Our study reports the evolution of DENV-4 in French Guiana and the French Caribbean islands since its last detection and circulation from 1993 to 1995. We studied 8 DENV-4 strains isolated from human sera from French Guiana in 1993 and 1995 (n = 6) and in 2004 and 2005 (n = 2) (Table). We also tested 3 human serum specimens from Martinique (n = 2) and Guadeloupe (n = 1) that were positive for DENV-4 during dengue surveillance in the fourth quarter of 2004.

DENV-4 infection was confirmed by using virus isolation on AP 61 cells as previously described (7). We then amplified from first-passage RNA a 1,940-bp region of the genome for the E gene and adjacent prM/M and NS1 junctions previously described (8) to analyze the phylogenetic relationships of these strains. Each PCR product was cloned by using the TOPO TA Cloning kit (Invitrogen, Paisley, UK). For each isolate, 3 clones were sequenced by Genome Express (Meylan, France).

We compared our 11 sequences to the 87 DENV-4 sequences available in the GenBank database. Sequences were aligned by using MacVector version 7.2 software (Oxford Molecular Ltd., Oxford, UK). Percentage of nucleotide identity between sequences was determined for all observed differences (including insertions and deletions) and was calculated for the 1,473 nt available for all 98 sequences analyzed. Sequence comparison of strains from Martinique, Guadeloupe, and French Guiana with those reported from Asia showed that nucleotide divergence is 8%–12%. The level of nucleotide divergence was lower for sequences from the Caribbean area (maximum nucleotide divergence 2.8%). All 11 newly obtained sequences were different from each other. These sequences exhibited a 0.06%–0.47% nucleotide divergence (strains isolated in 1993 and 1995) and a 0.06%–1.22% nucleotide divergence (strains isolated in 2004 and 2005) among themselves. These strains had amino acid changes at position 351 in the E gene (from isoleucine to valine) and position 52 in the NS1 gene (from lysine to arginine), which supports their inclusion in the modern Caribbean basin clade (8).

Figure

Figure. Phylogenetic tree of American dengue virus type 4 (DENV-4) strains generated by using the neighbor-joining algorithm with Kimura 2 parameter distance. Numbers above branches refer to bootstrap values generated by using...

We conducted phylogenetic analyses on nucleotide sequences by using distance and parsimony methods in PAUP* version 4.0b8 to increase the reliability of derived tree topologies (9). We also used the neighbor-joining distance matrix algorithm with the Kimura 2 parameter (Figure) and the heuristic algorithm for the maximum parsimony analysis (data not shown). Robustness of resulting topologies was examined by using bootstrap analyses (10). Both neighbor-joining and parsimony algorithms underwent 1,000 iterations.

The phylogenetic tree constructed with the neighbor-joining method on nucleotide sequences restricted to all American DENV-4 isolates available in GenBank (Table) is shown in the Figure. Our 11 strains segregate into 2 main clades that were identified in the 2 phylogenetic analyses from consistent topologic associations and high bootstrap values. The strains isolated from French Guiana in 1993 and 1994 clustered with strains from Suriname and Costa Rica, which were isolated at the same time, as well as with a previously reported strain from Martinique isolated in 1995. Our 5 strains isolated in 2004 and 2005 are related to a group of DENV-4 genotype II sequences isolated in the Bahamas in 1998. This finding suggests that strains circulating in French departments of the Americas in 2004 and 2005 probably evolved from strains detected in the Caribbean islands during the late 1990s. Our observations confirm that DENV-4 lineages in the Caribbean and nearby regions are grouped temporally rather than by the geographic origin of the isolates, as has been previously suggested (8).

The DEN-4 serotype was first reported in the Americas in 1981. At that time, isolates collected in the Americas were similar to strains collected in Southeast Asia. This lineage was designated DENV-4 genotype II (11). Since it was first detected in Dominica and the French islands of Saint Barthelemy and Saint Martin (12), this genotype has spread rapidly throughout the Caribbean and Latin America and caused DF and sporadic cases of DHF/DSS (3).

DENV-4 genotype II never stopped circulating in the Caribbean region, despite its being periodically locally eliminated (8). This genotype reappeared in French departments of the Americas in 2004 after the introduction of a strain circulating in neighboring Caribbean islands, rather than by reemergence of a strain detected 10 years ago. Furthermore, this DENV-4 genotype was detected in French Guiana (October 2004) a few weeks after it was detected in Martinique (August 2004), which suggests that the strain came from the Antilles (Table). A similar phenomenon had already been observed in 1999 for the DENV-3 genotype (J. Morvan, unpub. data). This model of viral spread may be caused by cultural and economic ties between French departments, which allows gene flow through the Caribbean islands and South America as persons with viremia move through the region. This reflects the general pattern of dengue evolution in the Americas (13,14).

Reintroduction of DENV-4 in the French Antilles and French Guiana in 2004 after a decade of absence highlights the risk for subsequent epidemics in persons with no immunity to dengue. As expected, the French Antilles had a major DENV-4 epidemic with cocirculation of DENV-2 from July 2005 to January 2006 (15). In contrast, only sporadic cases of DENV-4 infection have been observed in French Guiana since the beginning of 2005. This indicates that other factors such as mosquito density or human susceptibility, in addition to immune status of the host or silent transmission of the virus, can modulate the risk for dengue epidemics.