Background

Unlike mosquitocidal bacteria and viruses, ascomycete fungi can infect and kill insects without being ingested. As with chemical insecticides, tarsal contact alone is sufficient to kill mosquitoes [1]. Despite intensive efforts to develop entomopathogenic fungi as biocontrol agents against malaria vectors, the strains under investigation have not met expectations due to their poorer efficacy relative to cheaper chemical insecticides [2]. The United States Department of Agriculture (USDA) ARSEF collection (the world’s largest collection of entomopathogenic fungi) has more than 12,000 isolates of insect pathogenic fungi. Of these, only 156 are from sub-Saharan Africa (South Africa and Benin are the source of 40 and 36 isolates, respectively), with none from Burkina Faso. The mosquitocidal activity of Metarhizium has been enhanced by engineering them to express insect-selective neurotoxins [3,4,5], and a transgenic strain of Metarhizium pingshaense is being evaluated in semi-field trials in Burkina Faso [5]. We speculate that future development of transgenic fungi worldwide will preferentially use local isolates as these may be better adapted to kill local mosquitoes and survive harsh local conditions (i.e. rainy season heat, sunlight and humidity) than exotic strains. However, the distribution and properties of indigenous Burkinabe Metarhizium spp. have not been characterized. The first objective of this study was to prospect for the presence and distribution of local Metarhizium strains. As well as prospecting mosquitoes, we also sampled rhizosphere soils (i.e. the soil in the vicinity of plant roots that is influenced by root secretions), as some Metarhizium spp. are abundant in the rhizosphere and may function as symbionts promoting plant growth. The plant-beneficial effects of Metarhizium species correlate with their association with roots and are mediated via plant hormones [6]. The second objective was to evaluate the pathogenicity of local Metarhizium isolates against wild-caught, insecticide-resistant Anopheles coluzzii. Finally, we also assessed the pathogenicity of the local isolates against American cockroaches and honeybees as representative non-target or beneficial species.

Methods

Fungal collection, isolation and morphological identification

Collections were carried out on a monthly basis during the 2015 rainy season (from July to September) from plant roots and wild-caught mosquitoes. Our three collection sites were the Kou Valley (11°23'N, 4°24'W), a rice crop area; Bana (11°9'41"N, 4°10'30"W), a savanna and forested area; and Soumousso (11°04'N, 4°03'W), a savanna and corn crop area (Fig. 1). One hundred and fifty-five plants were sampled from these three different agro-ecological sites. We followed the protocol described in [7] to collect rhizosphere soil and isolate fungi. The fungal selective medium contained 42 g potato-dextrose agar, 0.5 g chloramphenicol and 0.6 g cetyl trimethylammonium bromide per liter.

Fig. 1
figure 1

Rhizosphere and mosquito collection sites

Full size image

Overall, 300 mosquitoes were collected from 3 types of resting sites (inhabited houses, abandoned houses and outdoor piles of wood). Mosquitoes were brought to the IRSS/Centre Muraz insectary, where they were fed on 6% sterile glucose ad libitum. Approximately 22% of collected mosquitoes (67 mosquitoes) died within 2 weeks and were plated on selective medium for fungal isolations.

Fungal isolates from rhizospheres or mosquitoes were identified using macro-morphological characters, such as conidiogenesis, estimation of radial growth, spore color and mycelia texture of the isolates on PDA media according to Humber [8]. In addition, we used microscopic morphology to identify Metarhizium spp. spores as described by Fernandes et al. [9]. Met_S10 and Met_S26 were confirmed as Metarhizium pingshaense through amplification and Sanger sequencing of the intron-rich region of translation elongation factor 1-α [10].

Fungal virulence on mosquitoes, honeybees and cockroaches

Initial screens on mosquitoes revealed two promising isolates (Met_S10 and Met_S26) isolated from mosquito cadavers from Soumousso and Bana, respectively, that readily grew on PDA and were highly virulent (Additional file 1: Table S1): these strains were therefore chosen for further characterization.

Bioassay on mosquitoes

For bioassays, we used An. coluzzii adult mosquitoes reared from larval collections at the Kou Valley, Burkina Faso. Mosquitoes from this area are known to be highly resistant to multiple insecticides [5, 11]. We carried out bioassays with local M. pingshaense isolates Met_S10 and Met_S26. A M. pingshaense strain that has been used as the foundation for development of transgenic mosquito control technologies was used as a positive control; this strain was engineered to constitutively express red fluorescent protein (RFP) [5]. Expression of RFP provides a fluorescent tag for following infection processes without altering virulence. We used an atomizer protocol for infections, as described previously [12]. Three serial concentrations were used: 1 × 108; 1 × 107; and 1 × 106 conidia/ml. We confirmed that this inoculation technique was able to deliver a repeatable inoculating dose (mean ± SE): 276 ± 16 spores per mosquito with 1 × 108; 211 ± 13 spores per mosquito with 1 × 107 spores/ml; and 44 ± 3 spores per mosquito with 1 × 106. Mortality was counted twice daily over two weeks.

Bioassay on non-target insects

We bioassayed Met_S10, Met_S26 and Met_RFP against a breeding line of honeybees, Apis mellifera adansonii (Latreille, 1804), as well as American cockroaches, Periplaneta americana (Linnaeus, 1758) caught in households from Soumousso. Spore doses were 1 × 108, 1 × 107 or 1 × 106 conidia/ml, as described previously [5]. Following treatment, insects were kept in our insectarium at 25.3 ± 1 °C and 70 ± 10% relative humidity. Mortality was counted twice daily over two weeks.

Results and discussion

Metarhizium spp. were isolated from rhizosphere soil samples across 3 sample sites: the Kou Valley, Bana and Soumousso. From the Kou Valley and Bana, we isolated 362 and 306 soil samples, respectively. Metarhizium spp. comprised 28.71% (n = 56) of the isolates from Bana and 30.72% (n = 94) of the total isolated fungi from the Kou Valley. We isolated 152 fungal strains from Soumousso; of these, 113 (74.34%) were Metarhizium, with a mean of 1.18 isolates/gram of soil (Additional file 2: Table S2). Soumousso is a savanna and corn crop area, and the higher proportion of Metarhizium fungi is consistent with previous studies that reported a strong association between Metarhizium spp. and soils from cultivated habitats, particularly field crops [13,14,15].

Isolates of Metarhizium spp. represented ~1% (8/801; 3 isolates from Culex spp. and 5 isolates from Anopheles gambiae (sensu lato) of the fungi isolated from mosquitoes. Fifteen colonies of Beauveria spp. were isolated on mosquitoes (5 isolates from Aedes aegypti and 10 isolates from Anopheles gambiae (s.l.) at Soumousso). Trichoderma was the predominant genus at all sites being isolated from 56% (Vallée du Kou) to 79% (Soumousso) of mosquitoes (Additional file 2: Table S2). However, two Metarhizium isolates (Met_S10 and Met_S26), collected from Anopheles gambiae (s.l.), in an inhabited house in Soumousso and in a woodpile in Bana, respectively, were more virulent against mosquitoes than other isolates, including those from rhizospheres (Additional file 1: Table S1). At 1 × 108 and 1 × 107 conidia/ml, both strains achieved lower LT50 than Met_RFP (LT50 of ~6 days) [16]. At the highest concentration (1 × 108 conidia/ml), the LT80 of Met_S10 (5.67 ± 0.17 days) was significantly lower than both Met_26 (LT80 = 7.50 ± 0.29 days; Welch t = -5.5, df = 3.2, P = 0.01) and Met_RFP (7.17 ± 0.17 days; Welch t = -6.364, df = 4, P = 0.003). At the lowest concentration (1 × 106 conidia/ml), Met_S10 still had a significantly (Welch t = -5.1962, df = 3.2, P = 0.011) lower LT80 (7.00 ± 0.29 days) compared to Met S26 and Met_RFP, which both had LT80’s of 10 days (Fig. 2, Table 1). At intermediate concentrations, all strains achieved 80% mortality, which is the threshold value from the World Health Organization Pesticide Evaluation Scheme (WHOPES) for successful control with insecticides [17]. Thus, our results revealed higher virulence for the native isolate Met_S10, against wild-caught, insecticide-resistant Anopheles coluzzii. The virulence of these isolates to mosquitoes is also higher than isolates from Benin and in Kenya where Metarhizium anisoplae strains were originally isolated from a white fly, Trialeurodes vaporariorum [16, 18].

Fig. 2
figure 2

Survival curves of mosquitoes infected with Burkina Faso Metarhizium pingshaense isolates at different concentrations: C1, 1 × 108 conidia/ml; C2, 1 × 107 conidia/ml; C3, 1 × 106 conidia/ml

Full size image
Table 1 LT80s and grouping LT80 values for Anopheles coluzzii adults treated with Burkina Faso local Metarhizium pingshaense strains (Met_10 and Met_26) compared with wild type Metarhizium pingshaense expressing red fluorescent protein (Met_RFP) at three different concentrations
Full size table

We bioassayed honeybees and cockroaches with the local strains and Met_RFP. However, even at the highest spore dosage (1 × 108 conidia/ml), these fungi did not significantly increase mortality compared to controls containing no conidia (Table 2). Fewer than 5% of honeybees and cockroaches died during the bioassays, and no mycosis was observed on any cadavers. This is in agreement with previous studies that report Met_RFP is a specialist to Culicidae [5]. The host ranges of different Metarhizium strains are chiefly controlled by recognition events on the cuticle [19], and the cuticles of honeybees, cockroaches and mosquitoes would likely have many topographical and chemical differences.

Table 2 Two week-survival and grouping survival values for non-target insects (Honeybees and Cockroachs) treated with Burkina Faso local Metarhizium pingshaense strains (Met_10 and Met_26) compared with wild type Metarhizium pingshaense expressing red fluorescent protein (Met_RFP) at three different concentrations and a control (0.01% Tween)
Full size table

Despite being more virulent than other WT Metarhizium strains, the Burkinabe Anopheles-derived isolates are still significantly less effective than transgenic strains expressing arthropod toxins [5]. However, our results suggest that these native Burkinabe Metarhizium strains would make attractive candidates for transgenic virulence enhancement and subsequent use as transgenic biocontrol agents.

Conclusion

Native fungal isolates may offer a superior alternative to introducing a foreign biocontrol strain, as they may be better adapted to both kill local mosquitoes and survive local conditions. There are also regulatory and ecological advantages to using strains already present in the country or in the ecosystem. This study provides a promising precedent for isolating local Metarhizium strains for application in mosquito biological control, and it lays a foundation for future transgenic biocontrol projects in Burkina Faso.