Phylodynamics of classical swine fever virus in Brazil

Abstract

The classical swine fever virus is the etiologic agent of one of the diseases with the greatest impact on swine farming worldwide. An extensive area of Brazil is considered free of the disease, but some states in Northeast Brazil have registered outbreaks since 2001. The objective of this study was to analyze the genetic variations of the virus and its spread over time and space. Partial sequences of the viral E2 protein obtained from samples collected during the Brazilian outbreaks were compared with sequences from the GenBank database (NCBI). The results demonstrated the continuous presence of the virus in the state of Ceará, with diffusion to at least two other states. The Brazilian Northeast virus presents specific polymorphisms that separate it from viruses isolated in other countries.

Introduction

Brazil is the world’s fourth-largest pork producer and exporter. It produced more than 3.9 million tons in 2018, about 3% of world production, and exported approximately 650 thousand tons (10% of the world total in volume) (Brazil, 2020). The size of the Brazilian pig herd and its economic importance raise concerns about its health status. One of the diseases of greatest concern is classical swine fever (CSF), whose etiologic agent is classical swine fever virus (CSFV), an RNA virus belonging to the family Flaviviridae, genus Pestivirus [1]. Brazil is considered disease-free in its main pork-producing areas; however, the virus persists in the northeast areas of the country, mainly within the context of subsistence of pig farming [2].

Since the middle of the last century, several official swine disease control and eradication programs have been implemented with the aim of raising the quality of swine production. After the success of the program for the eradication of African swine fever (ASF), the official Swine Pests Combat Program (SPCP) has turned to the control of CSF (or hog cholera) outbreaks. For this purpose, initially, the SPCP relied solely on vaccination. Unfortunately, vaccination alone could not eliminate the virus; consequently, many CSF outbreaks still occurred in areas throughout the country [3]. In 1992, a specific CSF control and eradication program were conceived in agreement with international rules to progressively combat CSF by avoiding vaccination. Initially, vaccination against CSF was prohibited in South Region states because swine production in this area of the country presented a high-level of technology and was carried out under reassuringly good sanitary and safety conditions. Then, gradually, vaccination was prohibited in all territories. In 2001, The CSF Eradication and Control Program (CSFECP) divided the country into CSF-Free Zone, where vaccination is not permitted, and CSF infected area. Currently, the CSF-Free Zone includes 15 states plus the Federal District (FD), with the federal capital Brasília. Concomitantly, Brazilian official laboratories have been investing substantially to improve swine disease diagnosis, looking at both the production for internal consumption and the production destined for export. The last CSF outbreak in the CSF-Free Zone occurred in 1997. Immediately, the stamping-out measures recommended by both the Office International des Epizooties (OIE) and the Ministry of Agriculture, Livestock and Supply were applied to all pig herds affected, regardless of preventive vaccination [3]. Brazil maintains CSF-free status in its main pork-producing areas; however, this has not precluded CSF outbreak recurrence in several Brazilian states outside of the CSF-free regions.

CSFV infections in domestic pigs persist in the North and Northeast Regions, mainly among smallholdings. From 2001 to 2009, 49 CSF outbreaks devastated seven states of non CSF-free areas. In 2001, 12 CSF-outbreaks were recorded: six in Pernambuco, three in Ceará (CE), two in Paraíba, and one in Rio Grande do Norte (RN). In 2003, four CSF outbreaks occurred in CE, while of the eight CSF outbreaks recorded in 2006, seven took place in CE and one in PB. Only one focus per year was reported, in 2007 (in CE) and in 2008 [in Maranhão (MA)]. Nevertheless, in 2009, 18 CSF outbreaks occurred: four in Amapá (AP) and two in Pará (PA) in the North Region, and as many as 12 of the total 18 CSF outbreaks occurred in RN in the Northeast Region, in the municipalities of Mossoró, Jucurutu, and Macaíba. No CSF outbreaks were registered from 2010 to 2018 [4]. Stamping out was the main sanitary measure adopted on these occasions, but in two federal states, AP and RN, emergency vaccination was also applied. In general, all these CSF outbreaks involved animals raised in subsistence pig farming which is a rustic model for rearing pigs, being carried out with low level of technology and taking place for the most part, in backyard pig holdings.

From 2018 to the end of 2019, 68 CSF outbreaks occurred in the Northeast Region; of these, 49 happened in CE, 19 in PI, and two in Alagoas State (AL). In these states, the animals affected were backyard pigs, which meant that a comparatively smaller production loss occurred (in comparison to the extensive loss experienced when large-scale pig farming had been affected in the south of the country), but also meant that already poor, small producers were reduced to even worse economic conditions [4].

Infection by CSFV occurs mainly by direct contact between animals, but contaminated fomites can also present considerable risks. Once infection has occurred, the virus is distributed throughout the organism and can be found in several organs. Typical clinical signs include depression and a high fever associated with severe leukopenia, bruising, nervous signs, and death [5]. The diagnosis needs to be quick to prevent the disease from spreading as much as possible, and is usually performed either serologically or by PCR [6].

The CSFV genome is composed of a single, positive strand of RNA with two untranslated regions (5 ′UTR and 3′ UTR) and an open reading frame encoding eight non-structural proteins (Npro, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B), and four structural proteins (C, Erns, E1, and E2) [7]. Among these, the E2 protein is one of the most widely used for phylogenetic analysis because of its variability, which allows CSFV strains to be divided into three genotypes [1]. Genotype 1 is the most commonly found in the Americas and the only one found in Brazil [2]. It can be subdivided into up to six subgenotypes according to classifications proposed in the literature [2, 8]. Genotypes have no direct association with virulence; however, studies have shown that most strains with high virulence belong to genotype 1. Genotypes 2 and 3 generally exhibit low to moderate virulence [9].

Phylogenetic studies are important auxiliary tools for understanding the spread of diseases. A previous survey carried out in Brazil suggested that three distinct subgenotypes were circulating in Northeast Brazil. Among these, subtype 1.5 was reported as the most prevalent, with 1.6 and 1.1 being the rarest, with only one record from each interval of the study from 2003 to 2009 [2]. Brazilian samples are genetically different from those of other countries. Recent studies have shown that subtype 1.6 diverges genetically from others in South American countries, such as Peru and Ecuador [2, 5–10]. Genotype 1.5 still presents only in Brazilian samples with high genetic divergence [10].

The objective of this work was to study the evolution of CSFV in Brazil using samples collected from outbreaks in Northeast Brazil between 2018 and 2020. The study was carried out by sequencing the E2 protein-encoding regions of the virus genome. Sequences were used for phylogenetic analyses and comparisons with sequences from other studies in Brazil and the rest of the world.

Material and methods

Surveying CSF outbreaks

CSF outbreak data were obtained from the records of the World Animal Health Information System (WAHIS), also known as OIE from the Federal Agriculture Defense Secretary, and from the sample records of Animal and Plant Health Laboratory and Inspection (LFDA). CSF outbreaks took place from 2018 to 2019; they commenced in CE with 48 foci, spreading fast to PI, where 18 CSF outbreaks were recorded, and two in AL in the Northeast Region. In 2020, only one CSF outbreak occurred in Parnaiba, PI. These federal states were outside the CSF-Free Zone. The CSF outbreak survey began after the first CSF outbreak was registered in the city of Forquilha (CE) on August 25, 2018. This was confirmed by laboratory analysis. All 70 outbreaks were reported to OIE. The municipalities affected by CSF outbreaks were plotted on a geographic map displaying the possible relationships among CSF outbreaks based on local proximity and trade (Fig. 1).

Fig. 1.

Municipalities from CE and PI where CSF outbreaks increased between 2018 and 2020. CE (number of outbreaks): Forquilha (2), Groaíras (6), Santa Quitéria (2), Varjota (3), Reriutaba (4), Cariré (6), Moraujo (4), Frecheirinha (3), Mulungu (1), Graça (1), Ipu (1), Hidrolândia (1), Martinópole (1), Tianguá (2), Coreaú (2), Sobral (1), Croatá (2), Granja (3), Viçosa do Ceará (2), Massapê (1). PI: Lagoa do Piauí (2), Cabeceiras do Piauí (5), Murici dos Portelas (4), Domingos Mourão (2), Brasileira (2), Milton Brandão (1), São João do Arraial (1), Parnaíba, (1), Luís Correia (1)

Samples

Organ tissue and blood samples collected from necropsied pigs, which were sick, suspected to be infected, or dead during the CSF outbreaks in CE and PI, were sent to official laboratories for CSFV diagnosis and differential examinations. Samples were usually collected from the following organs: tonsils, lymph nodes, spleen, brain, pancreas, small intestine, heart, and lungs. Samples were collected by the official veterinary service and sent to the official laboratory, where diagnosis was confirmed by real-time PCR [11] and sequencing.

RT-PCR

RNA was extracted using TRIzol reagent (Invitrogen, USA) as recommended by the manufacturer. Samples which had resulted positive for CSFV in RT-qPCR were treated according to a one-step RT-PCR protocol developed in this study, based on the One Step Ahead RT-PCR kit (Qiagen, Germany) using primers CSFV.E2.778.F (TGAGATAGGGCTACTYGGGG) and CSFV.E2.778.R (CATRGGGCCCAGTCTTTCAT) at a concentration of 0.4 µM in a total reaction volume of 25 µL. The thermal cycler was programmed at 95 °C for 10 min, 95 °C for 5 min, followed by 40 cycles at 95 °C for 20 s, 52 °C for 30 s, and 72 °C for 60 s. PCR products were analyzed on a 1.0% agarose gel stained with ethidium bromide; specific bands were cut, purified using MEGAquick-spin Total (iNtRON Biotechnology, South Korea), and subjected to sequencing on a 3500 genetic analyzer (Thermo Fisher Scientific, USA) using BigDye v 3.1 (Thermo Fisher Scientific, USA).

Phylogenetic tree

Sequences were analyzed and edited using BioEdit program [12]. The phylogenetic tree was built in MEGA X program [13], using Kimura 2-parameter model with 1000 bootstrap replicates. Forty-seven GenBank samples were included in the analysis to ascertain genotypes and sub-genotypes to which the samples belonged.

Positive selection

Calculations for evaluating positive selection were performed using the Selecton software [14] and Datamonkey with default parameters for a mixed effects model of evolution (MEME) [15]. Calculations were made using three sets of the phylogenetic tree: the first set encompassed all the sequences used in the present study, the second set contained the Brazilian clade of subgenotype 1.5 only, and the third set was composed of all sequences except those of subgenotype 1.5.

Phylogeography

Bayesian analyses were performed on Brazilian only samples with Beast 2.0 [16], using the HKY model and testing of three molecular clocks: relaxed normal log, relaxed exponential, and strict molecular clock. The chain size was 100,000,000. The results were evaluated in Tracer for an ESS minimum of 200. Then, SPREAD [17] was used to compute the phylogeographic data.

Haplotypes were generated in DnaSP [18] and analyzed in PopArt using the median joining network model to analyze the phylogenetic network [19]. Analyses were performed using a set of Brazilian sequences and the same sequences collected in GenBank, as described previously.

Results

Phylogenetic tree

The Brazilian samples sequenced (GenBank accession number MW368988-MW369005) in this study were grouped into two clades (Fig. 2). Subgenotype 1.5 represented the first clade, as previously described. Even within this group, the sequences showed a degree of divergence from sequences of previous outbreaks in northeastern Brazil [2], which determined an intragroup genetic distance of 0.02.

Fig. 2.

Phylogenetic tree built using MEGA X, the K2 model of nucleotide substitution, and the Maximum Likelihood model with 1000 replicas of bootstraps

All subgenotypes had a minimum genetic distance of 0.10 (Table 1). Samples whose genotype was not defined in GenBank were included in the comparison that were used to evaluate the distance between groups that were more distant phylogenetically and geographically.

Table 1.

Phylogenetic distances between groups and intra-groups

Group/genotype	Intra-group	Inter-group
		1.1	1.2	1.3	1.4	1.5	1.6-BR	1.6-P + E	Asian
1.1	0.04
1.2	0.00	0.1000
1.3	0.04	0.0981	0.1125
1.4	0.02	0.1188	0.1179	0.1229
1.5	0.02	0.1190	0.1764	0.1612	0.1746
1.6—Brazil	n/c	0.1006	0.1513	0.1483	0.1432	0.1312
1.6—P + E**	0.01	0.0945	0.1384	0.1300	0.1388	0.1401	0.1297
Asian	0.10	0.1815	0.1814	0.1858	0.1793	0.2090	0.1874	0.2100
Europe + India	0.11	0.1972	0.2163	0.2064	0.2083	0.2232	0.2058	0.2211	0.1529

^*Not calculated

^**Peru and Ecuador

A specific analysis was performed on samples classified as subgenotype 1.6 [10]. Taking into account all samples from Peru and Ecuador, the distance of the Brazilian sequence from the Peruvian and Ecuadorean sequence was found to be 0.13.

Positive selection

All groups analyzed using Selecton displayed positive pressure sites. Tyrosine at position 365 (GenBank accession number YP_009508222.1) appeared to be under strong positive pressure within subgenotype 1.5; however, it ceased to be under pressure when the analysis of all the sequences used in this work was performed without this clade. The rest of the phylogenetic tree presented six points (amino acids 738, 760, 779, 857, 886, 890) under positive pressure that are not present in genotype 1.5. The results obtained using MEME also indicated six positive pressure points for the rest of the phylogenetic tree. The same test adopting MEME was performed using only samples belonging to subgenotype 1.5 to verify the evolution of this specific group. Only two sites of subgenotype 1.5 were found to be under positive pressure.

Phylogeography

Phylogeography analyses indicate a continuous presence of CSFV in the state of CE since 2003. Results for each molecular clock model used are listed in Table 2. Molecular clocks calculated for the E2 gene ranged from 1073 × 10⁻³ to 1053 × 10⁻³. Our data indicate a constant population of the virus until a slight expansion occurred when outbreaks started (Fig. 3). The phylogenetic signal, when analyzed together with geographic data, shows that the root of the outbreaks in the last 20 years is to be found in the state of CE, where the virus shows the greatest genetic diversity within the largest population of samples collected (Fig. 4). The data also show that the virus had an older entry, before the 2000s. Since then, it has been detected continuously. The results of the strict molecular clock indicate the evolution of the virus at a constant rate throughout the tree. Between Brazilian samples, genetic diversity increased progressively, with a sudden number of variants appearing after the initial outbreaks of 2018.

Table 2.

Results for the molecular clock models tested in this study

Molecular clock	Prior	Posterior	Likelihood	Prior
Strict	Coaslecent constant population	− 1268.796	− 1189.3	− 79.496
Strict	Coaslescent Bayesian skyline	− 1284.779	− 1187.0	− 97.681
Relaxed log normal	Coaslecent constant population	− 1263.624	− 1187.6	− 76.011
Relaxed log normal	Coaslescent Bayesian skyline	− 1281.198	− 1185.947	− 95.25

Fig. 3.

Bayesian skyline reconstruction from the data generated using Beast 2.0 for subgenotype 1.5 samples. The result indicates a steady population with a slight increase from 2018 onwards

Fig. 4.

Phylogeographic analysis of samples of sub genotype 1.5 collected in foci of Northeast Brazil. The data demonstrate the spread of the virus in CE and its subsequent spread to PI and RN

The 16 haplotypes generated from the Brazilian samples presented the following characteristics: haplotype diversity equal to 0.9447, 448 invariable (monomorphic) sites, 65 variable (polymorphic) sites, and a total number of mutations of 67 (42 singleton variable sites and 23 parsimony informative sites). Brazilian CSFV sequences are separated from sequences of other countries, except for sequences from AP and MA [2] (Fig. 5). The results corroborate the Bayesian analyses demonstrating continued presence of the virus and show that there is no relationship between the outbreaks in the states of CE, PI, and RN with the outbreaks of MA and AP. The phylogenetic network of the Brazilian samples demonstrated a high number of haplotypes. The number of nucleotide variations was low between samples from recent outbreaks and high when nucleotides were compared between samples from recent outbreaks with samples from outbreaks dating back to before 2009 (Fig. 6).

Fig. 5.

Phylogenetic network generated in PopArt by the Median Joining Network model for samples from Brazil and other countries. The circles show the Brazilian samples of the subgenotype 1.5. The results demonstrate the clear separation of samples from recent outbreaks and those detected in the early 2000s in the Brazilian Northeast

Fig. 6.

Phylogenetic network generated in PopArt by the Median Joining Network model for Brazilian samples only. The circles were colored by state. The results demonstrate the genetic difference between the sequences of the states of AP, MA, and RN. The sequences of outbreaks from 2018 are phylogenetically closer to those that caused the outbreaks between 2001 and 2009 in the state of CE

Discussion

The damage caused by CSFV represents a permanent threat to pig farming. Even though the main pig-producing regions in Brazil are in the south and southeast (considered free of the disease without vaccination), there is still a great concern and constant surveillance in areas such as the Northeast, where the virus still circulates in smallholdings with small-scale production, usually focused on subsistence. Subsistence pig farming (or family farming) in the Brazilian Northeast Region represents a fundamental source of jobs, income, and food for the families involved [20]. The growth of pig farms has been sustained by using local pigs, which, on the one hand, is a positive fact as these animals retain genetic resources from native pig breeds that have adapted to and survived harsh local conditions [21]. However, on the other hand, these pigs are susceptible to CSFV infections, which represent a serious source of concern [20].

A previous study identified three subtypes belonging to the 1.0 genotype in Brazil [2]. Among them, the sequences belonging to subgenotype 1.1 were similar to those of other countries; however, there was a great divergence of the subgenotypes 1.5 and 1.6.

A recent study demonstrated that a sequence classified as belonging to subgenotype 1.6, found in the state of AP, was similar to others of the same group detected in Peru and Ecuador, but still stood out genetically [8]. The results of the present study suggest that samples from Peru and Ecuador may not belong to the same group as the Brazilian sample. The genetic distance between the Peruvian and Ecuadorian sequences when compared to the Brazilian sequence is 0.13, which is even greater than the distance between some subtypes within the 1.0 genotype. These values suggest that the samples did not belong to the same genetic group. Analyzing larger fragments of the genome or complete genome may clarify this issue.

The sequences classified as belonging to sub-genotype 1.5. formed a clade genetically distant from samples originating from other countries. Our data show an evolution rate ranging from 1073 × 10⁻³ to 1053 × 10⁻³/year, which is higher than that calculated in previous studies for the CSFV genome or for the complete E2 sequences [10, 22]. However, one study found similar evolution rates for E2 in CSFV isolated from wild boar in Germany [23]. These results were explained by the fact that surface proteins are more prone to selective pressures by the host, so a higher mutation rate is needed in order to evade the host's immune response.

CSFV is a relatively stable virus compared to other RNA viruses [24]. Its evolution can be driven by the use of vaccines, generating greater stability and slower genetic changes [25]. Vaccination against CSF is prohibited in all CSF-free zones in Brazil and is only allowed in areas not considered CSF-free with government permission when outbreaks occur. Without the influence of vaccination, the CSFV seems to have continued to evolve in the Brazilian Northeast, with new haplotypes emerging to the point that the first ones found in the early 2000s were replaced and not found after 2018.

Some of the positive selection sites described in the present study are located in regions of the viral sequence which had been predicted to be epitopes in a previous study [26], which explains and confirms our results. According to computational prediction methods, the site under strong positive pressure in subgenotype 1.5 lies within a B cell epitope [27, 28]. It is known that mutations in E2 can produce variants that can escape the immune response. Further studies would help to understand whether the genetic variations found in subgenotype 1 occur due to antibody pressure.

Our results demonstrate the rapid expansion of the CSFV population during the outbreaks taking place from 2018 onwards, with a high number of haplotypes, but little genetic difference between them, indicative of a rapid viral population expansion [29], which may be related to the sudden increase in outbreaks detected in the states of CE and PI. However, the differences between haplotypes before 2009 and the most recent haplotypes revealed that the virus continued to circulate in the state of CE during these years without being detected.

Our data demonstrate the continuous presence of CSFV in the state of CE, from where it probably spread to the state of PI in 2019. The virus has been constantly evolving, but it was not detected for almost 20 years before the sudden increase in pig population. It is important to continue surveillance in this region to understand how the virus has evolved in swine populations.

Author contribution

Bioinformatics analysis: Antônio Augusto Fonseca Júnior.

Virus isolation, serology, PCR, sequencing: Antônio Augusto Fonseca Júnior, Mateus Laguardia-Nascimento, Aline Aparecida Silva Barbosa, Valdenia Lopes da Silva Gonçalves.

Manuscript review: Tânia Rosária Pereira Freitas, Anselmo Vasconcelos Rivetti Júnior, Marcelo Fernandes Camargos.

Manuscript preparation: Antônio Augusto Fonseca Júnior.

Funding

This work was funded by Ministério da Agricultura, Pecuária e Abastecimento.

Data availability

Data are available upon request.

Code availability

Not applicable.

Declarations

Ethics approval

Samples used in this study were collected by the Brazilian Ministry of Agriculture as part of routine laboratory tests for the diagnosis of classical swine fever within the eradication program.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Conflict of interest

The authors declare no competing interests.