Geosmin Attracts Aedes aegypti Mosquitoes to Oviposition Sites

Abstract

Geosmin is one of the most recognizable and common microbial smells on the planet. Some insects, like mosquitoes, require microbial-rich environments for their progeny, whereas for other insects such microbes may prove dangerous. In the vinegar fly Drosophila melanogaster, geosmin is decoded in a remarkably precise fashion and induces aversion, presumably signaling the presence of harmful microbes [1]. We have here investigated the effect of geosmin on the behavior of the yellow fever mosquito Aedes aegypti. In contrast to flies, geosmin is not aversive but mediates egg-laying site selection. Female mosquitoes likely associate geosmin with microbes, including cyanobacteria consumed by larvae [2], who also find geosmin-as well as geosmin-producing cyanobacteria-attractive. Using in vivo multiphoton calcium imaging from transgenic PUb-GCaMP6s mosquitoes, we show that Ae. aegypti code geosmin in a qualitatively similar fashion to flies, i.e., through a single olfactory channel with a high degree of sensitivity for this volatile. We further demonstrate that geosmin can be used as bait under field conditions, and finally, we show that geosmin, which is both expensive and difficult to obtain, can be substituted by beetroot peel extract, providing a cheap and viable potential mean for mosquito control and surveillance in developing countries.

Figure 1.. Geosmin Mediates Egg Laying Selection in Aedes aegypti

(A) Geosmin has, to the human nose, the distinct smell of wet soil and is produced by a wide range of micro-organisms, both terrestrial and aquatic. Photo: M. Stensmyr (B) Plastic trays lined with filter paper used in oviposition experiments. On top, water with geosmin added is shown; on the bottom, water only control is shown. In the inset, close-up of a cluster of Aedes eggs in the geosmin-containing tray is shown. (C) Oviposition indices (OI) of wild-type (WT) (20 mosquitoes per trial; n = 6) and Orco⁵ mosquitoes (20 mosquitoes per trial; n = 6 trials) from a binary-choice test between water and water spiked with geosmin. Total number of eggs is as follows: WT 4,036, Orco⁵ 582; WT geosmin 593 ± 143 eggs, control 191 ± 58; Orco⁵ geosmin 167 ± 37, control 124 ± 23 (mean ± SEM). The edges of the boxes are the first and third quartiles, thick lines mark the medians, and whiskers represent data range. Preference was tested with one-sample Wilcoxon test, theoretical mean 0. Star denotes significantly different from 0; p < 0.05. (D) Feeding indices (FIs) from a CAFE assay of WT (n = 10) and Orco⁵ mosquitoes (n = 10) given a choice to feed from two capillaries with sucrose water (10%), one of which in addition containing geosmin (10⁻¹, 10⁻³, or 10⁻⁵). Boxplots and statistics as per (C) are shown. (E) Probing index (PI) from WT mosquitoes (20 mosquitoes per trial; n = 5) in a constrained contact assay over 6 min, provided with a choice to approach and probe two hands (from the same individual), one of which scented with geosmin (10⁻¹, 10⁻³, or 10⁻⁵). Shaded line indicates SEM. Statistics as per (C) are shown. (F) Schematic of the larval behavioral assay. Dashed lines denote the two zones in which time spent was measured. (G) Sample tracks of WT larvae with antennae (above) and with antennae removed (below). (H) Response indices of WT larvae with antennae (n = 44), without antennae (n = 33), and Orco⁵ mutants (n = 30) toward geosmin (10⁻⁵). Boxplots and statistics as per (C) are shown.

Figure 2.. Geosmin-Producing Cyanobacteria Mediate Egg Laying and Larval Attraction

(A) Flame ionization detection (FID) traces from a gas chromatography-mass spectrometry analysis of head space volatiles emitted by two strains of cyanobacteria. (B) Oviposition indices (OIs) of WT (20 mosquitoes per trial; n = 6) and Orco⁵ mosquitoes (20 mosquitoes per trial; n = 4) from a binary-choice test between growth medium and growth medium with a cyanobacteria strain (PCC6506) producing geosmin. Total number of eggs is as follows: WT 4,191, Orco⁵ 1,743; WT PCC6506 532 ± 223 eggs, control 166 ± 103; Orco⁵ PCC6506 242 ± 110, control 194 ± 92 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. (C) OI of WT mosquitoes (20 mosquitoes per trial; n = 4) from a binary-choice test between growth medium and growth medium with a cyanobacteria strain (PCC8913) not producing geosmin. Total number of eggs is as follows: 1,185; PCC8913 157 ± 40 eggs, control 140 ± 22 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. (D and E) OIs of WT mosquitoes (20 mosquitoes per trial; D, n = 4; E, n = 10) from a binary-choice test between PCC6506 and PCC8913, without (D) or with (E) geosmin (~10⁻⁷) added to the latter. Total number of eggs is as follows: (D) 5,112, (E) 6,445; (D) PCC6506 308 ± 80 eggs, PCC8913 203 ± 62; (E) PCC6506 822 ± 234, PCC8913+geosmin 789 ± 75 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. (F) Response indices from larvae (WT, n = 27; Orco⁵, n = 32) given a choice between agar mixed with growth medium and agar with PCC 6506. Boxplots and statistics as per Figure 1C are shown.

Figure 3.. Geosmin Elicits Robust Responses in the Aedes Antennae and AL

(A) Schematic of the electroantennogram (EAG) preparation (top). EAG responses from WT and Orco mutants to stimulation with 10⁻³ dilutions of geosmin, 1-octen-3-ol, and octanoic acid are shown. On the right, representative recordings are shown. Vertical scale bar: 0.25 mV; horizontal scale bar: 5 s. Statistical difference was measured via a Student’s t test. Star denotes significant difference (p < 0.05). (B) Schematic of the two-photon setup used to record calcium dynamics in the mosquito antennal lobe (AL). (C) Pseudocolor plot from a single preparation of ΔF/F₀ calcium responses (0 to 1 scale) to geosmin (10⁻³ dilution), at a depth of 75 mm from the surface of the AL. Geosmin evoked a strong response in one glomerular region of interest (highlighted in white). (D) Non-responsive AL glomeruli (gray) and the geosmin-responsive glomerulus (green; the third posterodorsal glomerulus [PD3]) tentatively registered and mapped to an AL atlas and cross-referenced to a previously published atlas [19]. (E) Glomerular responses (ΔF/F₀) to geosmin characterized at five depths (15, 30, 50, 75, and 90 μm) from the ventral surface of the AL. Each trace is the mean of one glomerulus; the PD3 response is shown in green. Vertical scale bar: 0.4%. Grey bar denotes stimulus duration (2 s). (F) Responses to geosmin across all sampled glomeruli; only the PD3 glomerulus (bar in green) showed significant calcium dynamics to geosmin compared to the solvent control (Kruskal-Wallis test: p < 0.05). Glomerular nos. 1–27 were tentatively mapped to PM1, PM2, V1–3, AM2–5, AL3, LC2, LC1, AC1, AL1, AL2, PL2, MC2, PL4, PD1, PD3, PC1, MD1–3, PD4, PD2, and AD1, respectively. Bars represent the mean ± SEM. (G) Dynamics of the calcium response to geosmin (green trace) and the solvent control (dipropylene glycol [DPG], black trace) for the putative PD3 glomerulus. Lines are the mean; shaded areas are the SEM. Grey bar denotes stimulus duration (2 s). (H) AL atlas showing the tentatively identified PD3 glomerulus (green), which is responsive to geosmin; the AL3 glomerulus (blue), responsive to nonanal; and the AM2 glomerulus (magenta), responsive to lilac aldehyde. (I) Concentration dependency of glomeruli tentatively identified as PD3, AL3, and AM2 to their cognate odorants (geosmin, nonanal, and lilac aldehyde, respectively). The glomeruli showed significantly different dose response curves (F_1,105 = 21.5; p < 0.05), with the PD3 glomerulus having the lowest EC₅₀ (10⁻⁹ concentration) compared to AL3 (10⁻⁵) or AM2 (10⁻⁴). Lines are the mean; shaded areas are the SEM. (J) Tuning curve for the PD3 glomerulus to a panel of 21 odorants, each tested at 10⁻² concentration. See also Figure S1.

Figure 4.. Geosmin as a Potential Mosquito Control Agent

(A) Map over greater Miami area with trap sites marked. Satellite image courtesy of Google Maps. (B) Oviposition trap used for the field experiments. (C–E) OIs from Miami mosquitoes offered a choice between control traps (water only) and traps baited with geosmin. Each data point represents the average OI from a single site (n = 11–14). Total number of eggs is as follows: (C) 2,240, (D) 2,594, (E); 2,946; (C) geosmin (10⁻³) 31 ± 6 eggs per trap, control 24 ± 5; (D) geosmin (10⁻⁴) 35 ± 6, control 26 ± 4; (E) geosmin (10⁻⁵) 39 ± 8, control 31 ± 6 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. (F) OIs of WT mosquitoes (20 mosquitoes per trial; n = 6 trials) from binary-choice tests between whole beetroot extract and water. Boxplots and statistics as per Figure 1C are shown. (G) FID traces from a gas chromatography-mass spectrometry analysis of head space comparing volatiles emitted from beetroot peel and beetroot pulp. (H) Pseudocolor plot of ΔF/F₀ calcium responses (0 to 1 scale) to beetroot peel and pulp, at a depth of 75 μm from the surface of the AL. (I) PD3 responses (ΔF/F₀) to the extracts of the beet rind (brown), pulp (purple), and solvent (methanol) control (blue). Grey bar denotes the time course of odor stimulus. Traces are the mean; area is the SEM (n = 3 mosquitoes). Shown in the inset are mean responses to the extract. Letters denote significant differences between stimuli (Kruskal-Wallis test: χ = 63.19, p < 0.0001; multiple comparisons: p < 0.05). (J) OIs of WT mosquitoes (20 mosquitoes per trial; n = 6 trials) from binary-choice tests between whole beetroot peel and pulp. Total number of eggs is as follows: 2,878; peel 322 ± 82 eggs, pulp 158 ± 41 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. (K) Brazil field site. Satellite image courtesy of Google Maps. (L) Oviposition trap constructed from painted PET bottles lined with filter paper used for the experiments in Brazil. (M) OIs from wild Brazilian mosquitoes offered a choice between control traps (water only) and traps baited with beetroot peel extract. Each data point represents a collection event. Total number of eggs is as follows: 1,630; peel 45 ± 7 eggs, control 18 ± 3 (mean ± SEM). Boxplots and statistics as per Figure 1C are shown. See also Figure S2.