Simultaneous circulation of genotypes I and III of dengue virus 3 in Colombia

As shown in the phylogenetic tree (Figure 1), in this study DENV-3 circulation in Colombia was detected since the beginning of 2002. The results were consistent between distance and character-based methods, with minimal differences in topologies (Figure 1, Additional file 1, and data not shown). The most important findings are the detection of genotype I (or Southeast Asia/South Pacific genotype) in Colombia and its co-circulation with genotype III (or Indian genotype) [6, 18] in three states from Colombia, La Guajira, Guaviare and Huila (Figure 2). Genetic diversity within 3'end of the E gene of DENV-3 throughout the world allowed resolution of previous clustering in four lineages (genotypes) [6], and the presence of a basal clade in genotype I, would be consistent with a fifth genotype [19].

Diversity within DENV-3 has been previously identified and classified [20], but they have found that genetic distance between genotypic groups is low when compared to genetic diversity in DENV-1 and -2, showing that the fixation rate is also lower [18]. By contrast, it has been published that DENV-3 has the higher substitution rate between the dengue viruses (about 7,48 substitutions/site/year) [21]. Our results shows that overall mean distance for DENV-3 as estimated for 84 sequences of 224 bp, with MEGA software is 0,070; for 104 DENV-1 sequences is 0,065 and for 60 DENV-4 sequences is 0,053. Overall mean distance for DENV-2 has not been determined in this study.

In the Americas, DENV-3 circulation was reported in the 1960's and 1970's, and all sequenced strains were clustered within genotype IV or American genotype [6, 18]. After these isolations, genotype IV has not been identified in any country and could be considered as an extinct genotype. In Colombia, circulation of DENV-3 was reported from 1975 – 1977 [12]. The identification was made by viral isolation in mosquito cells (C6/36) and indirect immunofluorescence, but molecular detection was not carried out. Therefore, sequences of isolated strains during this period have not been determined. It is highly probable that Colombian isolates from this period would cluster within genotype IV, like Puerto Rico strains isolated in the same year [GenBank: L11434].

In the present study, we attempted to amplify historical Colombian strains of DENV-3 isolated in 1977, but it could not be achieved, maybe due to poor samples, or improper maintenance or storage during this time. The recovery of these samples could enrich the basal clade of genotype IV, or might help in explaining the presence of an Asian genotype (genotype I) in Colombia at present if it had been circulating in the past, a very difficult hypothesis to corroborate.

The genomic region used to analyze the relation between strains has been evaluated and determined to be an informative region for genotyping [22]. Nevertheless, the complete E gene of some strains has been sequenced, and the topology results are newly confirmed (Additional file 1).

Since DENV-3 genotype III has been present in northeastern and southwestern Colombia since early 2002, different routes of introduction are possible. First, The Venezuelan origin is supported by high similarity of sequences and circulation of this genotype in Venezuela in August of 2001, when the largest epidemic caused by DENV there since the 1989 DENV-2 epidemic ended [17]. The Venezuelan origin of Colombian strains is also supported by the first isolation in La Guajira, along the frontier with Venezuela. Second, it is possible that DENV-3 genotype III had come across the frontier with Peru and/or Ecuador. The high similarity between a strain from Ecuador (DQ177898_Ecuador00) and a strain from Putumayo, Colombia (352_Putu02), a state along the border, offers hard support for this idea. Finally, the entry of genotype III into the Americas was first reported in Panama and Nicaragua in 1994 [13], so another possibility is its introduction into Colombia through the northwestern border with Panama. However, results do not support this hypothesis, not only due to the genetic distance between strains of Colombia and Panama but also mainly due to distribution of strains on northeast and southwest of Colombia. Surely not only one introduction event had occurred, but probably at least two events, via northeast and southwest of Colombia.

DENV-3 genotype I was recently described in the Americas from nine cases in Brazil, as a result of phylogenetic analysis using two fragments corresponding to C-prM and partially the E gene [23]. Here, we report the presence of this lineage in Colombia from a different region of the E gene, without recent closely-related sequences available on GenBank to date. Moreover, the related sequences corresponding to Asian strains were isolated in 1973 in Japan as an imported case and in 1980 in Guangxi, China (GenBank: AB111085 and AF317645). Samples that clustered in this lineage are located in a basal branch into genotype I, with high bootstrap support (86%) and mean distance between clades of 5%, estimated with Tamura-Nei model to be classified as a fifth different genotype, referred to as genotype V in [19]. Variability within genotype I has been demonstrated as the presence of into-clade nucleotide substitutions and branching in few years (Figure 1).

The presence of DENV-3 genotype I only in Colombia, and its close relation with Asiatic strains from 1973 and 1980, suggests that strains circulating in Colombia during the 1970's would have not been of genotype IV, like other American strains from that period, but, perhaps a strain of Asiatic origin that had been circulating without detection for over 25 years until 2002. This speculation needs more data to support it, because there is no evidence for genotype circulation in Colombia in the past, and explaining possible silent circulation without causing outbreaks for more than twenty years could be a challenge.

The presence of the Southeast Asia/South Pacific genotype has recently been detected not only in Colombia, but also in Brazil [23].

DENV-3 genotype IV was last reported in Puerto Rico in 1977 (as corroborated by sequencing) [24], but to date Colombian isolates from the same year have not been sequenced because of lack of good samples from these years. Reintroduction of other genotypes clearly has not signified displacement of genotype IV, probably because it was not present for more than twenty years, so co-circulation was not possible.

Intra-serotype recombination has been detected in natural populations of DENV [22, 25‐29]. Nonetheless, the significance of recombination events for increasing genetic diversity is unknown. The topology of the phylogenetic tree could be affected by recombination between strains, and then the results could be misinterpreted. Our findings obtained by using a short fragment could be a product of recombination. For this reason, we achieved sequencing of complete E gene of strains corresponding to both DENV-3 Colombian genotypes. The results of the phylogenetic reconstruction (Additional file 1) were consistent with the presence of genotypes III and I (genotype V according to [19]). Additionally, a recombination analysis using the complete E gene was carried out, but recombination events were not detected (data not shown).

As known, the potential for causing severe disease has been described for all four serotypes of DENV, and the main factors considered to explain its pathogenicity are host genetic susceptibility, antibody dependent enhancement and differences in virulence among strains [30]. It is evident that phenotype is not segregating with phylogeny, but is an evolutionary convergence, resulting from interaction of the viruses with hosts and moulded by selection to enhance its transmission and persistence [31].

Determinants of virulence have been located in three genomic regions [32] and have been tested in vitro [33], so the genotypes have been more or less related with potential to cause DHF. Recent studies have concluded that the spread of genotype III of DENV-3 from the Indian subcontinent to Africa and then to Latin America was correlated with an increase in severe cases of dengue disease [10, 34]. The ability of all serotypes to cause severe disease is an indicator of adaptive selection of this character during independent evolution of DENV serotypes. However, more efforts should be made to understand the role of viral genetics in human pathogenesis.

Although the origin of genotype I is uncertain, the co-circulation with genotype III could have epidemiologic implications if it has intra-serotype antigenic variation related with differential generation of protective antibodies and immune response [6]. It is important to take into account the low sample size, because the possibility of a more wide distribution of the genotypes I and III into the country (Figure 2).

The relevance of these results is the detection of two different genotypes in the same country, one of them of Asiatic origin, only recently described in the Americas [23]. The results underscore the need for a global strategy of genotypes circulation surveillance, because disease dynamic is more than a regional problem, involving neighbouring countries as well. The establishment of a Pan-American program would provide very useful epidemiological information about the potential of strains for causing outbreaks.

The strains included in the study, with locality, year and GenBank accession numbers, are listed in table 1. Samples were collected by local hospitals in Medellín (Colombia) and Public Health Laboratories of the National Network all around the country and remitted to the National Institute of Health (Colombia) for diagnostic and epidemiological surveillance. Serum or plasma was obtained and kept at -70°C until processing. The samples cover a period of four years since reintroduction and detection of DENV-3 in Colombia (2002 – 2005).

C6/36 cells cultured in Dulbecco's modified Eagle's medium (DMEM), were infected with 0.15 ml of samples and incubated for 10 days at 28°C, washed with PBS, removed by hitting the culture tubes manually and seeded on slides. Cells were then fixed with acetone and the indirect immunofluorescence procedure was carried out incubating the cells with serotype-specific monoclonal antibodies (kindly donated by Dr. Elizabeth Hunsperger, CDC Puerto Rico) for 60 minutes and then washed with PBS and incubated for another 60 minutes with a commercial secondary antibody conjugated with fluorescein isotyocianate.

Aliquots of 140 μl of serum or supernatants of cell cultures were placed into 540 μl of AVL buffer with Carrier RNA and used to extract the viral RNA with QIAamp Viral RNA Minikit (QIAGEN, Germany) as indicated by manufacturer. RNA obtained in 60 μl of AVE buffer was stored at -70°C and used in the RT-PCR. Alternatively, the total RNA of some samples was extracted by the use of TRIZOL^® LS (INVITROGEN, Inc., USA), and a final volume of 15 μl was recovered in these cases.

The RT-PCR and nested-PCR have been previously described [35]. When viral load was too low, nested-PCR was used to detect DENV directly on clinical samples, so sensitivity of detection was increased more than five logarithms and passage of viruses in cell cultures was avoided (data not shown). RT-PCR primers were designated to amplify an intergenic region E/NS1 of 776 bp, and nested-PCR to amplify an internal region of 350 bp.

Products of RT-PCR or nested-PCR were purified using QIAquick PCR Purification Kit (QIAGEN, Germany). Sequencing reactions on both strands were performed with 10 pmol of the primers used for the second round of amplification, and the ABI Prism Dye Terminator Cycle Sequencing Ready Reaction Kit (APPLIED BIOSYSTEMS, USA), and analysed using an ABI model 377 automated sequencer (APPLIED BIOSYSTEMS, USA).

Four sequences were obtained for each sample, two sequences with sense and two with antisense primer. Editing and consensus obtaining were performed with the SeqMan module of Lasergene (DNASTAR Inc. Software, Madison, Wis.).

Sequences on GenBank corresponding to different lineages of DENV-3 were downloaded and aligned with the consensus sequences obtained in this study, using Clustal W software [36]. Additionally, a visual correction of alignment was done. A fragment of 224 bp was used for phylogenetic reconstructions corresponding to the 3' end of the E gene (nucleotides 1256 to 1479). The portion of the NS1 gene amplified with the nested-PCR was excluded from the analysis due to the absence of this portion in the majority of reported sequences.

Alignment of the sequences obtained in the present study (n = 32) (Table 1) and homologous sequences for DENV-3 available on GenBank (n = 68) (Table 2) were used for phylogenetic reconstructions. Many sequences of different strains were completely identical to the fragment analysed, and so one sequence was used for analysis, corresponding to the first isolation.

The strain 359_Caqu03 was completely identical to 363_Caqu03, 366_Caqu03, 367_Caqu03, 368_Caqu03, and 464_2003; strain 449Meta05 was identical to 456_Guav05 and 461Guav05; strain 352_ Putu02 to 484_Putu02; strain 221_Guaj02 to 233_Guaj03, 517_Caqu03, 518_putu04, 247_Guav05 and 530_Guav05; and finally, strain 375_SAnd03 was identical to 389_Guaj03, 395_NSan04, 417_Guav04, 429_Huil04, 535_Huil04 and DV06_Ant05.

The phylogenetic trees were estimated for the 224 bp fragment, corresponding to the 3' end of the E gene. Initially, the neighbour-joining algorithm was used with 10000 bootstrap replicates and the Tamura-Nei model of nucleotide substitution with MEGA 3.1 software [37]. Maximum parsimony and Maximum Likelihood trees were obtained with PAUP* [38]. For selecting the model of substitution, MODELTEST software and current dataset were used and the resulting parameters were used for running maximum likelihood analysis. Trees were rooted using genotype IV, only for graphical purposes.