Characterization of the honeybee AmNaV1 channel and tools to assess the toxicity of insecticides

Pollination is important for both agriculture and biodiversity. For a significant number of plants, this process is highly, and sometimes exclusively, dependent on the pollination activity of honeybees. The large numbers of honeybee colony losses reported in recent years have been attributed to colony collapse disorder. Various hypotheses, including pesticide overuse, have been suggested to explain the disorder. Using the Xenopus oocytes expression system and two microelectrode voltage-clamp, we report the functional expression and the molecular, biophysical, and pharmacological characterization of the western honeybee’s sodium channel (Apis Mellifera NaV1). The NaV1 channel is the primary target for pyrethroid insecticides in insect pests. We further report that the honeybee’s channel is also sensitive to permethrin and fenvalerate, respectively type I and type II pyrethroid insecticides. Molecular docking of these insecticides revealed a binding site that is similar to sites previously identified in other insects. We describe in vitro and in silico tools that can be used to test chemical compounds. Our findings could be used to assess the risks that current and next generation pesticides pose to honeybee populations.


Supplemental results: sequence and features of the honeybee's NaV1 channel regulatory subunits
Five regulatory subunits for the NaV1 channel have been identified previously in Drosophila Melanogaster 1 . Interestingly, these subunits were reported to be similar to the regulatory subunits of mammalian calcium-activated potassium channels (KCa) which are distant cousins of Nav channels in the VGL-chanome 2 . Screening a HMM protein profile inferred from KCaβ subunits against the bee proteome resulted in the successful identification of the TipE protein and its homologs (e-values from 3,5E-66 to 5,8E-68).
Unlike the NaV1 sequence characterization, the TipE and TEH sequence characterizations were performed with the jackhammer algorithm. This algorithm consists in iterative hmmsearch-like queries. Each new query uses a protein profile inferred from the selected proteins associated with the significant matches of the previous query. This procedure was performed in order to make sure that all TipE homolog proteins were identified from the bee proteome. Using a similar approach with a protein profile inferred from mammalian regulatory subunits of Nav channels did not highlight any homologous protein in the bee proteome.
Comparison of the identified subunits with the known KCaβ (KCNMB) subunits shows that, these proteins belong to a single protein family ( Figure 1D). Moreover, based on sequence homology, the TEH family is divided into two subfamilies. TEH3 and TEH4 seem to feature sequence motifs slightly different from TipE, TEH1 and TEH2 All the regulatory subunits of AmNaV1 are located within a span of 2 686 kB in the LG1 chromosome ( Figure S2C). The TipE, TEH2, TEH3 and TEH4 genes are all located in closer proximity (within 74 kB). All the TEH subunits are oriented in the same direction.

Similar arrangements have been reported for the TipE and TEH subunits of the Drosophila
Melanogaster, Drosophila Pseudoobscura, Drosophila Yakuba and Anopheles gambiae 1 .
Such an arrangement either points to early gene duplication and inversion events in the phylogeny or to common regulation of the genes.
Using the SOSUI web server 3 , we identified two putative transmembrane segments in all regulatory subunit but TEH1. Given that all the putative transmembrane segments identified are in the same regions and that drosophila's TEH1 subunit is thought to feature two transmembrane segments 1 , it is probable that the honeybee's TEH1 also has two transmembrane segments in the same regions ( Figure S3).
As noted by Derst and collaborators for the drosophila's TipE and TEH subunits 1 , some cysteines thought to be involved in the formation of disulphide bridges in hKCaβ2 are conserved in the honeybee's NaV1 regulatory subunits 1 . Other cysteines not found in hKCaβ2 seem to be highly conserved in both drosophila's and honeybee's regulatory subunits ( Figure S3A-B). Those residues may be involved in the formation of disulphide bridges as well. Furthermore, TEH3 and TEH4 both feature two EGF-like domains. Those domains are typically associated with protein-protein interactions and calcium binding.
TEH3's first EGF-like domain seems to be involved in calcium binding as it features the D/N-x-D/N-E/Q-x1-D/N*-x2-Y/F consensus sequence (where x1 and x2 are of variable length and * represents β-hydroxylated residue). TEH4's first EGF-like domain, however, only features some residues of the consensus sequence.
No alternative sequences were found for the honeybee's TEH1, TEH2 and TEH3 subunits.
Nevertheless, alternative sequences for TipE and TEH4 were found as a result of genomic variations and RNA editing ( Figure S2B).

Supplemental methods: cloning AmNaV1 and its regulatory subunits
Total RNA from honeybee heads was extracted using Trizol kits (Sigma). cDNA was produced using Transcriptor first strand cDNA synthesis kits (Roche). The cDNA corresponding to two overlapping fragments of the NaV1, the TipE, and TEH1-4 proteins was obtained by PCR amplification. Prior to insertion in a plasmid with a standard digestion and ligation method, restriction sites were introduced at the beginning and end of each fragment cloned using PCR amplification with the following primers: The vectors containing the regulatory subunits were amplified in Escherichia coli XL2 Blue (Agilent) and were purified using GenElute HP Plasmid Maxiprep kits (Sigma).
The constructs were linearized with Not1, and T7 RNA polymerase was used to make sense RNA using mMESSAGE mMACHINE T7 kits (Ambion). mRNA was also generated using mMESSAGE mMACHINE kits with a custom DNA featuring a T7 promoter, the AmNaV1 channel sequence, the Xenopus laevis ß-globin 3′-untranslated region, and polyA and polyC tracts.
To identify sequence variations, genomic DNA from whole bees was extracted and was sequenced using QIAamp DNA mini kits (Qiagen).

Supplemental methods: oligonucleotide primer sequences for tissue expression analysis
The following primer sequences were used for RT-PCR amplifications in order to visualize the tissue expression pattern of the proteins under investigation:

Supplemental methods: Equations
The fraction of channels modified by pyrethroid insecticides was calculated using the following equation: where Itail is the maximal amplitude of the tail current measured in the 170s following the last conditioning pulse (the first 5ms are excluded to avoid the inclusion of a capacitive component in the current measured), Etail is the membrane potential at which the tail current was measured, Erev is the reversal potential of sodium ions as calculated using an I-V curve, and Idep is the peak current observed in response to a depolarization at the Edep potential.
The dose-response curves obtained from these calculations were fitted to a Hill curve using the following equation: where h is the Hill coefficient, Mmax is the maximal fraction of channels affected, EC50 is the half-maximal effective concentration, and dose is the concentration at which the observation was made.
The voltage-dependence of activation was fitted to the following Boltzmann equation: while the voltage-dependence of inactivation was fitted to the following equation: where V1/2 is the half-maximal voltage of activation (or inactivation), G is the conductance, I is the current measured at a given voltage (V), and k is the slope factor (in mV).
Current decay and tail current kinetics were fitted to the following equation when a single exponential was used: and to the following equation when double exponential was used: where y0 is the initial value, A is the weight of the exponential, and τ is the time constant of the exponential.
All fits were performed using the Levenberg-Marquardt algorithm.

Supplemental tables:
Table S1: Correspondence of residues in the honeybee NaV1 sequence with known kdrlinked mutations