Molecular detection of flaviviruses and alphaviruses in mosquitoes (Diptera: Culicidae) from coastal ecosystems in the Colombian Caribbean

Arboviruses belonging to the genera Flavivirus and Alphavirus were detected in mosquitoes in a rural area of San Bernardo del Viento (Córdoba, Colombia). A total of 22,180 mosquitoes were collected, sorted into 2,102 pools, and tested by generic/nested reverse transcription-polymerase chain reaction. Venezuelan equine encephalitis virus, dengue virus, West Nile virus, St. Louis encephalitis virus, yellow fever virus, and Culex flavivirus were detected and identified by sequencing. The detection of arboviral pathogens in this zone represents possible circulation and indicates a human health risk, demonstrating the importance of virological surveillance activities.


MATERIALS AND METHODS
Study area and mosquito samples -Mosquitoes were collected at eight sites in a rural area called "La Balsa" (9º 21´ 30.97"N, 75º 58´ 37.28" W) in San Bernardo del Viento (Fig. 1).San Bernardo del Viento (approximate population: 32,000) is two meters above sea level, has a mean annual temperature of 30ºC and is bordered by the Sinú River delta.The study area consisted of numerous mangrove trees (across 74 ha), such as "Mangle rojo" (Rhizophora mangle), "Mangle bobo" (Laguncularia racemosa), "Zaragoza" (Conocarpus erectus), "Mangle negro" (Avicennia germinans), and "Mangle piñuelo" (Pelliciera rhizophorae), with a canopy averaging 24 m in height (Rojas & Sierra-Correa 2010).The ground is well irrigated by streams transecting the study site that eventually form ponds and ditches.Small rice fields cultivated by local residents near their homes and inside the mangrove habitats were also inspected.Cattle, chickens, goats, horses, and donkeys were the most common peridomestic animals observed in farm areas close (within approximately 400 m) to the mangroves.Mosquitoes, namely resting and engorged females, were manually sampled.
Adult mosquitoes were collected on seven field trips between September 2011 and October 2013, during seasons (February-April, September-November) of peak bird migration.Each field trip consisted of five to seven days of sampling.Mosquitoes were collected using eight dry ice-baited CDC light traps operated for a 14 h period from 5:00 p.m. to 7:00 a.m.The light traps were placed in the mangrove forest, in small rice fields inside this forest, at the edge of mangroves, in secondary forests, and in peridomestic sites (Fig. 2).Additionally, manual and electric aspirators were used during the day (9:00 a.m.-4:00 p.m.) to collect mosquitoes in their resting sites.
rpm for 30 min.The supernatant was poured into Eppendorf tubes, and total RNA was extracted using an RNeasy Kit (Qiagen, Valencia, CA, USA).Generic (RT-PCR) and nested (RT-nested PCR) were performed using a One-Step RT-PCR kit (Qiagen, Valencia).The protocols used for flavivirus and alphavirus detection were those described by Sánchez-Seco et al. (2001, 2005) (Supplementary data, Table ).The viruses used as positive controls were DENV (New Guinea strain) and VEEV (Argentina strain) donated by Dr Luis Adrián Díaz (Universidad Nacional de Córdoba).Positive samples were sequenced in both directions using an ABI automatic sequencer (Macrogen, Korea), and the resulting sequences were compared with GenBank reference sequences.
Sequence and data analyses -Nucleotide sequences of positive samples were compared with those of reference strains from each viral group (Sánchez-Seco et al. 2001, 2005).Sequences obtained from the first product amplified by RT-PCR (Sánchez-Seco et al. 2005) were aligned using Bioedit v7.0 (Hall 1999).The neighbour-joining algorithm (Saitou & Nei 1987) and Kimura 2-parameter model (Nei & Kumar 2000) were used to construct a dendrogram in MEGA v6.0 software (Tamura et al. 2013).Sequences were compared with those available from GenBank using BLASTN (Altschul et al. 1997).Minimum infection rates (MIRs) were calculated as follows: MIR = number of positive pools by morphospecies/total number of morphospecies tested from that site × 1000 (Walter et al. 1980).
Ethics -Mosquitoes were collected according to the guidelines of Colombian decree number 1376 (2013), which regulates the collection of wild specimens of biological diversity for non-commercial research.No specific permits were required for this study.

RESULTS
A total of 2102 pools, representing 14 morphospecies, were screened.Mosquitoes were identified based on morphology and DNA barcoding.A total of 268 voucher specimens of mosquito species were deposited in the Francisco Luis Gallego Entomological Museum.
Sequences from alphaviruses and flaviviruses were amplified from 30 of the 2102 mosquito pools (1.43%) (Table).Mosquitoes belonging to the genus Culex according to morphology were collected most frequently.Deinocerites (De.atlanticus aff.), which were present in mangrove forests with numerous crab holes, were also common.
DNA barcodes were obtained for mosquitoes in the genus Culex; however, closely related species within the subgenera Culex and Melanoconion could not be distin- guished by their sequences, and an exact species could not be assigned to specimens other than Cx.(Culex) quinquefasciatus (KT766432-KT766453).
Nested RT-PCR was used to identify viruses in mosquito pools.However, generic RT-PCR with primers targeting flaviviruses (1.1 kb, NS5 gene) and alphaviruses (500 b, nsP4 gene) were also obtained and sequenced, for consensus alignments of 515 and 420 nt, respectively.These sequences were used for identification by BLASTN search and neighbour-joining dendrograms.
The MIR of SLEV, YFV, and WNV calculated for each morphospecies was low; the notable exceptions were CxFV in Cx. quinquefasciatus and DENV2 in Ae. aegypti.VEEV was detected in several mosquito species (Table ).
VEEV was mostly detected in mosquitoes collected in mangrove forests or in neighbouring ecosystems (mangrove edge, 11 pools; rice crops in mangrove forest, six pools; and mangrove forest, five pools).SLEV, WNV, and YFV were detected only in mangroves.DENV2 was present in peridomestic sites, and CxFV was only detected in secondary forests.

DISCUSSION
The presence of five arboviruses (genera Alphavirus and Flavivirus) indicates the risk for emergence or re-emergence of these pathogens in San Bernardo del Viento.
Many VEEV isolates have been obtained previously from the Magdalena valley (Ferro et al. 2003) and Norte de Santander (Ferro et al. 2008) (Ferro et al. 2008).Most strains of VEEV isolated from the Culex subgenus Melanoconion in Colombia have been categorized as enzootic subtype ID (Weaver et al. 1996(Weaver et al. , 2004)).Our sequences grouped with VEEV subtype ID strain 3880, which was isolated from a human case in Canito, Panama (Brault et al. 2002) In our study, Culex spp., Ps. confinnis, Ae. scapularis, and Ma.titillans were found to be infected with VEEV.These species are likely enzootic vectors of the virus, as has been reported in various geographic regions of Colombia and in nearby countries (Groot 1964, Hastriter & Lawyer 1998); however, the local vector status of these four species is unknown.Our study is the first report of a natural VEEV infection in De. atlanticus aff.This morphospecies was associated with crab holes in mangroves forests that are near the beach and in coconut groves.The adults of this species are reported to have nocturnal or crepuscular habits, resting by day in the upper parts of crab holes (Belkin & Hogue 1959).Their biting activity may extend to humans (Adames 1971).However, many essential aspects of De. atlanticus aff.biology remain unknown.In the Guajira department, De. atlanticus aff. is the dominant mosquito species.It has been collected inside houses and is a vector of VEEV in this geographic area (Ferro et al. 2015).
SLEV and WNV are emerging flaviviruses that have been detected previously in humans, equines, and birds in the Colombian Caribbean region (Mattar et al. 2005a, b, 2011, Goéz-Rivillas et al. 2010, Osorio et al. 2012).Both viruses detected in this study were molecularly characterised based on the NS5 and envelope genes and were found to be closely related to genotypes associated with attenuated virulence (WNV Texas genotype) (Hoyos-López et al. 2015a) and little activity in human populations (SLEV genotype IV) (Hoyos-López et al. 2015b).
The CxFV detected in this study in pools of Cx. quin-quefasciatus is the first evidence of this virus in Colombia, although it has been reported in the same species in other countries (Hoshino et al. 2007, Morales-Betoulle et al. 2008, Cook et al. 2009, Farfan-Ale et al. 2009, Kim et al. 2009, Huanyu et al. 2012, Machado et al. 2012).
A strong ecological association between CxFV and WNV was observed in Chicago, USA (Newman et al. 2011), suggesting that super-infection with WNV during CxFV infection may interfere with secondary viral infection with a similar virus (Farfan-Ale et al. 2009).However, experimental infections of Cx. quinquefasciatus with WNV strain Guatemala and CxFV strain Izabal do not support this hypothesis, because prior infection with CxFV had no significant impact on WNV replication, infection, dissemination, or transmission by this mosquito (Kent et al. 2010).It is possible that natural infection with CxFV affects the vector competence of Cx. quinquefasciatus through vertical transmission and persistence in mosquito progeny (Bolling et al. 2012), in- fluencing the ecology of viruses such as WNV and SLEV and even enhancing WNV transmission (Kent et al. 2010).Interestingly, Cx. quinquefasciatus was collected only in secondary forests and was absent from peridomestic areas.Possible explanations include the low density (approximately 30) of human households close to mangroves, saline water, and availability of few breeding sites (Forattini et al. 2002b, Correia et al. 2012).Potential breeding sites (many ponds and livestock watering devices) were found near the forest collection site.
In Colombia, vectors that participate in sylvatic YFV transmission include Haemagogus equinus and Haemagogus janthinomys (Morales 1968).Natural infection of Haemagogus sp.(Splendens section) with YFV may suggest an epidemiological risk for humans working in mangroves, as this species readily bites humans and is present in high densities in coastal mangrove forests (Arnell 1973).There have been previous reports of experimental and natural infection of YFV (Galindo et al. 1956, de Rodaniche et al. 1957) and SLEV in Haemagogus from Panamá (Kramer & Chandler 2001).In San Bernardo del Viento, cases of yellow fever have not been recorded; however, they are reported annually in the Caribbean region of the Cesar, Guajira, and Magdalena departments (Rojas-Álvarez 2008).
DENV2, and predominantly the American/Asian genotype (subtype IIIb), is frequently detected in patients in Colombia (Méndez et al. 2012).Although there are few cases in San Bernardo del Viento, with approximately 4-6 patients reported between 2011 and 2014 (Secretaría de Salud Municipal, San Bernardo del Viento, personal communication), asymptomatic human cases and clinical underreporting is likely common in other regions of Colombia (Méndez et al. 2006).
We report the first detection of flaviviruses and alphaviruses circulating in coastal mangroves and neighbouring ecosystems in Colombia, indicating a potential health risk to humans living or working in these zones.Interestingly, in the present study, the mangrove edge harboured the most positive pools of VEEV.Some authors have suggested an important role for the transitional zone between two adjacent ecological systems (e.g., mangroves and pastures) or "ecotones" in the transmission of emerging viruses and arboviruses (Estrada-Franco et al. 2004, Despommier et al. 2006).Mosquitoes that vector several viruses (i.e., Rift Valley fever, VEEV, YFV) have been found to be abundant in ecotones (Olival & Daszak 2005, Despommier et al. 2006), and they may have the potential to adapt and exploit new breeding sites in human-dominated ecosystems (Diallo et al. 2012).The presence of Culex subgenera Melanoconion and Culex may be important to the emergence and/or re-emergence of WNV, SLEV, and VEEV.Mangrove ecosystems, may be specifically acting as a source of emerging pathogens that infect nearby human populations (Holmes 2008).Furthermore, ecotones have been identified as potentially important for host seeking and host switching (Burg 2001, Méndez et al. 2001, Hoberg et al. 2002).Because of the high concentrations of reservoirs and potential hosts, jumps to alternative hosts or mosquitos with different levels of vector competence may be favoured (Barrera et al. 2002, Deardoff et al. 2011).A key factor is that RNA arboviruses have high rates of evolution that may facilitate the generation of progeny with variable fitness or virulence in alternative hosts (Holmes 2008, 2009, Hoyos et al. 2012).
Arbovirus detection is just one aspect of vector incrimination, and more data are needed to identify potential vectors, define vector competence, and identify specific habitats, biting behaviour, and blood feeding patterns of possible vectors.All of these aspects should be considered in a context of global climate change, which is another factor that may contribute to the emergence of new diseases (Naicker 2011, Butler 2012) and may be particularly important in Colombia, which experiences the effects of climatological oscillations such as ENSO (Poveda et al. 2001, Singh 2013).These abiotic changes may contribute to an increase in the size of mosquito populations and in cases of vector borne diseases (Poveda et al. 2000).These perspectives may contribute to the development of more effective disease prevention and control strategies.