Evolution of Dengue Virus Type 3 Genotype III in Venezuela: Diversification, Rates and Population Dynamics

In order to gain insight into the degree of genetic variability of DENV-3 genotype III strains circulating in Venezuela, 29 Venezuelan DENV-3 genotype III E gene sequences representing strains isolated between 2000 and 2007 in seven different Venezuelan geographic locations, were aligned with 58 sequences from DENV-3 genotype III E gene of DENV isolated in Latin America and 11 DENV-3 sequences from strains isolated elsewhere representing other DENV-3 genotypes (for strains included in these studies see Additional File 1, Table S1).

Once aligned, we first identified the optimal evolutionary model that best represent our sequence dataset (Akaike Information Criteria and Hierarchical Likelihood Ratio Test indicated that the GTR+Γ model fit the sequence data). Using this model, maximum likelihood phylogenetic trees were constructed and the robustness of each node of the tree was assessed by approximate Likelihood Ratio Test (aLRT). The results of these studies are shown in Figure 1.

All strains in the tree are assigned according to their genotype. Genotype III strains cluster together, strains belonging to other DENV-3 genotypes cluster separately (see Figure 1). These clusters are supported by very high values of aLRT. Inside genotype III cluster, three different clades can be observed for strains isolated in South America: one composed exclusively by strains isolated in Venezuela (Cluster A; Figure 1, bottom); one composed by strains isolated in Martinique, Brazil, Bolivia, Paraguay (Cluster B; Figure 1, middle); and one composed by strains isolated mainly in Ecuador and Peru (Cluster C; Figure 1 top). Clusters B and C also include strains isolated in the Caribbean region. Previous strains isolated in Nicaragua, Panama and Mexico from 1994 to 2000 were assigned to a different cluster inside genotype III (see Figure 1, middle). These strains circulated in the initial years after the introduction of DENV-3 genotype III in the continent, but show a great divergence with recent Latin American strains. This reflects a significant evolutionary change among strains isolated in the initial years and recent strains.

Interestingly, strains isolated in Venezuela are not only assigned to Cluster A, but also to Clusters B (strain DENV-3/VE/BID-V911/2001) and C (strains DENV-3/VE/BID-V1593/2005 and Gua2007), revealing at least three different introduction events of DENV-3 genotype III in that country (see Figure 1). The results of these studies also show the diversification of DENV-3 genotype III circulating in South America in three different genetic lineages (see Figure 1).

In order to confirm these findings, the same studies were done using a dataset containing all the 119 DENV-3 genotype III sequences isolated in Venezuela (for strains included in these studies see Additional File 1, Table S1). Using this dataset, we have found that apart from the three strains mentioned, the rest of the 119 DENV-3 genotype III strains isolated in Venezuela were assigned to Cluster A (Additional File 2, Fig. S2).

In order to gain insight into the diversification of DENV-3 genotype III in the South American region, full-length E protein amino acid sequences from strains isolated in Venezuela, representative of the three genotype III clusters observed, were aligned.

DENV-3 E protein resembles its homolog of DENV-2 in its dimeric structure and in the details of its protein folding [12]. Each monomer consists of three domains: domain I, an eight-stranded β-barrel, which organizes the structure; domain II, which contains 12 β-strands and two α-helices; and domain III, which is an IgG-like domain, with 10 β-strands. In solution and in the crystals, two monomers of E assemble head to tail to form a dimer [13].

As shown in Figure 2, Cluster A strains (composed entirely by genotype III strains isolated in Venezuela) present an amino acid substitution in position 329 (A→V), while strains from Clusters B and C share an Alanine at this position (see Figure 2, domain III). Interestingly, Cluster C strains present an amino acid substitution at position 132 (Y→H) with respect to Clusters A and B strains (see Figure 2, domain I), in a region previously described to be exposed at the surface of the protein [14]. Besides, the unique Venezuelan strain assigned to Cluster B present an amino acid substitution at position 346 (N→S) (see Figure 2, domain III). An asparagine (N) at position 388 was found in all strains included in these studies (see Figure 2). Previous studies in DENV-2 have shown that an N at this position appears to be associated to increased incidences of hemorrhagic fever (see Figure 2, domain III) [15].

These studies were repeated using all the 119 sequences of DENV-3 genotype III strains isolated in Venezuela. The same conclusions were obtained using this dataset (for detailed results of substitutions found in these strains in nucleotide and/or amino acids sequences, see Additional File 3, Table S2).

In order to map the amino acid substitutions found on the E protein structure of DENV-3 genotype III strains of the three clusters identified in this study, we employed the PDB ProteinWorkshop 3.6 [16], using as a reference the E protein structure of DENV-3 strain DENV-3/TH/CH53489D73-1/1973 [12]. The results of these studies are shown in Figure 3.

Amino acid substitution at position 329 of domain III, found in Cluster A strains, is situated in previously identified surface-exposed amino acids in DENV-3 E protein [12, 13, 17]. The amino acid at position 132 of domain II is also exposed on the surface of the E protein (see Figure 3) [14].

In order to gain insight into the evolutionary rate and mode of evolution of DENV-3 genotype III circulating in Venezuela, we used a Bayesian Markov chain Monte Carlo (MCMC) approach as implemented in the BEAST package [18], to analyze E gene sequences of DENV-3 genotype III of strains included in Cluster A (see Figure 1). The results shown in Table 1 are the outcome of the analysis for 20 million steps of the MCMC, using the GTR+Γ model, a relaxed clock [19] and the expansion population growth model [20].

As shown in Table 1, our results suggest that Cluster A, entirely composed by strains circulating in Venezuela, evolved from ancestors that existed around 1998. When the GTR+Γ model is used, a mean rate of 8.48 × 10^-4 nucleotide substitution per site per year was obtained for Cluster A strains (Table 1). This rate is similar to the ones obtained in similar studies for DENV-2 genotype III and DENV-4 genotype II (8.0 × 10^-4 and 8.3 × 10^-4 substitutions/site/year, respectively) circulating in the Americas [21]. This rate is also roughly similar to the others previously estimated for American DENV-3 genotype III strains (8,2 × 10^-4 and 1.03 × 10^-3 substitutions/site/year) [22, 33].

The new sequences of the E gene reported in this study are from Venezuelan strains of DENV-3, originally isolated in the "Department of Virology" from the "Instituto Nacional de Higiene Rafael Rangel - INHRR, Caracas, Venezuela, and from the "Laboratorio Regional de Diagnóstico e Investigación del Dengue y Otras Enfermedades Virales - LARDIDEV, Maracay, Venezuela. Acute-phase sera of patients identified as infected with DENV-3 using an established RT-PCR Multiplex protocol [35], were used to infect monolayers of the Aedes albopictus cell line C6/36. After growth for 7 days at 32°C, virus-infected supernatants were collected, clarified by centrifugation and stored at -70°C until use.

Viral RNA was extracted from 280 μl supernatant medium of virus-infected cells using the QIAamp^® Viral RNA System, according to the manufacturer's protocol (Qiagen^®, Chatsworth, CA, USA). Viral RNA was reverse transcribed to cDNA first strand in a 50-μl reaction final volume with Superscript II reverse transcriptase system (Invitrogen) and pd(N)6 random primers (Invitrogen, USA) or a specific primer (Additional File 4, Table S3). Reverse transcription was allowed to proceed at 42°C for 90 min followed by reverse transcriptase inactivation at 70°C for 15 min. cDNA amplification was performed with synthetic primers listed in Supplementary Material Table 1. Both sense and antisense primers were used to amplify a fragment containing the E coding region on the viral RNA. The reaction mixture contained, in a total volume of 50 μL, 2 to 10 μL of the cDNA, 1 mmol/L of MgSO₄, 0.3 mmol/L of each dNTP, 0.2 μmol/L of each sense an antisense primers, and 2.5 U of platinum^® pfx DNA polymerase (Invitrogen). The amplification was carried out in a thermal cycler (Eppendorff) as follows: 35 cycles at 94°C for 30 seconds, 60°C for 1.5 min, and 68°C for 2.5 min, followed by a final incubation at 68°C for 5 minutes. Each PCR was run with positive and negative controls and the fragments were separated by 1% agarose gel electrophoresis, stained with 1 μg/ml ethidium bromide, and detected under ultraviolet light.

Amplicons were purified using QIAquick PCR Purification Kit from QIAGEN, according to instructions from the manufacturers. The sequence reaction was carried out using the Big Dye DNA sequencing kit (Perkin-Elmer) on a 373 DNA sequencer apparatus (Perkin-Elmer). Alternatively, sequences were obtained from purified amplicons through a commercial sequencing service (Macrogene Inc, Seoul, Korea). Both strands of the PCR product were sequenced in order to avoid discrepancies. The E gene nucleotide sequences obtained from the DENV-3 Venezuelan strains were deposited in the GenBank Database under accession numbers [GenBank:HM348812] through [GenBank:HM348831], (see also Additional File 1, Table S1). Previously reported E gene sequences from DENV-3 isolated in Venezuela were obtained by means of the Virus Variation Database at the National Center for Biotechnology Information (NCBI) [36] (for strains names and accession numbers see Additional File 1, Table S1).

Different datasets were constructed to perform phylogenetic analyses. One included all DENV-3 genotype III E gene sequences isolated in Venezuela (n = 119) between 2000 and 2008, as well as 58 sequences from DENV-3 genotype III E gene of DENV isolated in Latin America and 11 DENV-3 sequences from strains isolated elsewhere representing other DENV-3 genotypes. We created another dataset that includes 29 selected Venezuelan DENV-3 genotype III E gene sequences representing strains isolated in seven different Venezuelan geographic locations between 2000 and 2007 and the same other strains, isolated elsewhere, included in the former dataset (for strain names and accession numbers, see Additional File 1, Table S1). For the selection of the 29 sequences of the latter dataset, we considered several variables in order to avoid biases in our results, including avoiding the presence of duplicates.

Sequences were aligned using the MUSCLE program [37]. Once aligned, we first tested whether a recombination event occurred on any of the sequences used in these studies. We used two approaches implemented in the SimPlot Program [38]: 1) a sliding window analysis of distances and 2) the bootscanning [39]. No recombinant strains were found in the dataset (not shown).

The program Modelgenerator [40] was used to identify the optimal evolutionary model (Akaike Information Criteria and Hierarchical Likelihood Ratio Test indicated that the GTR+Γ model fit the sequence data). Using this model, maximum likelihood trees were constructed using software from the PhyML program [41] (for details about the ML parameters, see Additional File 5, Table S4). As a measure of the robustness of each node, we employed an approximate Likelihood Ratio Test (aLRT), that assesses that the branch being studied provides a significant likelihood gain, in comparison with the null hypothesis that involves collapsing that branch but leaving the rest of the tree topology identical [42]. In order to gain insight into the evolutionary rate and mode of evolution of DENV-3 genotype III strains circulating in Venezuela, we used a Bayesian Markov chain Monte Carlo (MCMC) approach as implemented in the BEAST package v.1.4.8 [18], to analyze E gene sequences of DENV-3 genotype III of strains included in cluster A (Figure 1) isolated between 2000 and 2007. Using the GTR+Γ model and 20 million steps of MCMC, different population dynamic models were tested (constant population size, exponential population growth, expansion population growth, logistic population growth and Bayesian Skyline). Statistical uncertainty in the data was reflected by the 95% highest probability density (HPD) values. Results were examined using the TRACER v1.4 program [43] from the BEAST package. Convergence was assessed with ESS (effective sample size) values after a burn-in of 2 million steps. Models were compared by calculating the Bayes Factor (BF) [44] from the posterior output of each of the models using TRACER v1.4 program [42] as explained in BEAST website http://​beast.​bio.​ed.​ac.​uk/​Model_​comparison. A log BF (natural log units) > 2.3 indicates strong evidence against the null model. The expansion population growth model was the best fit to the data.

In order to map amino acid substitutions found in Venezuelan DENV-3 genotype III E protein, crystallography data of the E protein of the strain DENV-3/TH/CH53489D73-1/1973 [12] was imported from Protein Data Bank [PDB:1uzg]. Visualization was done using PDB ProteinWorkshop 3.6 [16].