Female A. gambiae mosquitoes are primarily attracted by oct-1-en-3-ol in human sweat and a previous study by another group revealed that the molecule does not interact with Orco but instead activates other ORs [48]. It has even been suggested that DEET repels mosquitoes by reducing the volatility of oct-1-en-3-ol rather than inhibiting any interactions with ORs [49,50]. We investigated the interaction between oct-1-en-3-ol and Orco by injecting 1 µL of the undiluted molecule and 1:1, 1:3 and 1:7 mixtures (volume basis) of oct-1-en-3-ol and Hank’s buffer into the assay buffer. We performed the study in Hank’s buffer containing 5 mM of Ca²⁺ but noticed insignificant differences in the fluorescence emissions from each sample (Figure 9 ). Subsequent addition of VUAA1 to a final concentration of 1 mM in the solutions activated Orco expression by the PP-Orco cultures equally (data not shown). These observations confirm that oct-1-en-3-ol does not interact with Orco of A. gambiae .

Figure 9: The Pichia biosensor does not respond to oct-1-en-3-ol , which is corroborates previous reports that Orco is not activated by the odorant. The dilutions correspond to volumetric ratios of oct-1-en-3-ol and Hank’s buffer containing 5 mM Ca²⁺ in an injection of 1 µL into the assay buffer. ‘None’ refers to no dilution.

Most mosquito and insect repellants that are presently on the market were discovered using an apparatus known as an olfactometer. It is quintessentially a Y-shaped chamber that studies the behavior of mosquitoes that are housed at the base of Y when they are exposed to odorant samples in both or either one of its two arms [51]. Not only are these experiments extremely cumbersome and slow to perform, but olfactometers are error-prone and offer limited control over critical experimental parameters [16,17]. The Pichia odor biosensor described in this study directly addresses these limitations. We have conclusively demonstrated that P. pastoris can functionally express Orco and that the level of expression can be modulated by varying the concentration of the transcriptional inducer. We have also identified culturing conditions that facilitate optimal expression of the protein. We then tested the biosensor by exposing it to VUAA1, one of the strongest agonists of Orco reported in literature. Although it is known that Orco is a cationic channel, we observed that it functions like a TRP cationic channel after activation, even when expressed heterologously. Moreover, the fluorescent signal from the assay can be titrated by adjusting the concentration of either the stimulatory odorant or concentration of extracellular Ca²⁺ in the assay buffer. The EC50s of VUAA1 for the Pichia expressed Orco were determined to be 0.83 mM and 0.41 mM when the Ca²⁺ concentrations in the assay buffer are 1 mM and 5 mM, respectively.

We also exposed the biosensor to citronella oil and oct-1-en-3-ol. Citronella oil is a widely used insect repellant, and oct-1-en-3-ol is a metabolite present in human sweat that is the primary attractant of mosquitoes. Previous studies have revealed that oct-1-en-3-ol does not interact with Orco [52]. However, the effect of citronella oil on the protein is poorly understood. In D. melanogaster , citronellal, the primary constituent of citronella oil, has been shown to be an agonist of Orco as well as TRPA1 receptors [46]. ThePichia biosensor corroborates that oct-1-en-3-ol does not interact with Orco of A. gambiae , but we determined that either citronellal or a minor constituent of citronella oil interferes with Orco. The ligand is either a negative allosteric modulator of Orco or an antagonist that is weakly competitive with VUAA1. We believe it is possible to further increase the sensitivity and signal-to-noise ratio of the biosensor by maintaining a higher concentration of Ca²⁺ in the assay buffer. P. pastoris has been shown to grow normally at extracellular Ca²⁺concentrations as high as 100 mM [53]. We are also confident that the fluorescence emitted by the cells is definitely a product of channel activation and not any other phenomena since the physiological concentration of Ca²⁺ ions in the cytoplasm ofP. pastoris ranges between 50 and 200 nM [53,54].

Similar biosensors have been constructed previously using HEK293 cells [32], Sf9 cells [55,56] and Xenopus oocytes [57]. Not only are these cell lines cumbersome and expensive to maintain, but they also require the use of patch clamping to assess receptor activation or deactivation, which is incompatible with high-throughput screening. Among these competing platforms, only the Xenopus system has been adapted to a relatively high-throughput microfluidics screening platform. However, the transformation of Xenopus oocytes is slow and has a low efficiency. In contrast, not only is the Pichiabiosensor comparably sensitive as these systems [28,58], but it is also easier to maintain and deploy and simpler to modify and optimize.P. pastoris also has a faster doubling time [59] and does not require the use of high doses of antibiotics, which is a significant advantage over other screening platforms [29,30,32]. Moreover, since all 79 ORs and Orco of female A. gambiae mosquitoes exhibit a high degree of structural and topological similarity within the membrane, the system is highly modular and can be used to investigate any of these proteins by co-expressing them with Orco.

In closing, the Pichia biosensor developed in this study is sensitive and could form the basis of miniaturized, high-throughput and highly precise assays for identifying chemical modulators of the mosquito’s sense of smell. The biosensor opens opportunities for medicinal chemistry to be performed, which could then facilitate systematic elucidation of structure-activity relationships and the subsequent identification of effective repellants through lead optimization [29]. Beyond repellant screening, the P. pastoris odorant sensor could also be used as an investigative tool in other fields such as entomology, agriculture and aromachology.

This work was supported by the Discovery Grants Program of the Natural Sciences & Engineering Research Council of Canada (NSERC). JNV was supported by a doctoral fellowship from NSERC’s Collaborative Research & Training Experience (CREATE) Program.

The authors declare no conflicts of interest.

We would like to acknowledge Dr. Sandip Pawar for his technical assistance. We would also like to thank Profs. Abby Collier and Philip Hieter from the University of British Columbia, who allowed us to use the ultracentrifuge and Western blotting equipment in their laboratories, respectively, and Prof. Steven Hallam, also from the University of British Columbia, for donating the P. pastorisGS115 strain and pPICZA plasmid used in this study.

VGY conceived the study. JNV performed the experiments and collected the data. JNV and VGY designed the experiments, analyzed the data and wrote the manuscript.