Evidence of circulation of Orthobunyaviruses in diverse mosquito species in Kwale County, Kenya

This study was carried out in Lunga Lunga sub-county, Kwale County (Fig. 1). Kwale County covers an area of 8270.2 km² and comprises of four sub-counties: Matuga, Msabweni, Kinango and Lunga Lunga. Lunga Lunga sub-county is located six kilometres from the Kenya border with Tanzania. The population of Lunga Lunga sub-county based on the 2019 census is approximately 198,423 persons. Vanga Island is a coastal fishing settlement, while Jego and Tsuini represent a rural village. On the other hand, Lunga Lunga is an urban area. The economic activities in this sub-county include, fishing, farming, sand harvesting and small-scale trading [34].

Mosquitoes were sampled in Vanga Island from 27th July–1st Aug 2017. A second sampling was done in Vanga Island from the 15th–17th Sep 2018, in Jego which is a rural setting on the Vanga mainland from 14th–19th Sep 2018 and in Tsuini from 18th Sep 2018. A third sampling was conducted in Lunga Lunga, from the 13th–20th Dec 2019. Lunga Lunga is an urban commercial centre located on the mainland and serving as the commercial center for the sub-county as well as the border crossing into Tanzania. The last site to be sampled was Tsuini on 20th Dec 2019. Adult mosquitoes were collected in sites randomly selected within the locations, while larvae were collected in Vanga Island and Jego from the 14th–19th Sept 2018, a dry season while in Lunga Lunga, the survey was done from 13th–20th Dec 2019, a wet period.

In households that were accessible, water-holding containers found indoors and outdoors were inspected for mosquito immatures. Samples were collected from each productive container using ladles or Pasture pipettes. A score was given to every container inspected as either wet negative or wet positive depending on presence of and number of Ae. aegypti immatures found [37]. The immature mosquitoes collected were reared to adulthood and identified to species [7] so that only Ae. aegypti mosquitoes were used to calculate larval indices. The following larval indices were calculated to determine the risk of arbovirus transmission,House index (HI): percentage of houses where breeding larvae and/or pupae where found. Container index (CI): percentage of water-holding containers infested with larvae or pupae. Breteau index (BI): number of positive containers per 100 houses inspected [29].

Adult mosquitoes were collected using CO₂-baited CDC light traps (John W Hock) and BG-Sentinel traps (Biogents). The CDC light traps were hung randomly on selected sites near houses or at the edges of the compound, 10 traps per night. Each trap was baited with 0.5 kg of CO_2, hung from dusk to dawn and retrieved in the morning. The BG- Sentinel traps were set from morning to evening outside houses and retrieved in the evening [23]. At a temporary laboratory set up in the field station, the mosquitoes collected from each of the traps were knocked down using triethylamine. The mosquitoes were sorted and identified morphologically using mosquito identification keys and pooled (≤ 25 mosquitoes per pool) by species, sex and collection site [10, 14]. Each mosquito pool was stored in a 1.5-ml cryogenic vial. The cryogenic vials were stored in liquid nitrogen and transferred to the level-2 biosafety laboratory based at the Kenya Medical Research Institute’s (KEMRI) centre for virus isolation.

The mosquito pools obtained were homogenized using Minimum Essential Medium supplemented with Foetal Bovine Serum (Gibco), L-Glutamine and antibiotics (10,000 units penicillin, 10 mg streptomycin and 25 μg amphotericin B per ml-Sigma) [23]. Homogenates were clarified by centrifugation at 12000 rpm for 10 min and the resultant supernatants inoculated in 24 -well plates of Vero cells (CCL-81™) in supplemented Minimum Essential Medium [27]. The cultures were incubated at 37 °C and monitored for cytopathic effect (CPE) daily for 14 days. Cultures showing CPE were harvested and viruses identified by RT-PCR and sequencing.

RNA was extracted from cell culture isolates using Trizol^®LS Chloroform method. The RNA extract was transcribed into cDNA using First Strand cDNA synthesis kit (Invitrogen) with random hexamers, followed by Polymerase Chain Reaction (PCR) using Amplitaq Gold PCR mastermix (Applied Biosystems). Amplification by PCR used universal arbovirus primers targeting the genus of Flavivirus, Orthobunyavirus and Alphaviruses [27].

Preliminary characterization of the isolates was performed by Sanger sequencing. Amplicons were cleaned using DNA Clean & Concentrator Kit (ZymoResearch, US). Chromatogram files were prepared and edited to generate consensus sequences using BioEdit v7.2.5 [15].

A subset of the samples was also prepared for whole genome sequencing. RNA extracts were used as input to perform library preparation by using Truseq mRNA Library Prep kit (Illumina, San Diego, CA, USA), following the manufacturer’s recommended protocol which was modified to exclude the mRNA clean-up steps [5]. The prepared libraries were sequenced on an Illumina Miseq platform (Illumina, San Diego, CA, USA) using a 2 × 300 base paired-end reads. Sequence analysis was performed by making use of NGS Mapper v1.5 pipeline (ngs_mapper). The pipeline performs a number of sequential steps on the raw sequence reads, which includes; adapter trimming, raw sequence read cleanup, reference-based mapping and assembly and finally consensus sequence generation. The sequences reported in this study are available in GenBank under accession numbers MW314021-MW314035.

Sequences generated in this study were analyzed using NCBI Blast and phylogenetic analysis was performed using MEGA7 [21]. Sequences falling under the different Orthobunyavirus serogroups were downloaded from Genbank and combined with those generated in this study. Alignment of the individual segments was achieved with Muscle v6 embedded in MEGA7, and phylogenetic analysis was performed on each of the segments using the Maximum Likelihood approach in MEGA7. Reassortment was evaluated by examining the topology of the phylogenetic trees generated for each of the individual segments.

All analyses were performed using R version 3.6.1 [30]. For larval data, we summarized the larval indices separately for each site. For adult mosquito data, we summarized the collections by collection method (i.e., CDC-light traps and BG sentinel traps) and by site. We also produced mosquito richness and relative abundance of species separately for each site and combined. Data on proportions (or percentages) were compared across the sites using Chi Square test. Multiple comparison of proportions was adjusted for using Holm’s method. All tests were performed at 5% significance level.

In Vanga Island, 164 houses were sampled, of which 5 had positive water containers for Ae. aegypti larvae/pupae with a HI of 3%. Six out of a total of 903 containers inspected indoors and outdoors were positive, with a CI of (0.66) and BI of (3.66). In Jego village, 2 out of 84 inspected houses were had positive water containers, with a HI of 2.4%. A total of 414 containers were inspected indoors and outdoors and 2 were positive, with a CI of (0.48) and a BI of (2.4). In Lunga Lunga an urban area, 118 houses were sampled of which 26 were positive giving a HI of 22.03%. A total of 413 containers were inspected indoors and outdoors and 35 were positive, with a CI of (8.5) and a BI of (29.7) (Table 1).

In the three sites, a total of 1,730 containers were inspected indoors and outdoors (Table 2). The most abundant containers were the buckets 42.7%, jerricans 32.9%, plastic drums 14.2% and plastic basins 5.8%. The most productive containers – measured in terms of proportions of inspected containers which are positive – were the concrete tanks (44.4%), plastic tank (22.2%), claypot (13.3%), plastic drums (8.9%), plastic basins (4%), jerricans (1.2%) and buckets (0.3%) (Fig. 2).

A total of 29,447 mosquitoes, belonging to 7 genera and 36 species, were collected by CDC light traps (20,236) and by BG sentinel traps (9,211). After identification and pooling, 1,950 pools were obtained. The greatest diversity was in the genus Aedes that recorded 13 species, followed by Culex (9), Anopheles (7), Coquilletidia (3), Mansonia (2), Eretmapodite (1) and Filcabia (1). The species richness was 33 in Lunga Lunga, 30 in Tsuini, 28 in Vanga and 28 in Jego. Seventeen species were recorded in all four sites, while 9 were recorded in single sites only. More Aedes, Culex and Mansonia were collected in Vanga Island than other sites while more Anopheles species were collected in Tsuini. Mosquito abundance varied significantly across the sites (Chi Sq. = 892.8, df = 3, p < 0.001): Vanga recorded significantly larger numbers (50.2%) than all other three sites (p < 0.001). Tsuini recorded higher abundance (19.2%) than Lunga Lunga (16%; p = 0.018) and Jego (14.6%; p < 0.001). There was no significant difference in abundance between Lunga Lunga and Jego (p = 0.229). Table 3 exhibits the relative abundance of these mosquitoes separately for each site and all sites combined. The table shows that Cx. quinquefasciatus was the overall most abundant species sampled in all the sites combined (17.8%), followed by Ae. pembaensis (11.9%), An. funestus (11.2%), Cx. annuloris (9%), Mn. uniformis (8.9%) and Ae. tricholabis (6.3%). The least overall sampled species were Ae. africana, Ae. vittatus, Coq. fuscopennata, Coq. metallicus and Cx. tigripes.

There was no virus isolated from immature mosquitoes. However, 12 isolates were obtained from the 1950 pools of adult mosquitoes that were tested. Ten of the pools tested positive for Bunyavirus using the California-Bunyamwera serogroup primers [22] while the two other isolates remained unidentified by Flavivirus, Bunyavirus and Alphavirus universal primers. Sanger sequencing of the successfully amplified isolates resulted in generation of sequences that showed approximately 99% sequence homology to Bunyamwera and Ngari viruses. Considering the primers we used only targets a region within the small segment (S segment) of the Bunyamwera serogroup, it was unclear whether the identity of the virus was a Bunyamwera or Ngari virus. Therefore, we performed whole genome sequencing on four of the Bunyavirus isolates including those obtained from Ae. pembaensis (2), Cx. quinquefasciatus (1), and Culex spp (1) as well as one isolate from the unidentified group that was obtained from Cx. zombaensis. Full genome sequences were obtained from all the 5 isolates, including all the three segments of the individual isolates. The depth of coverage across the entire segments of all the 5 genomes was well covered, with the lowest being 24X (Fig. 3). All the other coding regions of the genomes had a high coverage (Fig. 3). All the three segments of the 4 Bunyamwera isolates showed a high similarity to the Bunyamwera virus (NC_001925, NC_001926, NC_001927) with amino acid (aa) similarity of approximately 98.6% (M segment), 99.7% (L segment) and 99.5% (S segment). This, therefore, confirms the identity of this virus as a Bunyamwera virus. The virus from the unknown isolate showed a high similarity to Nyando virus (KJ867197.1, KJ867198.1, KJ867199.1) with percent (aa) similarity of 100% (S segment), 98.05% (M segment), and 98.94% (L segment). The sequences generated in this study were aligned and compared with 23 other sequences belonging to the different serogroups including Bunyamwera, Wyeomyia, California, Bwamba, Nyando and Simbu serogroups which were obtained from Genbank. Maximum likelihood phylogenetic analysis using the individual segments of this set of sequences placed the 4 isolates into the Bunyamwera serogroup and one isolate into the Nyando serogroup, particularly with close similarity to the Bunyamwera virus and Nyando virus respectively. There was no incongruities observed on the different phylogenies generated based on the individual segments (Fig. 4).