Figures and data
Comparative sea anemone stinging behavior.
A) Nematostella vectensis stings with tentacles while Exaiptasia diaphana also stings with acontia filaments that are ejected from its body for defense. Left: Nematostella nematocyte discharge was only observed in response to simultaneous prey chemicals and touch stimuli. Middle, Right: Exaiptasia nematocyte discharge from tentacles and acontia occurred irrespective of prey cues (touch alone). Scale bar = 50μm.
B) Nematostella nematocyte discharge was elicited by simultaneous touch and prey chemical stimuli (n = 10 trials). Exaiptasia tentacle (n = 10) and acontia (n = 13) nematocytes discharged only to touch, with or without prey chemicals. p < 0.05 for Nematostella, paired two-tailed student’s t-test. Data represented as mean ± sem.
Nematostella stinging is regulated by predation while Exaiptasia stings for defense.
A) Top: Nematostella burrows in the substrate and stings for predation. Bottom: We assumed the cost of stinging does not change with starvation state.
B) Left: Desirability of nutritional state, or reward, decreases with starvation. Three examples are shown: example 1, r(s) = 10 tan−1(1 − s); example 2, 
C) Examples of optimal (blue) versus random (black) predatory stinging. Each agent (anemone) starts with s = 0.9, and stings sequentially for many events (represented on the x axis). The random agent almost always reaches maximal starvation before time 50 events (grey lines, five examples shown). In comparison, the optimal agent effectively never starves due to a successful stinging strategy optimized for predation (blue lines, five examples shown, parameters as in panel B, curve with matching color).
D) Top: Exaiptasia diaphana relies heavily on endosymbiotic algae for nutrients and stings primarily for defense. Bottom: We assumed there are two states, safety (L), and danger (D). The state of safety can transition to danger, but not the other way around. We assumed the agent obtains reward 1 in state L and penalty −1 in state D.
E) Left: Cost function, which is assumed to increase linearly with the fraction of nematocysts discharged cs(a) = c0a The cost c0 for full discharge is constant as in panel A, c0 = 1 (example 1), or varies with starvation: c0(s) = s (example 2), c0(s) = 1 − (s − 1)2 (example 3) and c0 obtained by fitting the experimental data (example 4) (see fitting procedure in Supplementary Information). Right: Predicted optimal stinging obtained by solving equation (2) with numerical simulations (circles) and analytical solutions (lines). Colors match the corresponding cost in Left panels, and we assume p(a) = PM a(2 − a) and pM = 0.8 as before. Optimal defensive stinging is constant or decreases with starvation under broad assumptions (see Supplementary Information).
F) Left: Nematostella nematocyte discharge was affected by prey availability while Exaiptasia stung at a similar rate regardless of feeding. p < 0.0001 for Nematostella, two-way ANOVA with post hoc Bonferroni test (n = 10 animals, data represented as mean ± sem). Right: Normalized optimal nematocyst discharge predicted from MDP models for both Exaiptasia (orange circles, using the cost function in E, example 4) and Nematostella (blue circles, using the reward function in B, example 2) fit the experimental measurements well. We match the last experimental data point to s = 0.5, the precise value of this parameter is irrelevant as long as it is smaller than 1, representing that animals are not severely starved during the experiment.
Exaiptasia nematocyte voltage-gated Ca2+ currents have weak inactivation compared with Nematostella.
A) Touch-elicited Exaiptasia tentacle nematocyte discharge was blocked in the absence of Ca2+ (p < 0.01, paired two-tailed student’s t-test, n = 9) or by addition of the CaV channel blocker Cd2+ (500μM, p < 0.05, paired two-tailed student’s t-test, n = 6). Scale bar = 50μm.
B) Top: Representative patch clamp experiment from an Exaiptasia nematocyte. Scale bar = 20μm. Bottom: Nematocyte voltage-gated currents elicited by a maximally activating 0mV pulse were blocked by Cd2+ (n = 3, p < 0.01, paired two-tailed student’s t-test).
C) Nematocyte voltage-gated currents elicited by −120mV (black) or 0mV pulses (colored). Conductance-voltage curves for Nematostella nematocyte (Va1/2 = −26.54 ± 0.78mV, n = 3) and Exaiptasia nematocyte (Va1/2 = −12.47 ± 0.70mV, n = 3).
D) Nematocyte voltage-gated currents elicited by a maximally activating voltage pulse following 1 s pre-pulses to −110 mV (max current, black), −50 mV (colored), or 20 mV (inactivated, no current). Nematostella nematocytes inactivated at very negative voltages (Vi1/2 = −93.22 ± 0.42mV, n = 7) while Exaiptasia contained two populations of nematocytes: low-voltage threshold (Vi1/2 = −84.94 ± 0.70mV, n = 4), and high-voltage threshold (Vi1/2 = −48.17 ± 3.32mV, n = 3). Data represented as mean ± sem.
Exaiptasia expresses a CaV β subunit splice isoform that confers weak voltage-dependent inactivation.
A) ddPCR ratio of concentrations of CaV β subunit 1 and 2 mRNAs was similar in tentacle (n = 5), body (n = 5), and acontia (n = 4 animals) tissue samples.
B) EdCaVβ1 and EdCaVβ2 localized to distinct nematocytes in Exaiptasia tentacle cross section, as visualized by BaseScope in situ hybridization. Representative nematocyte expressing EdCaVβ1 (green) or EdCaVβ2 (red). Representative of 3 animals.
C) Voltage-gated currents from heterologously-expressed chimeric mammalian CaV (mCaV) with different β subunits: rat (Rattus norvegicus), Nematostella (Nve), Exaiptasia EdCaVβ1 or EdCaVβ2. Top: Currents elicited by voltage pulses to −120mV (no current, black) and maximally activating 0mV (colored). Bottom: Voltage-gated currents elicited by a maximally activating voltage pulse following 1 s pre-pulses to −110 mV (max current, black), −50 mV (colored), or 20 mV (inactivated, no current, black). Scale bars = 100pA, 50ms.
D) Exaiptasia CaV β subunit splice isoforms confer distinct inactivation: Nematostella β subunit (Vi1/2 = −68.93 ± 1.53mV, n = 5) and Rat β subunit (Vi1/2 = −2.98 ± 13.51mV, n = 12) and EdCaVβ1 (Vi1/2 = - 56.76 ± 3.18mV, n = 8), and EdCaVβ2 (Vi1/2 = −18.84 ± 8.00mV, n = 5). Data represented as mean ± sem.
E) Genomic alignment of Exaiptasia β subunit isoforms showed that alternative splicing of the N-terminus region was associated with distinct inactivation: CaVβ1 (long N-term) had strong inactivation similar to Nematostella, while CaVβ2 (short N-term) exhibited weak inactivation similar to its mammalian orthologue. Genomic loci listed above genomic sequence.
Cnidarian CaV β subunit N-termini confer unique inactivation properties.
A) Voltage-gated currents from heterologously expressed CaV channels with Nematostella-rat chimeric β subunits demonstrate that the Nematostella N-terminus is sufficient to drive inactivation at negative voltages. Currents shown in response to 10 mV voltage pulses following 1 s pre-pulses to −130 mV (max current, black), −50 mV (colored), or 0 mV (inactivated, no current, black). Scale bars = 100pA, 50ms.
B) Diagram of CaV Nematostella-rat β subunit domain swaps and resulting Vi1/2 values. The Nematostella β subunit N-terminus is required and sufficient for uniquely hyperpolarized CaV inactivation properties (p < 0.001 for average Vi1/2 values across mutant beta subunits, one-way ANOVA with post-hoc Tukey test, n = 2-8 cells).
C) Phylogenetic tree of β subunit sequences obtained from several species of cnidarians. Abbreviations of species: Nve, Nematostella vectensis; Ed, Exaiptasia diaphana; Cc, Cyanea capillata (jellyfish); Pp, Physalia physalis (siphonophore); Ch, Clytia hemisphaerica (jellyfish); Cx, Cassiopea xamachana (jellyfish); r, Rattus norvegicus.
D) Top: Percentage of identity between amino acid sequences across β subunit protein domains for NveCaVβ, EdCaVβ1, EdCaVβ2, CcCaVβ, PpCaVβ, ChCaVβ, CxCaVβ2, rCaVβ2. Bottom: Fraction of identity of amino acids across sites of the β subunit protein.
E) Cnidarian CaV β N-termini shift weak voltage-dependent inactivation of CaV channels containing EdCaVβ2. Voltage-dependent inactivation (Vi1/2) of heterologously-expressed CaVs with WT EdCaVβ2, β subunits from the indicated cnidarians, and chimeras with their N-termini on EdCaVβ2 (p < 0.0001 for average Vi1/2 values with multiple comparisons against WT EdCaVβ2 mean, one-way ANOVA with Bartlett’s test and post-hoc Tukey test, n = 4-9 cells). Data represented as mean ± sem.
Modulation of Nematostella and Exaiptasia stinging is not due to changes in the abundance of nematocytes.
Nematocytes were highly abundant in tentacles from Nematostella (top) and Exaiptasia (bottom) before and after starvation. Representative of n = 3 animals. Scale bar = 50μm.
Transcriptomic and molecular analyses of Exaiptasia β subunit isoforms.
A) mRNA expression (transcripts per million, TPM) of voltage-gated calcium (CaV) channel α and β subunits in Exaiptasia tentacle (nematocyte abundant, blue), body (nematocyte non-abundant, red), bleached (minimal symbionts) tentacle (light blue), bleached body (light red) tissues. NompC, the putative mechanoreceptor in Nematostella nematocytes (Schüler et al., 2015; Weir et al., 2020), was also detected in Exaiptasia tentacles.
B) Representative plots of fluorescent amplitude across event number (droplet events) from amplification of unique regions of EdCaVβ1 (Ch1, Top) and EdCaVβ2 (Ch2, Bottom) sequences using droplet digital PCR (ddPCR, Bio-Rad Laboratories). Individual lanes correspond to tentacle RNA, body RNA, acontia RNA, and no template control (NTC). Blue and green points indicate positive PCR droplets after thresholding and gray points indicate negative droplets.
Voltage-dependent activation of CaV channels is conserved across cnidarian β subunits.
A) Top: Voltage-gated currents from heterologously-expressed chimeric CaVs with the indicated β subunits elicited by voltage pulses to −120mV (no current, black) and 0mV (colored). Abbreviations of species: Nve, Nematostella vectensis; Ed, Exaiptasia diaphana; Cc, Cyanea capillata (jellyfish); Pp, Physalia physalis (siphonophore); Ch, Clytia hemisphaerica (jellyfish); Cx, Cassiopea xamachana (jellyfish); r, Rattus norvegicus. Bottom: Voltage-gated currents elicited by a maximally activating voltage pulse following 1 s pre-pulses to −110 mV (max current, black), −50 mV (colored), or 20 mV (inactivated, no current, black). Scalebars = 100pA, 50ms.
B) Activation and inactivation curves for heterologously-expressed chimeric CaVs with different β subunits. Activation: rCaVβ2 Va1/2 = −19.76 ± 1.16mV, n = 12; NveCaVβ Va1/2 = −23.07 ± 1.16mV, n = 5; EdCaVβ1 Va1/2 = −18.27 ± 1.08mV, n = 8; EdCaVβ2 Va1/2 = −14.22 ± 1.46mV, n = 5; CcCaVβ Va1/2 = −18.47 ± 1.59mV, n = 6; CxCaVβ Va1/2 = −28.89 ± 1.54mV, n = 15; PpCaVβ Va1/2 = −15.29 ± 1.23mV, n = 10; ChCaVβ Va1/2 = −10.30 ± 1.04mV, n = 12. rCaVβ2 Vi1/2 = −2.98 ± 13.51mV, n = 12; NveCaVβ Vi1/2 = −68.93 ± 1.53mV, n = 5; EdCaVβ1 Vi1/2 = −56.76 ± 3.18mV, n = 8; EdCaVβ2 Vi1/2 = −18.84 ± 8.00mV, n = 5; CcCaVβ subunit Vi1/2 = −47.81 ± 5.57mV, n = 6; CxCaVβ Vi1/2 = −87.75 ± 1.72mV, n = 15; PpCaVβ Vi1/2 = −99.80 ± 0.92mV, n = 10; ChCaVβ Vi1/2 = −70.25 ± 4.67mV, n = 12.
C) Diagram of CaV β subunit domain swaps and the length of the N-terminus swapped in amino acids.
D) Cnidarian CaV β N-termini do not greatly affect voltage-dependent activation of CaV channels containing EdCaVβ2. Voltage-dependent activation (Va1/2) of heterologously-expressed CaVs with WT EdCaVβ2, β subunits from the indicated cnidarians, and chimeras with their N-termini on EdCaVβ2, p = 0.5830 for average Vi1/2 values across mutant beta subunits, one-way ANOVA with Bartlett’s test and post-hoc Tukey test, n = 4-7 cells. Data represented as mean ± sem.
