If the RSCU of one amino acid shows no bias, that is, the codon usage frequency is close to the expected frequency, and then the RSCU values of codons are equal to 1. If a codon has an RSCU value greater than 1, the codon use frequency is higher than the expected frequency and vice versa (Peden, 2000). The RSCU of vaccine and non-vaccine strains and the host swine were then calculated, using full genomes, E2 genes from China, E2 genes from other CSFV isolates (Table S5) and 2953 CDS (1,168,059 codons) of Susscrofa genes from the Codon Usage Database ( The relationship among these sequences was calculated using a squared euclidean distance (dik ¼ Rpj¼1 ðX ij  X kj Þ2 ) implemented in the software Q-mode Cluster Analysis (Nie et al., 1975).

4.3. Detecting recombinant sequences Predictions of recombination in the aligned genome sequences were explored using the RDP4 package with Bootscan, RDP, GENECONV, Maximum Chi, Chimaera, SiScan and 3Seq algorithm (Martin et al., 2010). The putative parental sequences and the breakpoint position were also estimated. Putative breakpoints were supported by five or more methods with p-values < 105. Simplot software was used to confirm all recombinants, through similarity plot analysis, Bootscan plot analysis, and the Findsite subprogram (Lole et al., 1999). Recombination breakpoints were considered statistically significant when p < 0.05 (Lancaster and Seneta, 1969). 4.4. Estimation of evolutionary rates and past epidemic history Twelve genes and full length genomes (Table S5) with available date were used to infer the evolutionary rate and divergence times using an MCMC approach implemented in BEAST version 1.7.1 (Drummond et al., 2012; Lemey et al., 2009). The E2 gene (188 sequences) and the Erns gene (104 sequences) were used to analyze the past population dynamics of the vaccine and non-vaccine related groups. Based on the cladogram of E2 sequences, the vaccine-related group (39 sequences) and the non-vaccine-related group (149 sequences) were chosen to infer the evolutionary rates and population dynamics. The evolutionary rate and dynamics were estimated with the HKY + G4 and GTR + G6 substitution models; a strict clock model and an uncorrelated lognormal distribution (UCLD). The MCMC computation was run for a total chain length of 2 ⁄ 108, sampled every 103 generations on average. The best fit model was determined by Bayesian factors calculated from their posterior distribution (Table S2). The coalescent Bayesian skyline plot (BSP) was used to estimate population dynamics. Population dynamics were analyzed using Tracer version 1.5 (http:// The maximum clade credibility (MCC) tree for the E2 gene (86 sequences) from China was calculated in Tree Annotator version 1.7.4 ( TreeAnnotator). 4.5. Selection analysis Selection analysis was conducted on all of the sequences, except the putative mosaic strains, because these can result in a false positive. Genome sequences and datasets for each gene were divided

Acknowledgements This study is supported by Ji’nan City College Independent Innovation Project (to H.C.Q). We thank Qi Hou for preparing graphics in Matlab software. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at 2014.04.008. References Agriculture, U., 2010. UK Agriculture livestock Classical swine fever. . Aynaud, J., Lejolly, J., Bibard, C., Galicher, C., 1970. Studies of the properties of cold induced classical swine fever virus mutants. Application to vaccination. Bull.Office Int. Epizooties 75, 654–659. Baker, J.A., 1946. Serial passage of hog cholera virus in rabbits, proceedings of the society for experimental biology and medicine. Society for experimental biology and medicine (New York, NY). R. Soc. Med., 183–187. Barboza, D., 2007. Virus spreading alarm and pig disease in China. N. Y. Times World Bus. Belongia, E.A., Irving, S.A., Waring, S.C., Coleman, L.A., Meece, J.K., Vandermause, M., Lindstrom, S., Kempf, D., Shay, D.K., 2010. Clinical characteristics and 30-day outcomes for influenza A 2009 (H1N1), 2008–2009 (H1N1), and 2007–2008 (H3N2) infections. JAMA: J. Am. Med. Assoc. 304, 1091–1098. Bognar, K., Meszaros, J., 1963. Experience with a lapinized hog cholera virus strain of decreased virulence 1963. Acta Vet. Acad. Sci. Hung. 13, 429–438. Cole, C.G., 1962. History of Hog Cholera Research in the US Department of Agriculture: 1884–1960. Agricultural Research Service, US Department of Agriculture.

W. Ji et al. / Infection, Genetics and Evolution 25 (2014) 69–77 Drummond, A.J., Nicholls, G.K., Rodrigo, A.G., Solomon, W., 2002. Estimating mutation parameters, population history and genealogy simultaneously from temporally spaced sequence data. Genetics 161, 1307–1320. Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. Edwards, S., Fukusho, A., Lefevre, P.-C., Lipowski, A., Pejsak, Z., Roehe, P., Westergaard, J., 2000. Classical swine fever: the global situation. Vet. Microbiol. 73, 103–119. Lin Tc Fau - Shieh, C.M., Shieh Cm Fau - Su, J.F., Su, J.F., 1974. Virus multiplication in pigs inoculated with lapinized hog cholera live vaccine. Chin. J. Microbiol. 7, 13– 19. Food, Agriculture, O, 1990. Animal health yearbook. 1989 (1990). Rome u.a. Francki, R.I.B., Fauquet, C., Knudson, D., Brown, F., 1991. Classification and nomenclature of viruses. Fifth report of the International Committee on Taxonomy of Viruses. Guillot, S., Caro, V., Cuervo, N., Korotkova, E., Combiescu, M., Persu, A., AubertCombiescu, A., Delpeyroux, F., Crainic, R., 2000. Natural genetic exchanges between vaccine and wild poliovirus strains in humans. J. Virol. 74, 8434– 8443. He, C.-Q., Ding, N.-Z., Chen, J.-G., Li, Y.-L., 2007. Evidence of natural recombination in classical swine fever virus. Virus Res. 126, 179–185. He, C.-Q., Meng, S.-L., Yan, H.-Y., Ding, N.-Z., He, H.-B., Yan, J.-X., Xu, G.-L., 2012. Isolation and identification of a novel rabies virus lineage in china with natural recombinant nucleoprotein gene. PLoS One 7, e49992. He, M., Ding, N.-Z., He, C.-Q., Yan, X.-C., Teng, C.-B., 2013. Dating the divergence of the infectious hematopoietic necrosis virus. Infect. Genet. Evol. Hon, C.-C., Lam, T.T.-Y., Yip, C.-W., Wong, R.T.-Y., Shi, M., Jiang, J., Zeng, F., Leung, F.C.-C., 2008. Phylogenetic evidence for homologous recombination within the family Birnaviridae. J. Gen. Virol. 89, 3156–3164. Hulst, M., Westra, D., Wensvoort, G., Moormann, R., 1993. Glycoprotein E1 of hog cholera virus expressed in insect cells protects swine from hog cholera. J. Virol. 67, 5435–5442. Hulst, M., Van Gennip, H., Moormann, R., 2000. Passage of classical swine fever virus in cultured swine kidney cells selects virus variants that bind to heparan sulfate due to a single amino acid change in envelope protein Erns. J. Virol. 74, 9553– 9561. Koprowski, H., James, T.R., Cox, H.R., 1946. Propagation of hog cholera virus in rabbits, proceedings of the society for experimental biology and medicine. Society for experimental biology and medicine (New York, NY). R. Soc. Med. Kwang, J., Littledike, E.T., Donis, R.O., Dubovi, E.J., 1992. Recombinant polypeptide from the gp48 region of the bovine viral diarrhea virus (BVDV) detects serum antibodies in vaccinated and infected cattle. Vet. Microbiol. 32, 281–292. Lancaster, H.O., Seneta, E., 1969. Chi-Square Distribution. Wiley Online Library. Launais, M., Aynaud, J., Corthier, G., 1978. Hog cholera virus: active immunization of piglets with the Thiverval strain in the presence and absence of colostral passive immunity. Vet. Microbiol. 3, 31–43. Leifer, I., Hoeper, D., Blome, S., Beer, M., Ruggli, N., 2011. Clustering of classical swine fever virus isolates by codon pair bias. BMC Res. Notes 4, 521. Leifer, I., Ruggli, N., Blome, S., 2013. Approaches to define the viral genetic basis of classical swine fever virus virulence. Virology 438, 51–55. Lemey, P., Salemi, M., Vandamme, A.-M., 2009. The Phylogenetic Handbook: A Practical Approach to Phylogenetic Analysis and Hypothesis Testing. Cambridge University Press Cambridge. Lin, T.T., Lee, R.C.T., 1981. An overall report on the development of a highly safe and potent lapinized hog cholera virus strain for hog cholera control in Taiwan. Natl. Sci. Counc. Liu, X., Zhang, Y., Fang, Y., Wang, Y., 2012. Patterns and influencing factor of synonymous codon usage in porcine circovirus. Virol. J. 9, 1–9. Lole, K.S., Bollinger, R.C., Paranjape, R.S., Gadkari, D., Kulkarni, S.S., Novak, N.G., Ingersoll, R., Sheppard, H.W., Ray, S.C., 1999. Full-length human immunodeficiency virus type 1 genomes from subtype C-infected seroconverters in India, with evidence of intersubtype recombination. J. Virol. 73, 152–160. Lunais M, A.J., 1977. Classical swine fever: duration of immunity obtained in the sow with‘‘Thiverval’’strain under field conditions. Recl. Med. Vet. 153, 349–352. Lunais M, A.J., Corthier, G., Laude, H., 1974. Swine fever: characteristics of immunity, innocuity for pregnant sows and geneticst ability of the‘‘Thiverval’’strain. RevMedVet(Toulouse) 125, 175–194. LunaisM, A., 1972. Classical swine fever: properties of a clone (strain‘‘Thiverval’’) isolated from a cellular culture at a low temperature: application in vaccination. RevMedVet(Toulouse) 123, 1537–1554. Martin, D.P., Lemey, P., Lott, M., Moulton, V., Posada, D., Lefeuvre, P., 2010. RDP3: a flexible and fast computer program for analyzing recombination. Bioinformatics 26, 2462–2463.


Minin, V.N., Bloomquist, E.W., Suchard, M.A., 2008. Smooth skyride through a rough skyline: bayesian coalescent-based inference of population dynamics. Mol. Biol. Evol. 25, 1459–1471. Nie, N.H., Bent, D.H., Hull, C.H., 1975. SPSS: Statistical Package for the Social Sciences. McGraw-Hill, New York. Nielsen, R., Yang, Z., 1998. Likelihood models for detecting positively selected amino acid sites and applications to the HIV-1 envelope gene. Genetics 148, 929–936. Olah, P., Palatka, Z., 1967. Immunobiological study of lapinized hog cholera virus strains. Acta Vet. Acad. Sci. H 17, 239. Pasma, T., Joseph, T., 2010. Pandemic (H1N1) 2009 infection in swine herds, Manitoba. Can. Emerg. Infect. Dis. 16, 706. Paton, D., McGoldrick, A., Greiser-Wilke, I., Parchariyanon, S., Song, J.-Y., Liou, P., Stadejek, T., Lowings, J., Björklund, H., Belak, S., 2000. Genetic typing of classical swine fever virus. Vet. Microbiol. 73, 137–157. Peden, J.F., 2000. Analysis of Codon Usage. Citeseer. Pérez, L.J., Díaz de Arce, H., Perera, C.L., Rosell, R., Frías, M.T., Percedo, M.I., Tarradas, J., Dominguez, P., Núñez, J.I., Ganges, L., 2012. Positive selection pressure on the B/C domains of the E2-gene of classical swine fever virus in endemic areas under C-strain vaccination. Infect. Genet. Evol. 12, 1405–1412. Rice, C., 1996. Flaviviridae: the viruses and their replication. Fields Virol. 3, 931– 959. Sasahara, J., 1970. Hog cholera: diagnosis and prophylaxis. Natl. Inst. Anim. Health Q. 10, 57–81. Sasahara, J., Kumagai, T., Shimizu, Y., Furuuchi, S., 1969. Field experiments of hog cholera live vaccine prepared in guinea-pig kidney cell culture. Natl. Inst. Anim. Health Q. 9, 83. Seef S Fau - Jeppsson, A., Jeppsson, A., 2013. Is it a policy crisis or it is a health crisis? The Egyptian context – analysis of the Egyptian health policy for the H1N1 flu pandemic control. Pan. Afr. Med. J. Shen, H., Pei, J., Bai, J., Zhao, M., Ju, C., Yi, L., Kang, Y., Zhang, X., Chen, L., Li, Y., 2011. Genetic diversity and positive selection analysis of classical swine fever virus isolates in south China. Virus Genes 43, 234–242. Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739. Tang, F., Pan, Z., Zhang, C., 2008. The selection pressure analysis of classical swine fever virus envelope protein genes Erns and E2. Virus Res. 131, 132–135. Tao, P., Dai, L., Luo, M., Tang, F., Tien, P., Pan, Z., 2009. Analysis of synonymous codon usage in classical swine fever virus. Virus Genes 38, 104–112. Tc, Z., 1980. Progress of classical swine fever virus and the prevention of the disease (Part II). Chin. J. Vet. Sci. Technol. 5, 30–39. Thiel, H., Stark, R., Weiland, E., Rümenapf, T., Meyers, G., 1991. Hog cholera virus: molecular composition of virions from a pestivirus. J. Virol. 65, 4705–4712. Tian, K., Yu, X., Zhao, T., Feng, Y., Cao, Z., Wang, C., Hu, Y., Chen, X., Hu, D., Tian, X., 2007. Emergence of fatal PRRSV variants: unparalleled outbreaks of atypical PRRS in China and molecular dissection of the unique hallmark. PLoS One 2, e526. Tong, G.-Z., Zhou, Y.-J., Hao, X.-F., Tian, Z.-J., An, T.-Q., Qiu, H.-J., 2007. Highly pathogenic porcine reproductive and respiratory syndrome. China Emerg. Infect. Dis. 13, 1434. Tu, C., Lu, Z., Li, H., Yu, X., Liu, X., Li, Y., Zhang, H., Yin, Z., 2001. Phylogenetic comparison of classical swine fever virus in China. Virus Res. 81, 29–37. Van Oirschot, J., 2003. Vaccinology of classical swine fever: from lab to field. Vet. Microbiol. 96, 367–384. Wensvoort, G., Boonstra, J., Bodzinga, B., 1990. Immunoaffinity purification and characterization of the envelope protein E1 of hog cholera virus. J. Gen. Virol. 71, 531–540. Wright, F., 1990. The ‘effective number of codons’ used in a gene. Gene 87, 23–29. Yang, Z., 1998. Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution. Mol. Biol. Evol. 15, 568–573. Yang, Z., 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol. Evol. 24, 1586–1591. Yin, Z., Liu, J.H., 1997. Animal Virology, 2nd edition. Science Publisher, Beijing, China. Zhou, T., 1980. Progress of classical swine fever virus and the prevention of the disease (Part I). Chin. J. Vet. Sci. Technol. 4, 23–33. Zhou, Y.J., Hao, X.F., Tian, Z.J., Tong, G.Z., Yoo, D., An, T.Q., Zhou, T., Li, G.X., Qiu, H.J., Wei, T.C., 2008. Highly virulent porcine reproductive and respiratory syndrome virus emerged in China. Transboundary Emerg. Dis. 55, 152–164. Zhou, J.-H., Gao, Z.-L., Zhang, J., Chen, H.-T., Pejsak, Z., Ma, L.-N., Ding, Y.-Z., Liu, Y.-S., 2012. Comparative the codon usage between the three main viruses in pestivirus genus and their natural susceptible livestock. Virus Genes 44, 475– 481.

Vaccination influences the evolution of classical swine fever virus.

Classical swine fever is a serious, economically damaging disease caused by classical swine fever virus (CSFV). The CSFV is composed of two clades, ac...
2MB Sizes 0 Downloads 5 Views