Multiple mechanisms of action for an extremely painful venom

Abstract

Evolutionary arms races can lead to extremely specific and effective defense mechanisms, including venoms that deter predators by targeting nociceptive (pain-sensing) pathways. The venom of velvet ants (Hymenoptera: Mutillidae) is notoriously painful. It has been described as "Explosive and long lasting, you sound insane as you scream. Hot oil from the deep fryer spilling over your entire hand."¹ The effectiveness of the velvet ant sting against potential predators has been shown across vertebrate orders, including mammals, amphibians, reptiles, and birds.^2,3,4 This leads to the hypothesis that velvet ant venom targets a conserved nociception mechanism, which we sought to uncover using Drosophila melanogaster as a model system. Drosophila larvae have peripheral sensory neurons that sense potentially damaging (noxious) stimuli such as high temperature, harsh mechanical touch, and noxious chemicals.^5,6,7,8 They share features with vertebrate nociceptors, including conserved sensory receptor channels.^9,10 We found that velvet ant venom strongly activated Drosophila nociceptors through heteromeric Pickpocket/Balboa (Ppk/Bba) ion channels, through a single venom peptide, Do6a. Drosophila Ppk/Bba is homologous to mammalian acid-sensing ion channels (ASICs).¹¹ However, Do6a did not produce behavioral signs of nociception in mice, which was instead triggered by other venom peptides that are non-specific and less potent on Drosophila nociceptors. This suggests that Do6a has an insect-specific function. In fact, we further demonstrated that the velvet ant's sting produced aversive behavior in a predatory praying mantis. Together, our results indicate that velvet ant venom acts through different molecular mechanisms in vertebrates and invertebrates.

Keywords: DEG/ENaC; Dasymutilla; Drosophila; degenerin; epithelial sodium channel; nociception; pain; pickpocket channels; velvet ant; venom.

Figure 1.. Velvet ant venom activates larval sensory neurons with a potent nociceptor-specific component that requires Pickpocket/Balboa

(A) Venom was obtained from collected individuals of the female Scarlet Velvet Ant (Dasymutilla occidentalis). (B) Still images from time-series optical recording of cIV da (arrow) and cIII da (arrow heads) neurons expressing the genetically encoded Ca²⁺ sensor GCaMP6f (w; ppk1.9-Gal4 UAS-GCaMP6f) 10 s before and 20 s after application of diluted venom. Scale bar represents 20μm. Pseudo-coloring for fluorescence intensity (F). (C) Calcium imaging of diluted venom or vehicle application at time 0 s. cIV da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f/+) respond to venom but not vehicle application. cIII da neurons (w; nompC-Gal4/+; UAS-GCaMP6f/+) do not respond to venom application. n = 6–12 neurons. (D) Dose-response to serial dilutions of pooled venom samples in cIII da and cIV da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f). A nociceptor-specific component potently activates cIV da neurons at >100 fold greater dilution relative to venom that activates cIII da neurons. Data represented as mean ± SEM fit with a Sigmoid curve, at each dilution n = 1–14 neurons from 3 venom pools. See also Figure S1. (E) Calcium imaging of AITC application at time 0 s. cIII da neurons (w; nompC-Gal4/+; UAS-GCaMP6f/+) do not activate strongly in response to AITC. Expression of dTRPA1-C or dTRPA1-D in cIII da neurons (w; nompC-Gal4/UAS-dTRPA1-C; UAS-GCaMP6f/+ or w; nompC-Gal4/UAS-dTRPA1-D; UAS-GCaMP6f/+) renders them responsive to AITC. n = 12 neurons. (F) Calcium imaging of venom application at time 0 s. Expression of dTRPA1-C or dTRPA1-D in cIII da neurons (w; nompC-Gal4/UAS-dTRPA1-C; UAS-GCaMP6f/+ or w; nompC-Gal4/UAS-dTRPA1-D; UAS-GCaMP6f/+) does not render them responsive to venom. n = 12 neurons. (G) Calcium imaging of AITC application at time 0 s. cIV da neurons with RNAi against ppk (w; ppk1.9-Gal4 UAS-GCaMP6f/UAS-ppk RNAi; UAS-dicer2/+) or against bba (w; ppk1.9-Gal4 UAS-GCaMP6f/UAS-bba RNAi; UAS-dicer2/+) respond to AITC similarly as genetic controls (w; ppk1.9-Gal4 UAS-GCaMP6f/attP; UAS-dicer2/+ or w; ppk1.9-Gal4 UAS-GCaMP6f/attP2; UAS-dicer2/+), showing that knock-down of ppk or bba does not interfere with the neurons’ overall ability to respond to stimuli. n = 6 neurons. (H) Calcium imaging of venom application at time 0 s. RNAi against ppk (w; ppk1.9-Gal4 UAS-GCaMP6f/UAS-ppk RNAi; UAS-dicer2/+) or against bba (w; ppk1.9-Gal4 UAS-GCaMP6f/UAS-bba RNAi; UAS-dicer2/+) eliminates the response to venom in cIV da neurons, while genetic controls (w; ppk1.9-Gal4 UAS-GCaMP6f/attP; UAS-dicer2/+ or w; ppk1.9-Gal4 UAS-GCaMP6f/attP2; UAS-dicer2/+) respond normally to venom. n = 6–7 neurons. C, E-H: Data represented as mean ± SEM. For statistical comparisons see Table S1. See also Video S1.

Figure 2.. Pickpocket/Balboa channels are sufficient for neuronal response to venom

(A-B) Still images (A) and quantification (B) of diluted venom application at time 0 s on cIII da neurons expressing ppk alone (w; nompC-Gal4/+; UAS-GCaMP6f/UAS-ppk), which do not respond to venom. n = 12 neurons. (C-D) Still images (C) and quantification (D) of diluted venom application at time 0 s on cIII da neurons expressing bba alone (w; nompC-Gal4/+; UAS-GCaMP6f/UAS-bba-mCherry), which do not respond to venom. n = 12 neurons. (E-F) Still images (E) and quantification (F) of diluted venom application at time 0 s on cIII da neurons co-expressiing ppk and bba (w; nompC-Gal4/+; UAS-GCaMP6f/UAS-ppk UAS-bba-mCherry), which respond to venom. n = 16 neurons. (G-H) Still images (G) and quantification (H) of vehicle application at time 0 s on cIII da neurons co-expressing ppk and bba (w; nompC-Gal4/+; UAS-GCaMP6f/UAS-ppk UAS-bba-mCherry), which do not respond to vehicle. n = 12 neurons. B, D, F, H: Data represented as mean ± SEM. For statistical comparisons see Table S1. See also Video S2.

Figure 3.. Four venom peptides activate larval sensory neurons but only one peptide, Do6a, mediates activation through Pickpocket/Balboa

(A) Exemplar calcium-imaging traces in cIV da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f) in response to all venom peptides tested at 100 μM and applied at time 0 s. Four peptides, Do6a, Do10a, Do12a, and Do13a, activate cIV da neurons. The eighteen other peptides do not activate larval sensory neurons. Two peptides, Do14a and Do17a, were not tested due to poor solubility. (B) Exemplar calcium-imaging traces in cIII da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f) in response to venom peptides tested at 100 μM and applied at time 0 s (from the same trials shown in (A)). Peptides Do10a, Do12a, and Do13a activate cIII da neurons, but peptide Do6a does not. The remaining peptides also do not activate cIII da neurons. (C) Calcium imaging of Do6a (20 μM) application at time 0 s. RNAi against ppk (w; ppk1.9-Gal4 UAS-GCaMP6f/UAS-ppk RNAi; UAS-dicer2/+) or against bba (w; ppk1.9-Gal4 UAS-GCaMP6f/ UAS-bba RNAi; UAS-dicer2/+) eliminates the response to Do6a in cIV da neurons, while genetic controls (w; ppk1.9-Gal4 UAS-GCaMP6f/attP; UAS-dicer2/+ or w; ppk1.9-Gal4 UAS-GCaMP6f/attP2; UAS-dicer2/+) respond normally to Do6a. Data represented as mean ± SEM, n = 6 neurons. (D) Calcium imaging of Do6a (20 μM) application at time 0 s. cIV da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f/+) respond to Do6a application but cIII da neurons (w; nompC-Gal4/+; UAS-GCaMP6f/+) do not. Co-expression of ppk and bba in cIII da neurons (w; nompC-Gal4/+; UAS-GCaMP6f/UAS-ppk UAS-bba-mCherry) renders them responsive to Do6a. Data represented as mean ± SEM, n = 6–12 neurons. For statistical comparisons see Table S1. See also Figure S2. (E) Dose-response curves in cIV da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f) for Do6a, Do10a, Do12a, and Do13a. Half-maximal effective concentration (EC₅₀) is 113 nM for Do6a, 74.5 μM for Do10a, 2.45 μM for Do12a, and 48.7 μM for Do13a. Data represented as mean ± SEM fit with a Sigmoid curve, n = 3 neurons per concentration. (F) Dose-response curves in cIII da neurons (w; ppk1.9-Gal4 UAS-GCaMP6f) for Do6a, Do10a, Do12a, and Do13a (from the same trials as in (E)). Half-maximal effective concentration (EC₅₀) is 20.7 μM for Do10a, 2.29 μM for Do12a, and 14.7 μM for Do13a. Data represented as mean ± SEM fit with a Sigmoid curve, n = 3–10 neurons per concentration. See also Video S3 and Table S2.

Figure 4.. Whole venom and venom peptides are noxious to mice, and velvet ant stings deter an insect predator

(A) Intraplantar injection of venom unilaterally into the mouse hind paw produced nocifensive behaviors compared to control PBS injection. Nocifensive behaviors were measured over 10 min following injection of venom or PBS into the hind paw. ****p<0.0001, unpaired two-tailed t-test. (B) Venom decreases paw withdrawal thresholds (force in g) in the ipsilateral (injected) paw compared to PBS injection. Withdrawal thresholds were measured at the pre-injection baseline (BL) and at various time points after the intraplantar injection. Two-way repeated measures ANOVA with main effect between groups F(1, 10) = 15.57, p = 0.0027, Sidak’s multiple comparison test, ****p<0.0001, ***p<0.001. (C) Intraplantar injections (same mice shown in B) did not alter paw withdrawal thresholds in the contralateral (intact) paw. Two-way repeated measures ANOVA with no main effect between groups F(1, 10) = 0.6405, p = 0.4421, Sidak’s multiple comparison test with no significant differences. (D) Intraplantar injection of peptides Do10a and Do13a but not Do6a produced nocifensive behavior in the injected paw. Do13a produced maximal nocifensive behavior compared to all other groups. Two-way ANOVA with main effect between groups F(3, 18) = 121.7, p < 0.0001, Tukey’s multiple comparison test, ****p<0.0001, ***p<0.001, ns no significant difference. (E) Intraplantar injection of peptides Do10a and Do13a but not Do6a decreased mechanical paw withdrawal thresholds in the injected paw. Two-way repeated measures ANOVA with main effect between groups F(3, 28) = 42.94, p < 0.0001, Tukey’s multiple comparison test, ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05. + denotes differences between PBS and Do10a, * denotes differences between PBS and Do13a. (F) None of the treatments altered paw withdrawal thresholds in the contralateral (intact) paw (same mice shown in E). Two-way repeated measures ANOVA with no main effect between groups F(3, 28) = 0.1698, p = 0.9159, Tukey’s multiple comparison test with no significant differences. A-F: Data represented as mean ± SEM, n = 6–11 mice per group. (G) Still frames from Video S4 showing the defensive sting inflicted by the velvet ant to the foreleg femur/trochanter of the praying mantis, followed by the velvet ant’s escape a few seconds later. See also Figure S3 and Table S2.