Recent Emergence of Dengue Virus Serotype 4 in French Polynesia Results from Multiple Introductions from Other South Pacific Islands

Background Infection by dengue virus (DENV) is a major public health concern in hundreds of tropical and subtropical countries. French Polynesia (FP) regularly experiences epidemics that initiate, or are consecutive to, DENV circulation in other South Pacific Island Countries (SPICs). In January 2009, after a decade of serotype 1 (DENV-1) circulation, the first cases of DENV-4 infection were reported in FP. Two months later a new epidemic emerged, occurring about 20 years after the previous circulation of DENV-4 in FP. In this study, we investigated the epidemiological and molecular characteristics of the introduction, spread and genetic microevolution of DENV-4 in FP. Methodology/Principal Findings Epidemiological data suggested that recent transmission of DENV-4 in FP started in the Leeward Islands and this serotype quickly displaced DENV-1 throughout FP. Phylogenetic analyses of the nucleotide sequences of the envelope (E) gene of 64 DENV-4 strains collected in FP in the 1980s and in 2009–2010, and some additional strains from other SPICs showed that DENV-4 strains from the SPICs were distributed into genotypes IIa and IIb. Recent FP strains were distributed into two clusters, each comprising viruses from other but distinct SPICs, suggesting that emergence of DENV-4 in FP in 2009 resulted from multiple introductions. Otherwise, we observed that almost all strains collected in the SPICs in the 1980s exhibit an amino acid (aa) substitution V287I within domain I of the E protein, and all recent South Pacific strains exhibit a T365I substitution within domain III. Conclusions/Significance This study confirmed the cyclic re-emergence and displacement of DENV serotypes in FP. Otherwise, our results showed that specific aa substitutions on the E protein were present on all DENV-4 strains circulating in SPICs. These substitutions probably acquired and subsequently conserved could reflect a founder effect to be associated with epidemiological, geographical, eco-biological and social specificities in SPICs.


Introduction
Almost all tropical and subtropical regions of the world are concerned by the risk of dengue. Every year, dengue virus (DENV) causes more than 50 million infections, 500 000 hospitalizations and 12 500 deaths, mostly children [1]. Currently there is no vaccine or effective anti-viral therapy and vector control measures regularly fail to prevent the emergence of dengue epidemics. DENV is a single-stranded, positive-sense RNA virus belonging to the genus Flavivirus, family Flaviviridae, transmitted by Aedes mosquitoes, principally Aedes (Stegomyia) aegypti but also Aedes albopictus and some endemic vectors, like Aedes polynesiensis in the Polynesian triangle [2]. Infection with dengue virus may result in a wide spectrum of clinical manifestations accounting for the classification of the disease as dengue (fever, rash, headache, myalgia, arthralgia, nausea, vomiting) with or without warning signs (abdominal pain, persistent vomiting, clinical fluid accumulation, mucosal bleed) or severe dengue (characterized by plasma leakage) with or without hemorrhage [3]. There are four distinct serotypes of dengue virus (DENV-1 to -4) and infection with one serotype provides no long-term cross-protective immunity against the three others. Based on the sequence of the envelope gene (E), each serotype may be divided into distinct phylogenetic clusters or genotypes [4]. Both epidemiological observations and in vitro studies suggest that some genotypes may have different epidemic potential [5][6][7].
In the South Pacific, the earliest dengue epidemics may have occurred during the nineteenth century [8]. However, dengue only The first cases of severe dengue in the Pacific were reported in the 1970s [11][12][13]. Beginning in 2000, outbreaks of DENV in South Pacific Island Countries (SPICs) have been due to DENV-1 [14][15][16][17][18][19] but in 2008, DENV-4 circulation was detected [20][21][22] for the first time since the 1980's [23,24]. In January 2009, about 20 years after the last reported case, DENV-4 was detected in two members from the same family returning to FP from New Caledonia (NC) where this serotype had recently been detected. Two months later FP experienced a large dengue outbreak.
Here, we report the epidemiological and molecular characteristics of the introduction, spread and genetic microevolution of DENV-4 in FP.

Epidemiological and biological data
Laboratory diagnosis of dengue was performed on sera from dengue suspected patients using ELISA tests for the detection of either DENV NS1 (Platelia TM Dengue NS1 Ag, Bio-Rad Laboratories, Marnes-la-Coquette, France) or anti-DENV IgM (Dengue IgM Capture ELISA, Panbio Diagnostics, Brisbane, Australia). A DENV serotype-specific multiplex real-time RT-PCR using primers and probes designed by Johnson et al. (2005) [25] was performed on sera containing DENV NS1 or from patients who have travelled recently to areas where dengue is known to occur. Data from January 2009 to December 2010 were recorded by the Health Department (Bureau de Veille Sanitaire, Tahiti, French Polynesia) based on reports transmitted by Institut Louis Malardé (ILM, Tahiti, French Polynesia) and the principal hospital in Tahiti (Centre Hospitalier de la Polynésie Française).

Viruses
The details of the viruses employed are shown in Table 1. Briefly, DENV-4 strains collected in FP in the 1980s were obtained from the DENV collection of ILM. At the time they were collected, these strains were amplified by one passage on C6/36 Ae. albopictus cells and cell supernatants were then preserved at 280uC. DENV-4 strains recently collected in FP were obtained from sera positive by DENV-4 real-time RT-PCR performed by the clinical laboratory from ILM. One of these sera was recovered from a patient recently returned from NC [ [22]. Additional E gene sequences of 73 strains were retrieved from GenBank.

Ethics statement
All dengue virus strains were obtained from sera initially sampled for dengue diagnostic and surveillance purposes (under medical prescription by a physician), and archived at ILM. The only information provided to the research laboratory was the sample laboratory ID number, the date of collection and the district/island of collection (no personal information on the patient was provided). The use of biological samples and the collection of information were performed in accordance with the French regulations.

Nucleotide sequencing
Sera or whole blood on filter paper cards (LDA 22 , France) were used as a source of DENV. RNA was extracted from sera using the QIAampH Viral RNA Mini Kit (Qiagen, Germany) or the Easymag extraction system (Biomérieux, France) according to manufacturers' instructions. For whole blood, five filter paper spots per patient were incubated for 20 min at room temperature in lysis buffer from the QIAampH Viral RNA Mini Kit. After centrifugation for 10 min at 8000 rpm, the extraction process was pursued using the supernatant, according to the manufacturer's instructions. Amplification of the E gene was then performed using   [27]. The robustness of the original tree was tested with 1000 bootstrap replications [28]. Sequences available on GenBank and corresponding to strains belonging to all major branches of previously published DENV-4 phylogenies were included in the analysis and selected to maximize the representation of genotypes, countries and years of isolation. Those that were very closely related to one another in the data set were removed from the analysis. Mean distances between and within groups were calculated for all sequences (MEGA5).

Epidemiology
Two patients with DENV-4 infections were detected in January 2009 in Tahiti, Society Islands. The patients were members from the same family recently returned from NC where a DENV-4 outbreak had recently commenced concomitantly with DENV-1 transmission. These were the first cases of infection with DENV-4 detected in FP for more than 20 years. The French Polynesia Public Health and Hygiene Department immediately applied vector control measures by spraying insecticide against adult mosquitoes and destroying breeding sites around the home of these patients. The insecticide treatment was repeated 10 days later. No new DENV-4 cases were detected in Tahiti in the following weeks. However, at the end of February, DENV-4 was detected in three patients living in Taha'a, five in Bora-Bora, one in Tahiti (Society Islands) and one in Ua Pou (Marquesas Islands). None of these patients had traveled out from FP during the two previous weeks. An additional patient from Taha'a with an onset of symptoms at the end of January was found subsequently, by RT-PCR, to also have had a DENV-4 infection. From January to March 2009, the number of patients with DENV-4 infections increased dramatically and DENV-4 had displaced DENV-1 3 months later (Table 2). From January 2009 to December 2010, ILM reported 8375 presumptive cases of dengue to the Health Department, of which 2629 were confirmed subsequently by serological or virological testing. One hundred of these were due to DENV-1 and 763 to DENV-4 infections (Figure 1). During this period, 87 hospitalizations were reported of which three were dengue severe cases.

Molecular epidemiology of DENV-4 in FP
The 64 E gene sequences generated in the present study were combined with 5 E gene sequences from DENV-4 strains collected in the South Pacific between 2008 and 2009 and a data set of 73 E gene sequences available on GenBank representing the global genetic variability of DENV-4 ( Figure 2). The phylogenetic tree shows four distinct clusters named genotype I, II, III and IV. These four clusters have been previously described as genotype I or Southeast Asia, II or Indonesia, III, and IV or Sylvatic or Malaysia [6,[29][30][31]. All sequences of DENV-4 collected in the South Pacific belonged to genotype II. The average distances between genotype II and the three others varied from 0.073 to 0.156, with standard errors (SE) of 0.005 and 0.011 respectively. Within genotype II, viruses clustered into two distinct clades previously defined as IIa and IIb [32]. The average distance between these two clades is 0. Based on the Solomon Island [SB07/EU448462] E gene sequence, 53 sites exhibited a nucleotide (nt) substitution of which 18 were present on more than one strain but didn't lead to an amino acid (aa) substitution ( Table 3). The analysis of the 142 E gene sequences included in this study revealed 79 aa mutated sites, of which one aa substitution (T365I) was specifically encountered within the Oceania group and another (V287I) was only found in the PF80s group (Table 4).

Discussion
The recent emergence of DENV-4 in the SPICs occurred after a decade of active DENV-1 circulation and about 20 years after the previous circulation of DENV-4 in the region [23,24]. This observation is consistent with previous epidemiological data suggesting that, in the region and particularly in FP, sustained transmission of a predominant serotype/genotype occurs following a 20 to 25 years periodic cycle [33,34]. DENV-1 was still circulating in FP when DENV-4 started to be transmitted locally, but from July 2009, by performing DENV serotype-specific RT-PCR on all sera previously tested positive using NS1 test, we only found DENV-4 ( Table 2). Up to November 2011, the only DENV-1 case reported was a traveler just coming back from San-Martin, Caribbean (data not shown). Previous studies already demonstrated that the sensitivity of the Platelia TM Dengue NS1 Ag test depends on the infecting DENV serotype, however the sensitivity of the test was found to be at least identical or higher for DENV-1 compared to DENV-4 [35,36]. Our results suggest that DENV-4 displaced DENV-1. Consistently, the displacement of the DENV serotype previously circulating by the newly introduced serotype is a characteristic that has been constantly observed in FP over time [33,34].Since the second semester of 2008, at the time the transmission of DENV-4 in the South Pacific region was dramatically increasing, dengue surveillance was reinforced in FP. However, despite an active tracking, health authorities did not reach to prevent and control the introduction of DENV-4 in FP. Beside the two DENV-4 cases imported from NC to Tahiti in mid-January, additional introduction events that remained undetected might have occurred, particularly in the Leeward islands (Society Islands) where the firsts cases due to local transmission were reported ( Table 2). Additional DENV-4 cases were then   progressively reported in other FP archipelagos, the spread of the virus being supported by the high frequency of inter-island travel, mostly by plane. Reminding that the DENV-1 epidemic in 2001 also started in the Leeward Islands [23], this place should be considered as a main entry point for new emergent DENV serotypes. Consistently, the Leeward Islands, notably Bora Bora, are major touristic stations in FP. In order to complement the epidemiological data and get a better understanding of the events that might have led to the introduction and spread of DENV-4 in FP, we conducted a phylogenetic analysis on the complete E gene of DENV-4 strains collected recently and in the 1980s in FP. We also included in the study some strains collected in other SPICs in 2008-2009 and additional sequences available on GenBank (Figure 2). The resulting phylogenetic tree displays the four distinct genotypes already described [4,[29][30][31]. Furthermore, it shows that DENV-4 strains collected in FP, both in the 1980s and in 2009-2010, belong to genotype II and are distributed in clades IIa and IIb [4,32]. These two clades are distant from more than 6% of genetic divergence and could be therefore reasonably classified in two distinct genotypes [4]. Clearly, these clades evolved separately from a common ancestor originating from South-East Asia. Interestingly, only clade IIa continued to circulate beyond the year 2000 in Southeast Asia and in the Pacific region. In contrast strains belonging to clade IIb were isolated in Central and South America and in the Caribbean at least until 2008 [37,38]. Indeed, DENV-4 genotype IIb is still predominating in the Americas. The replacement of genotype IIb by the genotypes predominating in Southeast Asia (I, IIa, III, IV) did not occur; although introductions of genotype I were reported in Brazil in 2007 [39]. This contrasts with the situation observed for DENV-2, where the Southeast Asian genotype totally replaced the American genotype previously circulating [40,41].
Our results corroborate previous studies showing that both in the 1980s and in the early year 2000, DENV-4 had been introduced into the region from Southeast Asia [22,24,30]. In addition, we found that all FP strains along with other DENV-4 strains recently collected in the SPICs form a distinct group rooted by the Solomon Island strain [SB07/EU448462], we called Oceania group (Figure 3) (Table 3). Moreover, within each cluster, all FP strains share one additional nt substitution. Particularly, all FP strains in cluster PF/WF exhibit a nt substitution (C1465T) in the genomic E/NS1 junction region which is propitious to the study of evolutionary characteristics [4]. All of these nt substitutions are synonymous, showing that no major mutational event occurred on the E gene from the first emergence of DENV-4 genotype IIa in the Solomon Islands in 2007 to the current circulation in FP. Interestingly, some specific missense mutations resulting in aa substitutions occurred on the E gene of the strains collected in Kiribati and Tonga (Table 4).
A more global analysis of the aa changes on the E protein based on the consensus aa sequence of 142 DENV-4 strains reveals mutations that are specific to strains collected in the South Pacific. Within genotype IIb, all the strains composing the PF80s group, except [PF79/U18438], exhibited the same substitution V287I within the domain I of the E gene [42]. Within genotype IIa, all the strains belonging to the Oceania group show two substitutions (V335I and T365I) already described [22,42]. These mutations occurred within the domain III that mediates DENV binding to cellular receptors and has been proved to be highly immunogenic [42,43]. Our results show that in the 1980s like in the recent emergence of DENV-4 in the region, specific aa substitutions might have been acquired and subsequently conserved on all strains circulating in the SPICs during periods ranging from 3 (2007-2010) to 5 (1983-1988) years. In a previous study, investigating the microevolution of DENV-1 during 6 years of sustained circulation in FP, two specific aa substitutions on the E protein, that appeared during the endemic transmission period and were rapidly stabilized, were suggestive of an adaptation to the mosquito vector [18]. In the present study, the aa substitutions on the E protein found on all DENV-4 strains collected in the SPICs could reflect a founder effect, to be associated with either one or several factors characterizing the ''South Pacific Islands context'': insularity, climate, presence of endemic vector species, small human population sizes and low population flows.